You are on page 1of 14

SPE 84277

Sensitivity Study of Flow Unit Definition by Use of Reservoir Simulation


Anne-Kristine Stolz, SPE, and Ramona M. Graves, SPE, Colorado School of Mines

Copyright 2003, Society of Petroleum Engineers Inc.


Interpretation of flow units based on petrophysical
This paper was prepared for presentation at the SPE Annual Technical Conference and properties, logs, and stratigraphy are routinely done for
Exhibition held in Denver, Colorado, U.S.A., 5 – 8 October 2003.
characterization of a reservoir. Unless numerical simulation is
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
carried out to confirm the flow unit assignment, errors occur.
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to The use of erroneous flow models leads to incorrect prediction
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 of flow performance in the reservoir.
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is Introduction
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous Different methods are available, in the literature, for the
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
interpretation of flow units based on their corresponding
petrophysical properties. But, how much information is
necessary to interpret flow units? How do the number and
Abstract size of flow units affect the performance prediction? Trying to
The concept of flow units has been developed to integrate find the answers to these questions was the origin of
geological and petroleum engineering data. A flow unit is this study1.
defined as a group of reservoir rocks with similar properties In the last years, as a result of the increased interaction
that affect fluid flow. between geologists and engineers in resolving reservoir issues,
Different methods can be applied for the definition of flow geologists have become more aware of the need to vary the
units and their corresponding petrophysical properties. The scale and geologic description according to particular
amount of information required by each method varies engineering applications2. A flow unit is defined as a group of
depending on the data and the tools available. This leads to rocks that have similar properties that affect fluid flow. The
different flow unit interpretation of the same reservoir. The division of the reservoir into flow units or “flow facies”
objective of this study is to compare seven methods for flow provides a better input into numerical flow simulation when
unit definition to a known answer using numerical simulation compared to lithologic or depositional facies3.
to predict the flow performance of the reservoir. The main objectives of this study are:
The data set corresponds to the research well CSM Strat - define different methods to interpret flow units by
Test #61, which is located near Dad, Wyoming, and belongs to varying the amount of information, interpretation tool, and
the Lewis Shale of the Greater Green River Basin. time used,
Conventional and specialized well log information is available - apply different flow unit definition methods to the same
for the 1700-ft deep well, in addition to petrophysical set of data to obtain different study cases or models,
measurements on approximately 170 selected core plug - build a simple simulation model that will predict the
samples taken from 600 ft of core. The data for the well performance of the flow unit models,
includes: gamma ray log, bulk density log, caliper, sonic log, - analyze and compare the results among the flow
compensated neutron log, nuclear magnetic resonance, air unit models.
permeability (Klinkenberg-corrected), helium core porosity, It is important to point out that this study is intended as a
minipermeameter readings, capillary pressure analysis, and simple approach in order to keep constant as many variables
neural network synthetic permeability. as possible. In this way, the number of factors influencing the
A simple simulation model is developed using the software results is reduced, and a better comparison among the different
ECLIPSE 100, Black Oil. Fluid properties, dimensions of the flow unit interpretations can be made. Therefore, the data of
reservoir, grid size, pressure and production data are the same only one well was used, and the methods for flow unit
for all the runs. A waterflood of an oil-water system is definition were based on classical theory trying to vary one
simulated as a means of best observing how flow unit information variable per flow unit model.
delineation affects the results. The results of the reservoir The specific formation of interest for determining
simulation indicate that predicting flow performance varies petrophysical properties in this study, researched by Pyles4 and
depending on the flow unit model used. The flow in more detail by Witton5, is located near Dad, Wyoming, and
performance is a strong function of the average permeability, belongs to the Lewis Shale of the Greater Green River Basin.
and therefore of the flow unit definition method used. In order to verify geologic cross-section interpretations made
by Witton5, a behind-out-crop, shallow research well was
2 SPE 84277

drilled. This 1700-ft deep well, the CSM Strat Test #61, was wells. Average petrophysical rock properties are calculated
drilled in September 1999. Research by Goolsby6 will further for the different flow units.
investigate Witton’s5 interpretations and construct a detailed A summary of the theory behind each method for flow unit
geologic model using core and well log information from the definition is presented. Because the methods lead to the
research well CSM #613. Maglio-Johnson3 developed a “flow models, the words are used interchangeably, but the numbers
facies” framework through the use of petrophysically defined are consistent throughout the paper:
flow units to be input into a reservoir flow simulator. 1. Homogeneous
Conventional and specialized well log information was 2. Gamma-Ray and Density Logs
collected from the well. Approximately 600 ft of core was 3. Flow Zone Indicator (FZI)
retrieved, and petrophysical measurements were run on 4. Winland’s r35
approximately 170 selected plug samples. The data for the 5. Capillary Pressure Curves
well includes: gamma-ray log, bulk density log, caliper, sonic 6. Cumulative Flow and Storage Capacity Curves
log, compensated neutron log, nuclear magnetic resonance, air 7. Stratigraphic Modified Lorenz Plot (SMLP)
permeability (Klinkenberg-corrected), helium core porosity,
minipermeameter readings, capillary pressure analysis, neural Model 1: Homogeneous. A very generalized and simple
network synthetic permeability. model can be obtained by considering the entire depth of the
The first five are routine well logs. The nuclear magnetic well as one big flow unit with homogeneous properties. A
resonance (NMR) is a specialized log, and was mainly used to single permeability (k) and porosity (φ ) value can be
obtain the NMR clay-corrected effective porosity. A storage calculated by taking an average of all values available.
capacity curve created using the NMR clay-corrected porosity
was one of the key tools used to define the flow units in Model 2: Gamma-Ray and Density Logs. Gamma-ray logs
Maglio-Johnson’s3 work. measure natural radioactivity in formations. In general, it can
As routine core measurements, air permeability, help differentiate shales (high radioactivity) from sands,
Klinkenberg-corrected permeability, and helium porosity were carbonates, and anhydrites (low radioactivity)9.
measured. The Klinkenberg correction was made to account The formation density log is a porosity log that measures
for gas slippage effects. As special core measurements, electron density of a formation. It can be used to identify
minipermeameterreadings were made for the whole cored evaporite minerals, detect gas bearing zones, determine
interval using a probe minipermeameter7. hydrocarbon density, and evaluate shaly-sand reservoirs and
Another specialized core test carried out for the CSM Strat complex lithologies9.
Test #61 is capillary pressure analysis, which aids in reservoir The gamma-ray log can be used as a simple tool to
characterization by delineating pore-scale heterogeneities that distinguish between sandstones and shales, for example, and
influence flow behavior3. Capillary pressure measurements of therefore establish zones with different flow properties. A fast
17 selected samples were available from Maglio-Johnson’s3 estimate can be obtained by applying a gamma-ray cut-off,
work. An additional 11 samples were tested for the which separates the entire depth of the well into two groups.
present work. By observing similarities in the gamma-ray signature it is
One important tool in the study of this well is the neural possible to further divide into subgroups, which can be
network approach (NNA), which was used to create synthetic considered as flow units. The density log can be used to
permeability. The neural network synthetic permeability was confirm the flow unit zonation and to calculate average
generated using the modeling program NNLAP. Known porosity values for the depth intervals.
permeability measurements (minipermeameter) from the cored A k-φ plot can be made and the best-fit line along the
intervals and the well log responses from gamma ray, neutron relative trend of the data points can be established. A classical
porosity, and bulk density logs were used as input for linear correlation as shown in equation1 is usually obtained.
the program.
k = Aφ B
Theoretical Development
The flow unit approach to reservoir description provides a
means of uniquely subdividing reservoirs into volumes that or
approximate the architecture of a reservoir at a scale consistent
with reservoir simulations. logk = Blogφ + logA …....………....(1)
There is no universally applicable set of rules by which to
define flow units8. Several methods have been proposed in the Equation 1 represents a straight line in the form
literature, and they are all based on similar principles:
identifying variations in pore geometrical attributes, which y = mx + b ………….………….(2)
affect the fluid transport properties within the rock.
Petrophysical and available well log data are analyzed. An where y is a function of permeability, y=f(k), x is a function of
initial division of the zone into flow units is made based on porosity, x=f(φ), and m and b are functions of the rock type,
differentiating rocks with similar properties. Further analysis (m,b)=f(rock type).
allows the confirmation of the number of flow units, and their
extension to other uncored depths of the well or other uncored
SPE 84277 3

Model 3: Flow Zone Indicator (FZI). This method is based where the units for RQI and FZI are µm, and φz is
on a rearrangement of the Kozeny-Carman10-11 equation made dimensionless. Equation 9 represents a straight line in
by Amaefule et al.12 and the concept of mean hydraulic radius. the form:
The generalized form of the Kozeny-Carman10-11 relationship y = mx + b …………….……… (10)
is given by the following equation:
where y is a function of permeability, y=f(k), x is a function of
φ e3  1  …………………(3) porosity, x=f(φ), and m and b are functions of the rock type,
k=  2 2
(1 − φ e )  Fs τ S gv 
2 (m,b)=f(rock type).
A log-log plot of RQI versus φz results in a straight line
with unit slope because m=1. The value of FZI, which is the
where k is permeability, φe is effective porosity, Fs is the shape
y-intercept, occurs when logφz=0 or φz=1. According to
factor, τ is tortuosity, and Sgv is the surface area per unit
Amaefule et al.12, samples that lie on the same straight line
grain volume.
with similar FZI values (y-intercept) have similar pore throat
The term Fsτ2 is classically referred as the Kozeny attributes and, thereby, constitute a flow unit. Samples with
constant. According to Amaefule et al.12, the Kozeny constant different FZI values will lie on other parallel lines and
varies between hydraulic (flow) units, but is constant within a constitute different flow units12.
given unit. The issue of variability of the Kozeny constant is
addressed by rearranging the variables in the Kozeny- Model 4: Winland’s r35. H. D. Winland used mercury
Carman10-11 relationship (equation 3) to obtain equation 4: injection capillary pressure curves and multiple regression
analysis to develop an empirical equation using porosity, air
k  φ  1  permeability, and the pore aperture corresponding to a
=  e   ……………….. (4)
φ e 1 − φ e   Fs τS gv  mercury saturation of 35% from over 300 sandstone and
limestone samples. He ran regressions for other percentiles
(30, 40, and 50), but the best correlation (highest R2) was the
To simplify the nomenclature, Amaefule et al.12 defined 35th percentile13. The Winland equation (11) was used and
the terms in equation 4 as: published by Kolodzie14:
Flow Zone Indicator (FZI),
log r 35 = 0.732 + 0.588 log k − 0.864 log φ ..…. (11)
1 …..………..……. (5)
FZI =
Fs τS gv where r35 is the pore aperture radius (µm) corresponding to
the 35th percentile mercury saturation, kair is uncorrected air
Reservoir Quality Index (RQI), has a constant to account permeability (md), and φ is porosity (%)13.
for unit conversion, By rearranging Winland’s equation (11), we obtain
k …………………. (6) equation 12, which represents a line in the form (equation 13):
RQI = 0.0314
φe
0.588 log k = 0.864 log φ + (log r 35 + 0.732 ) …. (12)
Nominal Porosity Index (φz),
y = mx + b ……………………..(13)
 φ 
φ z =  e  ………………………(7) where y is a function of permeability, y=f(k), x is a function of
 1 − φe  porosity, x=f(φ), and m and b are functions of the rock type,
(m,b)=f(rock type).
which is the pore volume-to-grain volume ratio, as derived in A reservoir can be subdivided into several flow units based
equation 8. Vp is the pore volume, Vb is the bulk volume and on its Winland’s pore-throat radius (r35). An r35 curve can be
Vg is the grain volume. used to identify intervals (flow units) of similar pore-throat
radii and to discriminate between flow units whose pore-throat
Vp size yields different inflow performance15.
 φe  Vp Vp ………….(8) The preferred means of establishing pore-throat radii is
φ z =   = Vb = =
 1 − φ e  1−
Vp Vb − Vp Vg directly from capillary pressure data16. However, calculating
Vb r35 using Winland’s equation allows intervals to be
characterized as flow units without special core analysis
By substituting these terms into equation 4 we obtain: (capillary pressure measurements); r35 can be calculated from
core porosity and permeability15.
RQI = φ z FZI According to Martin et al.15, four petrophysical flow units
with different reservoir performances are distinguished by
or ranges of r35. One additional range is included in the paper
by Porras and Campos17, leading to the following five
log RQI = log φ z + log FZI ……………… (9) petrophysical categories:
4 SPE 84277

Megaporous, defined by a pore throat radius bigger than This dimensionless J(S)-function serves quite well in many
10 microns cases to remove discrepancies in the capillary pressure versus
Macroporous, defined by a pore throat radius between 2.5 saturation curves and reduce them to a common curve20. The
and 10 microns J(S)-function always maintains the original shape of the
Mesoporous, defined by a pore throat radius between 0.5 capillary pressure curve20.
and 2.5 microns Notwithstanding, the various complexities that contribute
Microporous, defined by a pore throat radius between 0.2 to capillary behavior, for engineering purposes, the J(S)-
and 0.5 microns function provides a practical approach to classifying pore
Nanoporous, defined by a pore throat radius smaller than structure on a global basis using a curve method. While
0.2 microns correlated curves are rarely identical, they can be grouped
according to the same arbitrary definition of closeness in
Model 5: Capillary Pressure Curves. Mercury capillary shape. The J(S)-function curves can be used to group cores,
pressure data directly measure the percentage of pore space which have operationally similar geometries22.
within a rock that can be filled with a given fluid when a given Thomeer’s Hyperbola and Swanson’s Parameter. In
amount of pressure is applied. In general, data are commonly order to characterize the capillary pressure curves by
measured by applying pressure to a core plug to force mercury parameters, different mathematical models have been
into all of the pore space. Samples undergo a range of proposed (Thomeer23, Swanson24, Brooks and Corey25, van
increasing pressure so that the cumulative amount of mercury Genuchten26). Thomeer23 described the curve as rectangular
forced into the rock at each pressure can be recorded. hyperbola of the form
Initially, low pressure is applied and only pore spaces
connected by large pore throats are filled. As pressure  Pc   Sb 
increases, mercury is forced into pore spaces connected by Log   Log   = −C 2 ………….(16)
 Pd  Sb
 ∞
ever decreasing sizes of pore throats. In this way, the range of
pore throat sizes is measured for each sample18. Mercury
injection curves are basically primary drainage curves (from where Pc is the mercury-air capillary pressure (psia), Pd is the
0% initial saturation of the non-wetting phase to residual extrapolated mercury-air capillary displacement pressure
saturation of the wetting phase). (psia), Sb is the bulk volume occupied by mercury (%), Sb∞ is
Leverett’s J(S)-function. The fact that the capillary the bulk volume occupied by mercury at infinite capillary
pressure-saturation curves of nearly all naturally porous pressure or total interconnected pore volume (%), and C2
materials have many features in common has led to attempts defines the shape of the curve. It is related to the pore
to devise some general equation describing all those curves. geometrical factor, G, by
Leverett19 approached the problem from the standpoint of
dimensional analysis20. Leverett21 first derived a relationship C2 = G …………….…… (17)
2.303
between average pore radius, rp, and the permeability and
porosity of a porous medium by applying Poiseuille’s law to
the ideal model given by a bundle of cylindrical tubes of equal Swanson24 developed another parameter that can be used
area, and, hence, radius22 (equation 16). as single point method for comparing capillary pressure
curves. He stated that the maximum curvature of the
hyperbolic equivalent of a capillary pressure curve is found at
8k ……………………….(14)
rp = the intersection of the hyperbola with a 45° line passing
φ through the origin of the hyperbolic axes. The ratio of the
coordinates of this point, (Sb/Pc)A,Hg, has a maximum value at
Reasoning that capillary pressure should depend on the apex of the hyperbola. According to Swanson24 the capillary
porosity, interfacial tension, and on some sort of mean pore pressure at this point corresponds to the pore sizes effectively
radius, Leverett19 later defined the dimensionless function of interconnecting the total major pore system and thus those that
water saturation, which he called the J(S)-function (equation dominate fluid flow. Wells and Amaefule27 presented another
15). In doing so, Leverett interpreted the ratio of permeability method for calculating the Swanson parameter (Sb/Pc)A,Hg.
to porosity as being proportional to the square of the mean This technique is an extension of both the Thomeer23 and
pore radius22. Swanson24 procedures.
They observed that a plot of (Pc/Sb)1/2 versus logSb
Pc ( S ) k …..…………….. (15) resulted in a well defined minimum, which is a unique
J( S ) =
σ φ petrophysical parameter (ΨHg) for a given sample.

1/ 2
where Pc is capillary pressure (psia) at a saturation S  Pc  ………………….(18)
ψ Hg =  
(fraction), σ is the interfacial tension (dyn/cm), k is  Sb 
permeability (md), and φ is porosity (fraction). An
appropriate constant has to be included to account for
unit conversion.
SPE 84277 5

This parameter is related to the Swanson parameter by k1 ( h1 − h0 ) + k 2 ( h2 − h1 ) + ... + k i ( hi − hi −1 ) …..(23)


( kh )cum =
Σk i ( hi − hi −1 )
−1 / 2
  Sb   …………….(19)
ψ Hg =   
φ1( h1 − h0 ) + φ 2 ( h2 − h1 ) + ... + φi ( hi − hi −1 ) .…(24)
  Pc   ( φh )cum =
 A ,Hg 
Σφi ( hi − hi −1 )
Pittman11 determined the apex of a plot of mercury
saturation/capillary pressure versus mercury saturation as a where k is permeability (md), h is the thickness of the sample
means to determine the apex of Thomeer’s23 hyperbola. This interval (ft) and φ is porosity (fraction). With this method,
apex is equivalent to the petrophysical parameter ΨHg both flow and storage capacities are normalized to 1.0, which
presented by Wells and Amaefule27. The difference is that serves as the basis of comparison if several wells are
because of the square root in Wells and Amaefule’s27 being analyzed3.
(equation 24), the curve becomes smoothed. The relationship One way to define flow units in a well is to compare the
between Pittman’s apex and the petrophysical parameter cumulative flow/storage capacity versus depth curves with the
ΨHg is: gamma-ray log signature to determine the lithologic
correlation to flow facies. Changes in flow and storage
−1 / 2 capacity, which represent changes in flow properties, are
 φS  …..………...…(20) indicated by deviations in the slope of the curves. Each break
ψ Hg =  Hg 
 Pc  in slope of the curve is interpreted as a break between
flow units3.
Since mercury bulk volume saturation (Sb) and the
Model 7: Stratigraphic Modified Lorenz Plot (SMLP). The
mercury saturation (SHg) are related by equation 27,
Stratigraphic Modified Lorenz Plot (SMLP) is a plot of
Pittman’s13 apex (26) will be the same as the petrophysical
cumulative flow capacity versus cumulative storage capacity,
parameter ΨHg (equation 24). organized stratigraphically. Inflections in the SMLP
correspond to changes in flow behavior. These changes are
φS Hg = S b ……………………...(21) interpreted using straight-line segments, which represent
individual flow units within the reservoir3. The SMLP offers a
Swanson24, Wells and Amaefule27, and Pittman’s13 guide as to how many flow units are necessary to honor a
parameters are essentially the representation of the maximum geologic framework29.
curvature of the capillary pressure curve or Thomeer’s23 shape
factor. Curves with similar shape factor can be considered as Methods for Flow Unit Definition
belonging to the same flow unit. Seven study cases or models have been chosen for the
Pore throat sorting (PTS), is a number that measures the purpose of the study: the first case, Model 1, is the case
sorting of the pore throats within a rock sample. It ranges interpreted using the most common tools, and, the following
from 1.0 (perfect sorting) to 8.0 (no sorting). The technique models build in complexity by adding more information and
yields similar results to the pore geometrical factor (G) tools to interpret the flow units. The last case, Model 7, is the
developed by Thomeer23. The resulting number provides a one that comprises the most information and tools available
measure of the curvature of the plotted curve, which is in turn for well CSM Strat Test #61. Figure 1 is a symbolic
related to pore geometry. In a similar fashion, PTS provides a representation of how the models were constructed. The
measure of pore geometry by applying a numerical value to volume of each piece of the pyramid represents the
the slope of the plateau found on the semi-log plot of capillary information used to define the flow units in Models 1 through
pressure data. The PTS is defined by the following equation: 7, indicating increasing complexity when going from one
model to another. Complexity refers to amount of
1/ 2 information, the tools and the time required to build the
 3rdQuartile  models; not difficulty of use.
PTS =   . ……………..(22)
 1stQuartile  Two assumptions were made during the development of
the methods: 1) although flow units can be the same at
where the first and third-quartile pressures are obtained different depths, they were considered different if they were
directly from the capillary pressure curve and they reflect the located at different depth intervals, because the data available
25 and 75% mercury saturation adjusted to belonged to a single well; 2) the term “flow unit” was used for
irreducible saturation28. both flow and very low permeability zones
A summary of the steps followed in each method is
Model 6: Cumulative Flow and Storage Capacity Curves. presented next.
Flow capacity is defined as the product of permeability and
thickness (kh), and storage capacity as the product of porosity
and thickness (φh)29. To obtain cumulative flow capacity and
cumulative storage capacity, the following equations are used:
6 SPE 84277

by core sample i, n is the number of samples, φj is average


porosity based on thickness, and φi is the porosity of core
sample i.

Method to define Model 2: GR. The flow units for Model 2


were interpreted using the gamma-ray (GR) log, a density log,
and k-φ crossplot using core plug data. The method applied
included the following steps:
1. A shale/sandstone cut-off line was drawn on the GR-
log at 90 API. By observing how much of the GR signature
curve was above or below the cut-off line, an initial division
of the depth into intervals was made. Zones with similar GR
signature were grouped and interpreted as flow units.
2. A flow unit number was assigned to each of these
groups, the highest number corresponding to the deepest depth
interval Figure 2.
3. The density log was corrected for shale, in order to
obtain effective porosity values.
4. Permeability values were calculated using a k-φ
correlation obtained using the values from core plug
measurements. For depths where the GR signature indicated
GR gamma-ray log
shale, a low permeability value (0.015 md) was assigned. This
ρlog density log low value was obtained taking the average of the core plug
core air and Klinkenberg-corrected permeability, helium porosity measurements at the tight sands.
GRFU interpretation of flow units using GR 5. For each flow unit, an arithmetic average of the
FZI Flow Zone Indicator, Amaefule et al. (1993)
Winland’s equation
effective porosity was calculated based on the porosity data
Winland
NNA synthetic permeability created by Neural Network Approach from the corrected density log.
Pc capillary pressure measurements 6. Permeability for each flow unit was calculated taking a
kh/φh flow/storage capacity geometric average of the values within the flow units.
NMR nuclear magnetic resonance log
SMLP Stratigraphic Modified Lorenz Plot

Figure 1: Symbolic representation of the amount of data and the


tools used to define flow units in the models.

Method to define Model 1: Homogeneous. The data source


for Model 1 consisted of core measurements of Klinkenberg-
corrected permeability and helium porosity from 170 core
plug samples. The method for flow unit definition followed
these steps:
1. The entire depth of the well was considered to be
homogeneous, as if it was one big flow unit, with one value of
permeability and one value of porosity for the entire depth.
2. The permeability values were obtained by taking a
geometrical average (equation 25) of the 170 plug core
measurements available. The porosity was calculated as a
arithmetic average (equation 26) of the core samples.

∑ log k i
log k j = i =1 ……..………….. (25)
n

∑φ h i i
……..…..………….(26)
φj = i =1
n

∑h
i =1
i

where kj is geometrical average permeability, ki is the


Figure 2: Division into flow units by using a GR cut-off of 90 API
permeability of core sample i, hi is the thickness represented for Model 2.
SPE 84277 7

Method to define Model 3: Flow Zone Indicator (FZI). 6. Flow unit number, depth interval, permeability and
The method for the flow unit zonation in Model 3 was porosity for each flow unit were tabulated.
developed using the Flow Zone Indicator (FZI) concept12. One of the important tools used in Model 4 is the Neural
According to Amaefule et al. 12, all samples with similar FZI Network Approach (NNA), which was used in a previous
values belong to the same flow unit because of the similarities work3, to generate synthetic permeability for the entire depth
in their pore throat attributes. The data sources for Model 3 of the well. The petrophysical neural network is most
were core plug measurements, air permeability (kair), commonly used to build a neural net in one well where all
Klinkenberg-corrected permeability (kkl), and helium porosity input and output data are known30. The created neural net is
for 170 selected samples. then applied to other wells in the field where the output curve
The steps followed by this method are listed below: is unknown. In the case of well CSM Strat Test #61, the
1. The FZI for each sample was calculated using kkl and neural net was used to model a pseudo-permeability in the
helium porosity. uncored portions of the well where no permeability data were
2. FZI ranges were established. A representative FZI and present. This use of neural network modeling in only one
color were assigned to each range. Each sample was given the well is slightly different than a typical use for the neural
corresponding color of its FZI value. network approach3.
3. The samples were grouped by depth taking into
account the predominant color in the different depth intervals.
These groups constituted the main flow units.
4. The main flow units were plotted on the gamma-ray
and porosity logs according to their depth intervals.
5. In order to interpret flow units in the uncored depths of
the well, the main flow units were extrapolated based on the
signature of the GR and the density logs.
6. Once the entire depth was divided into flow units, each
flow unit was assigned a number.
7. Permeability and porosity for the main flow units were
calculated taking the arithmetic average value of the core
measurements for the corresponding depth interval.
8. To obtain the porosity of the flow units corresponding
to uncored depths, an average value was calculated from the
density log at the corresponding depth interval.
9. Permeability of the flow units corresponding to
uncored depths was calculated with an equation developed by
Amaefule et al.12 and the arithmetic averaged porosity values,
calculated in the previous step.

Method to define Model 4: Pore Throat Radius r35. The


data source for defining Model 4 was synthetic permeability
generated by a Neural Network Approach (NNA), porosity
Figure 3: Classification according to pore throat size on the r35
from the density log and the pore throat radius (r35) defined versus depth plot; the green lines indicate the boundaries of the
by Winland’s equation14. The steps included in the method can predominant pore throat size in the depth interval.
be summarized as follows:
1. The r35 was calculated from k-φ values for the entire Method to define Model 5: Capillary Pressure Curves.
depth of the core using Winland’s equation. The permeability The data source for Model 5 was core data, mercury injection
had to be adjusted to air permeability as required by the data (27 capillary pressure curves), NNA permeability data,
equation by using a kair/kkl correlation. gamma-ray, and density logs corrected for shales. The steps
2. A plot of r35 as a function of depth was made (Figure followed to define this method are explained below:
3). The r35 radii were classified into megaporous, 1. The capillary pressure curves were normalized using
macroporous, mesoporous, microporous and nanoporous. the J(S)-function, and the k-φ values for each sample obtained
3. The entire depth was classified into mega, macro, from core data.
meso, micro or nano groups (Figure 3) by observing the 2. The J(S)-function was plotted as a function of bulk
predominant pore throat size at the different depths. volume mercury saturation (Sb). By visual inspection, the
4. The groups were interpreted as flow units and curves with similar shapes were grouped.
numbered according to depth, with the highest number 3. To interpret the flow units, both similarity in shape and
assigned to the deepest depth. location along the depth of the well were taken into account.
5. Permeability and porosity for each flow unit was Curves that were similar and located within a certain depth
calculated as the average of the NNA permeability values and interval were considered to be a primary flow unit. Five
density log values respectively for the corresponding primary flow units were defined.
depth interval. 4. To confirm this initial flow unit definition, the
petrophysical parameter ΨHg and the pore throat sorting (PTS)
8 SPE 84277

of the samples were calculated where applicable. These A plot of r35 obtained from capillary pressure analysis
values were compared to check that they were similar within versus depth was used to verify the flow units obtained by the
the flow units. other methods.
5. Permeability and porosity for the primary flow units
were calculated taking the geometric and arithmetic average
values respectively, of the core measurements for the
corresponding depth interval.
6. The primary flow units were extrapolated to the
uncored depths of the well by using the gamma-ray and the
corrected density logs. Secondary flow units were defined by
finding similar log signatures as the ones from the primary
flow units. For some of the primary flow units, the inclusion
of adjacent depths with similar log signature increased their
total depth interval.

Method to define Model 6: Cumulative Storage and


Capacity Curves. The flow units for Model 6 were
interpreted in a previous work3. The basic steps are
explained below.
1. Using Swanson’s24 correlation, permeability was
obtained from the capillary pressure measurements. The r35
values for the capillary pressure data points were calculated
as well.
2. Synthetic permeability in the parts of the well where no
Figure 4: Cumulative storage and flow capacity curves were
permeability data was present was created using the Neural calculated from the effective porosity and synthetic
Network Approach (NNA). The well logs and the points where 3
permeability, respectively .
permeability was known were input to build the neural
network. The core Klinkenberg-corrected permeability was 1.0
Flow Unit 1

used for this purpose. 35-99'

0.9
3. Effective porosity was obtained from nuclear magnetic Flow Unit 2

resonance (NMR) data. 0.8


99'-227'

4. Cumulative flow capacity (kh) and storage capacity Flow Unit 3


227'-353'
(φh) curves versus depth were created. A flow unit 0.7

interpretation was done by considering the inflections of the


curve as breaks between flow units. This led to seven flow 0.6

units (Figure 4). 0.5 Flow Unit 4


5. Permeability was calculated as the geometric average 353'-830'

of the NNA values and permeability was obtained taking an 0.4


arithmetic average of the density porosity log values corrected
for shales. 0.3

Flow Unit 5
0.2
Method to define Model 7: SMLP. The flow units for 830'-975'

Flow Unit 6
Model 7 were also interpreted in the previous work by 0.1
975'-1062'

Maglio-Johnson3. The following steps are continuing on from Flow Unit 7


1062'-1132'
Model 6. The flow unit interpretation using the cumulative 0.0
storage and flow capacity curves is considered as preliminary. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

6. Core permeability, core porosity, and the r35 values Cumulative Storage Capacity (phi x h)
were used to develop a multiple regression. The resulting Figure 5: Stratigraphic Modified Lorenz plot built using effective
equation (equation) was used to fill the in the missing points in porosity to create cumulative storage capacity and synthetic
the well with r35 data. permeability to create cumulative flow capacity. Interpreted flow
1
units are represented by straight line segments .

log r 35 = 0.624 log k − 1.247 log φ + 0.753 …… (27) Figure 6 presents an overview of the location of the flow
units versus depth of for Models 1 through 7. The different
7. A Stratigraphic Modified Lorenz Plot (SMLP) was patterns are used to indicate different flow units, however,
built plotting cumulative flow capacity (kh) versus cumulative they do not show the differences in properties. Models 2, 6
storage capacity (φh) (Figure 5). The flow unit boundaries and 7 have a total of seven flow units each, Models 3 and 5
were determined by inflections in the curve. have 9 flow units each, Model 4 has eight flow units and
Model 1 has one flow unit.
SPE 84277 9

k1, φ 1, h1 1
Injector Producer
Well Well

k2, φ 2, h2 2

k3, φ 3, h3 3

Figure 7: Example of subdivision of the reservoir into three flow


units that have different properties (modified from Willhite, 1996).

Hypothetical Reservoir. The reservoir was modeled as a


cross section with one water injector well at one end and one
oil producer well at the other end. Both were open-hole. The
cross section was 2000 ft long and 1114 ft thick, so that the
total area was 2.22*106 ft2 (51 acres approximately). The
datum depth was located at 3500 ft and the initial pressure at
this depth was 1500 psia. The reservoir and fluid properties
are listed in Table 1.

Table 1: Rock and fluid properties for the hypothetical reservoir


Figure 6: Overview of the location of the flow units in the well for Property Value Unit
all models. Irreducible Water Saturation (Swir) 25 %
Residual Oil Saturation (Sor) 30 %
Numerical Simulation Maximum water relative permeability (krw*) 0.20 md
As a means of comparing the “correctness” of the models Maximum oil relative permeability (krow*) 0.92 md
defined by different flow unit interpretations, numerical Oil Gravity (γo) 45.5 API
simulation was carried out. A base case that reflected “truth” Oil Formation Volume Factor (Bo) 1 RB/STB
was established, and was used as reference for the comparison. Oil Viscosity (µo) 2 cp
By “truth” we mean the base case model provided a fine grid Water Formation Volume Factor (Bw) 1 RB/STB
representation of the flow properties obtained from well logs, Water Viscosity (µw) 0.5 cp
which models no flow units grouping. The software used was *maximum value of curve
ECLIPSE 100, Black Oil Simulator, from GeoQuest
Schlumberger. Study Cases. Seven study cases were established for the
numerical simulation. One case was built to use as a
Simulation Model. The simulation model was built on a two- reference, and was named BaseCase. The seven cases were
dimensional, cartesian grid (x-z), with the x-direction built based on the previous models defined by the different
representing the width of the reservoir, and the z-direction flow definition methods. These were named Models 1
representing the thickness of the reservoir. The width and through 7, as previously discussed. A short name defining the
the length of the grid were constant for all study cases. method used was added to the name of the model to facilitate
A two-phase, water-oil system was selected for this study their differentiation (for example, Model 5-Pc).
because an immiscible process would be best for observing The base case model was built dividing the reservoir into
displacement differences in the reservoir characterization very fine layers of approximately 2.5 feet. The idea behind
models. The system was simulated as a linear waterflood, this was to describe in more detail the petrophysical properties
with one injector well and one producer well. A basic scheme of the well by using a finer grid. In this way, the influence of
of how the reservoir is represented in the simulation is shown flow and non-flow zones could be accounted for, and the
in Figure 7. The reservoir is divided into flow units, with effect of grouping into flow units would be reduced. The data
petrophysical properties ki, φi , and hi, which are different for source for porosity was the density log corrected for shales.
each flow unit. As an example, the reservoir in Figure 7, has The data source for permeability was the synthetic
been divided into three flow units. Note that flow unit 2 could permeability created by neural network approach (NNA) based
be a barrier. on minipermeameter measurements (NNKmp). There were
It is assumed that the displacement performance of the three reasons for this selection of permeability values: First,
reservoir is controlled primarily by vertical heterogeneities31. the synthetic data was available for the entire well at intervals
The initial saturation is uniform for the entire reservoir. One of 0.1 ft; second, the synthetic data created using the NNA
set of relative permeability curves is assumed to represent matched very well the measurements made using the
fluid/rock interactions. minipermeameter based on visual inspection, and third, the
synthetic data was very close to the permeability obtained by
10 SPE 84277

nuclear magnetic resonance (NMR) measurements, which is Discussion of Results. The sensitivity among the models was
one of the most reliable logs for obtaining permeability. done comparing recovery factors, cumulative oil production,
For the base case model, an average of the permeability oil and water production rates, water cut, and pressure. The
and porosity values was calculated at every 2.5 ft for the entire results are presented below.
depth of the well. For permeability, a geometric average of all Recovery Factor. A plot of recovery factor (RF) as a
the values was calculated. For porosity, an arithmetic average function of the injected pore volume is shown in Figure 8.
from the shale corrected porosity was taken. A total of 445 Table 3 shows that the recovery for all models is within a
layers was obtained, making the total number of cells 8900 range of 39-60%. Model 3-FZI and Model 4-r35 show
(20*1*445) for BaseCase. recovery factors far below the BaseCase, with a difference of
Models 1 through 7 had the same grid, i.e. the same 13% and 11%, respectively. All other models are within the
reservoir size and the same number of cells in x- and z normal recovery rate for a secondary recovery, such as
direction. The number of cells in the x-direction was 20, and waterflood in this case. The difference compared to BaseCase
the number cells in the z-direction was 100, making a total of is less than 10%.
2000 cells for the grid (20*1*100).
To represent the flow units, the number of cells in the z- 70

direction was distributed according to the number of flow


units in the models and their corresponding thickness. For 100 60

layers in the z-direction, the thickness of each cell was 11.14


feet. A summary of the properties of each flow unit within 50

each model, which were used as input for the simulation


model, is presented in Table 2. The vertical permeability was 40

calculated as one tenth of the horizontal permeability.


30

Table 2: Properties used in the simulation models. BaseCase


Model Flow Unit (i) kxi (md) kzi (md) φi (fraction) hi (ft) 20
Model 1-homogeneous
Model 2-GR
Model 1- Model 3-FZI
1 53 5.3 0.268 1114
homogeneous Model 4-r35
10 Model 5-Pc
1 28 2.8 0.180 179
Model 6-kh/phih
2 0.015 0.0015 0.052 135 Model 7-SMLP
3 48 4.8 0.202 250
0
Model 2-GR 4 0.015 0.0015 0.071 85
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
5 86 8.6 0.226 220
Injected PV
6 0.015 0.0015 0.024 73
7 8 0.8 0.131 172 Figure 8: Recovery factor as a function of injected pore volume
1 546 54.6 0.166 207 for all models.
2 0.005 0.0005 0.018 62
3 154 15.4 0.164 180
4 72 7.2 0.217 115
Table3: OOIP; cumulative oil production (Np) after 15 years and
Model 3-FZI 5 0.11 0.011 0.079 90 the difference with respect to BaseCase
6 33 3.3 0.213 160 At an Injected Pore Volume of 2
7 547 54.7 0.269 52 Case OOIP a)
R.F. Difference
8 0.03 0.003 0.033 78 (MMSTB)
9 9 0.9 0.131 170 (%) (%)
1 11 1.1 0.179 182 BaseCase 2.30 52 0
2 0.6 0.06 0.102 230
Model 1 2.30 60 8
3 137 13.7 0.225 151

Model 4-r35
4 2 0.2 0.071 86 Model 2 2.28 52 2
5 193 19.3 0.226 219
6 0.05 0.01 0.026 74 Model 3 2.29 39 13
7 90 9.0 0.187 83
Model 4 2.29 41 11
8 4 0.4 0.080 89
1 239 23.9 0.164 212 Model 5 2.31 55 3
2 0.01 0.001 0.015 57
3 114 11.4 0.183 260
Model 6 2.31 55 3
4 99 9.9 0.203 35 Model 7 2.30 57 5
Model 5-Pc 5 0.59 0.06 0.071 85
a)
6 154 15.4 0.213 170 with respect to BaseCase
7 285 28.5 0.270 50
8
9
0.09
32
0.009
3.2
0.027
0.131
70
175
Cumulative Oil Production, and Oil and Water
1 93 9.3 0.170 164 Production Rates. The general trend of the curves in the
2 11 1.1 0.077 150 cumulative oil production plot (Figure 9) is an initial linear
3 62 6.2 0.205 230
Model 6-kh/φh 4 4 0.4 0.105 120 increase for all the models. This is an effect of the producer
5 138 13.8 0.228 200 well being controlled by oil production rate. Later, because of
6 1 0.1 0.038 80
7 51 5.1 0.131 170
water breakthrough and the reservoir not being able to
1 80 8.0 0.157 63 maintain the pressure, the rate is controlled by bottom-hole
2 106 10.6 0.182 128 pressure. The differences in water breakthrough times and
3 7 0.7 0.056 126
Model 7-SMLP 4 35 3.5 0.179 467 production rates cause the curves to separate from each other.
5 16 1.6 0.151 155 These effects are more noticeable in Figure 10, which shows
6 110 11.0 0.181 87
7 25 2.5 0.080 88 oil production rate as a function of time. The oil production
rate is constant for a certain period of time and then begins to
SPE 84277 11

decline at different times depending on the model. In general


terms, the water production rate (Figure 11) increases rapidly
at the beginning and then stays constant throughout the rest of
the simulation period. Model 5-Pc and Model 6-kh/φh behave
similar to BaseCase regarding cumulative oil production and
oil production rate as a function of time (Figure 9 and Figure
10). In the figure of water production rate in time (Figure 11)
Model 5-Pc matches BaseCase very well.

Figure 10: Oil production rate for all models.

Figure 9: Cumulative oil production for all models.

The differences among the models are a consequence of


the differences in velocity of the flow unit flood fronts moving
through the reservoir. Model 1-homogeneous has one big
flood front, which slowly moves towards the end of the
reservoir. After it reaches the production well, breakthrough
occurs and oil production rate drops. Water flows easily
through the reservoir because it is homogeneous, and
constantly displaces oil. Therefore the production rate Figure 11: Water production rate for all models.
declines slowly. This also happens with Model 7-SMLP, Model 3-FZI and Model 4-r35 have the most and the
where although there is a division into layers (flow units), thickest barriers. As a consequence, they have the lowest
there is flow through most of its flow units. All the other lowest RF and cumulative oil production of all the models
models have flow units that act as barriers, which do not (Figure 8 and Figure 9). In Model 3-FZI, there is a big
contribute to flow. As a consequence, flow through the difference in the properties from one flow unit to the other.
reservoir is more difficult, and rate declines more rapidly. This contrast influences the overall flow through the reservoir.
The displacement of the flood fronts is affected by the The flow units of Model 4-r35 were defined using the same
thickness and the permeability of each flow unit. Model 2- data source as BaseCase. However, in the method for
GR, Model 5-Pc and Model 6-kh/φh have some flow units with defining Model 4-r35, there was a selection of properties
very low permeability (k < 1 md) compared to the other units, based on the predominant pore throat radius r35. Only
which act as barriers to the flow. Their behavior with regard permeability values corresponding to the predominant r35
to recovery factor is similar to BaseCase, because the latter were taken into account for the calculation, and this could
also has some low permeability layers. As mentioned before, have underestimated the average permeability for the well.
Model 7-SMLP has flow through all its flow units, and Water Cut. Water breakthrough takes place within 1000
therefore its RF is higher than for BaseCase. Model 1 - days for all cases as can be seen as the first sign of water
homogeneous has the highest RF and its curve is furthest from appearance in the water production rate plot (Figure 11). It
BaseCase in the cumulative oil production plot (Figure 9), can also be observed in a plot of water cut as a function of the
because flow takes place everywhere in the reservoir. injected pore volume of water (Figure 12). Water
The reason that Model 7-SMLP does not have low breakthrough in BaseCase occurs soon after water starts being
permeability zones is due to flaws in the method for flow unit injected, and this is caused by water flowing very fast through
definition for the model. According to Maglio-Johnson the some layers with high permeability (k ~ 400 md). In Model 3-
interpretation from the Stratigraphic Modified Lorenz Plot FZI, and Model 4-r35 high water cut is obtained at low
(SMLP) for Model 7-SMLP did not differentiate major shale injected pore volume. The volume of oil in the reservoir that
(low capacity) breaks3. can be displaced by water is reduced by the presence of flow
12 SPE 84277

barriers, and therefore the available volume is filled up


quickly. Model 1-homogeneous, on the other hand, has the
biggest volume to be filled up with water, and requires higher
injected pore volume to obtain the same water cut values as
BaseCase. Model 2-GR, Model 5-Pc, and Model 6-kh/φh have
similar water cut behavior as that of BaseCase.
1.0

0.9

0.8

0.7

0.6

0.5

0.4 BaseCase
Model 1-homogeneous
0.3 Model 2-GR
Model 3-FZI
0.2 Model 4-r35 Figure 13: Field pressure as a function of time for all models.
Model 5-Pc
0.1 Model 6-kh/phih
Model 7-SMLP
Summary of Results. The differences in flow unit
0.0 interpretation for the models are clearly reflected in the
0 0.2 0.4 0.6 0.8
Injected PV
1 1.2 1.4 1.6
simulation results. None of the models matches the behavior
of BaseCase. Model 5-Pc has the most similar behavior,
Figure 12: Water cut for as a function of injected pore volume for
all models. which could be due to the fact that mercury injection capillary
pressure analysis best reflects the pore characteristics/fluid
Pressure. The pressure of the reservoir as a function of behavior within the rock, and therefore constitutes an
time is shown in Figure 13. The general trend is a fast important tool in interpreting flow units. Model 6-kh/φh also
increase at the beginning of the production period as a behaves similar to BaseCase, which is very important, because
consequence of the pressure from water injection. The it is the model that best honors the geology. Model 2-GR is
pressure required to start moving the flood front is high and the next most similar in behavior to BaseCase in terms of
decreases slowly with time by the effect of water breakthrough recovery factor and water cut, but not in the other plots. This
and the water moving through the reservoir. The pressure is probably due to the properties obtained through the linear
drops to low values as the flood front reaches the end of the correlation of permeability and porosity. However, it is
reservoir and the rate becomes a function of pressure. These interesting to mention that in both Model 6-kh/φh and Model
low pressures are then maintained at constant values during 2-GR the primary tool for interpreting the flow units was the
the rest of the simulation time. It is clearly shown in Figure gamma-ray log, and their division into flow units is almost
13 that the pressure maintenance in all models is different. identical. However, permeability and porosity within the
Model 2-GR depletes the fastest, because the average flow units is different due to differences in data source and
permeability is the lowest, while Model 3-FZI maintains the permeability averaging.
pressure for a longer period of time due to higher average
permeability. The weighted average permeability with respect Conclusions
to depth for each model is presented in (Table 4). For the This paper has given some answers to the importance of flow
period of time where the flow rate is kept constant, a higher unit interpretation in the prediction of flow performance of a
average permeability represents higher capacity of the fluid to reservoir. The following conclusions are presented:
flow through the reservoir and therefore requires less pressure 1. Interpretation of flow units based on petrophysical
difference. On the other hand, when the average permeability properties, logs, and stratigraphy are routinely done for
is low, flow through the reservoir is more difficult, and reservoir characterization. Unless numerical simulation is
requires a larger difference in pressure. carried out to confirm the flow unit assignment, errors can
occur. The use of erroneous flow models can lead to
Table 4: Weighted average of permeability for the models. inaccurate predictions of flow in the reservoir.
Model Weighted k Average (md) 2. Results of the numerical simulation are a strong function
BaseCase 116
of average permeability, and therefore of the flow unit
1 53
2 33 definition method used. The flow capacity of a reservoir
3 165 determines the rate with which fluid can be recovered.
4 66 3. All methods used in this research for flow unit
5 117
interpretation derive from the equation of a straight line y =
6 61
7 45 mx + b, where y is a function of permeability, x is a function
of porosity, and the constants m and b are a function of the
rock type.
SPE 84277 13

4. Empirical equations for pore throat radius such as µm


Winland’s r35, which is based on mercury injection capillary RF = recovery factor, (%)
pressure measurements, cannot be universally applied. The rp = pore throat radius, µm
best correlation based on the available data has to be RQI = Reservoir Quality Index, µm
established for an individual reservoir. Sb = mercury saturarion relative to bulk volume,
5. This research demonstrates the importance of fraction
integrating the work of geologists and petroleum engineers in Sb∞ = bulk volume occupied by mercury, %, fraction
reservoir characterization and flow performance projects. Sgv = specific surface area per unit grain volume
Geologists need to account for flow behavior in the SHg = mercury saturation, fraction
description of a reservoir, while petroleum engineers need to SMLP = Stratigraphic Modified Lorenz Plot
match geological characterization for simulating Sor = residual oil saturation, fraction
flow performance. Sw = water saturation, fraction
Swir = irreducible water saturation, fraction
Acknowledgements Vb = bulk volume, cc
This paper is a partial review of a Master of Science thesis Vg = grain volume, cc
prepared at the Petroleum Engineering Department of the Vp = pore volume, cc
Colorado School of Mines. The authors would like to thank ΨHg = petrophysical parameter defined by Wells and
Dr. John Fanchi, Dr. Neil Hurley, and Steve Goolsby for their Amaefule27
valuable contribution. φe = effective porosity, fraction
φh = storage capacity, ft
Nomenclature
φz = Normalized Porosity Index, pore volume-to
A, B, = constants
grain volume ratio, dimensionless
Bo = oil formation volume factor, RB/STB
φι = porosity for flow unit i, fraction
Bw = water formation volume factor, RB/STB
Fs = shape factor µw = water viscosity, cp
FU = flow unit µo = oil viscosity, cp
FZI = Flow Zone Indicator, µm τ = tortuosity, dimensionless
G = pore geometrical factor, dimensionless
GR = gamma-ray log References
h = height (or thickness), ft 1. Stolz, A.-K.: “Sensitivity Study of Flow Unit Definition by
J(S) = Leverett’s19 J-function as a function of Use of Reservoir Simulation”, MS Thesis, Colorado School
of Mines, Golden, CO (2003).
saturation, dimensionless
2. Slatt, R.M., and Hopkins G. L.: “Scaling Geologic Reservoir
k = permeability, md, µm2 Description of Engineering Needs,” JPT (February 1990).
kair = uncorrected air permeability, md 3. Maglio-Johnson, T.: “Flow unit Definition using
kh = flow capacity, md-ft Petrophysics in a Deep Water Turbidite Deposit, Lewis
ki = permeability of flow unit i, md Shale, Carbon County, Wyoming,” MS Thesis, Colorado
kj = average permeability based on thickness, md School of Mines, Golden, CO (2001).
kkl = Klinkenberg-corrected permeability, md 4. Pyles, D..: “A High-Frequency Sequence Stratigraphic
krow = oil relative permeability to water, dimensionless Framework for the Lewis Shale and Fox Hills Sandstone,
Great Divide and Washakie Basins, Wyoming,” MS Thesis,
krow* = maximum value of the oil relative permeability
Colorado School of Mines, Golden, CO (2000).
to water curve, dimensionless 5. Witton, E.: “Outcrop and Subsurface Characterization of the
krw = water relative permeability, dimensionless Lewis Shale, Carbon County, Wyoming,” MS Thesis,
krw* = maximum value of water relative permeability Colorado School of Mines, Golden, CO (1999).
curve, dimensionless 6. Goolsby, S.: “Petrophysical and Reservoir Properties of
kxi = permeability in the horizontal direction for flow Turbidite Sandstones in the Lewis Shale of the Green River
unit i, md Basin, Wyoming,” unpublished PhD Dissertation, Colorado
kzi = permeability in the vertical direction for flow School of Mines, Golden, CO (in progress).
unit i, md 7. Jones, S.C.: “The Profile Permeameter: A New, Fast,
Accurate Minipermeameter,” paper SPE 24757 presented at
m,b = constants
the 1992 SPE Annual Technical Conference and Exhibition,
NMR = Nuclear Magnetic Resonance Washington D.C., Oct. 4-7.
NNA = Neural Network Approach 8. Ebanks, W.J., Scheihing, M.H., and Atkinson, C.D., 1992:
NNKmp = NNA created using core minipermeameter “Flow Units for Reservoir Characterization,” in: Morton-
permeability, md Thompson, D., and Woods, A.M., 9eds.), Development
OOIP = original oil-in-place Geology Reference Manual: AAPG Methods in Exploration
Pc = capillary pressure, psia Series No. 10, pp. 282-285.
Pd = extrapolated mercury-air capillary displacement 9. Asquith, G. B., with Gibson, C. R.: Basic Well Log Analysis
pressure, psia for Geologists, Methods in Exploration Series, American
PTS = Pore Throat Sorting 28 Association of Petroleum Geologists, Tulsa (1982).
R2 = correlation coefficient, dimensionless
r35 = pore throat radius at 35% mercury saturation,
14 SPE 84277

10. Kozeny, J.: “Über Kapillare Leitung des Wassers im Boden, 26. van Genuchten, M. T.: “A Closed-Form Equation for
Sitzungsberichte,” Royal Academy of Science, Vienna, Predicting Hydraulic Conductivity of Unsaturated Soils,”
Proc. Class I (1927), v. 136, pp. 271-306. Soil Sci. Soc. Am. J. (1960), vol. 44, 492-498.
11. Carman, P.C.: “Fluid Flow through Granular Beds,” Trans. 27. Wells, J.D., and Amaefule, J.O.: “Capillary Pressure and
AIChE (1937), v. 15, pp. 150-166. Permeability Relationships in Tight Gas Sands,” paper SPE
12. Amaefule, J.O., Altunbay, M., Tiab, D., Kersey, D.G., and 13879 presented at the 1985 SPE Low Permeability
Keelan, D.: “Enhanced Reservoir Description: Using Core Reservoirs Symposium, Denver, CO, May. 19-22.
and Log Data to Identify Hydraulic (Flow) Units and Predict 28. Jennings, J.B.: “Capillary Pressure Techniques: Application
Permeability in Uncored Intervals/Wells,” paper SPE 26436 to Exploration and Development Geology,” AAPG Bulletin,
presented at the 1993 SPE Annual Technical Conference Vol. 71, No. 10, October 1987, pp. 1196-1209.
and Exhibition, Houston, TX, October 3-6. 29. Gunter, G.W., Finneran, J.M., Hartmann, D.J., and Miller,
13. Pittman, E.D.: “Relationship of Porosity and Permeability to J.D.: “Early Determination of Reservoir Flow Units Using
Various Parameters Derived from Mercury Injection- an Integrated Petrophysical Method,” paper SPE 38679
Capillary Pressure Curves for Sandstone,” AAPG Bull., presented at the 1997 SPE Annual Technical Conference
(1992) 191-198. and Exhibition, San Antonio, TX, Oct. 5-8.
14. Kolodzie, S.: “Analysis of Pore Throat Size and Use of the 30. Olson, T. M.: “Porosity and Permeability Prediction in Low
Waxman-Smits Equation to Determine OOIP in Spindle Permeability Reservoirs from Well Logs using Neural
Field, Colorado,” paper SPE 9832 presented at the 1980 Networks,” paper SPE 39964 presented at the 1998 SPE
Annual Technical Conference and Exhibition of the Society Rocky Mountain Regional/Low Permeability Reservoir
of Petroleum Engineers of AIME, Dallas, TX, Symposium and Exhibition, Denver, Colorado, 5-8 April.
September 21-24. 31. Willhite, G. P.: Waterflooding, SPE Textbook Series
15. Martin, A.J., Solomon, S.T., and Hartmann, D.J.: Vol. 3 (1986).
“Characterization of Flow Units in Carbonate Reservoirs,”
AAPG Bulletin (May 1997), Vol. 81, No. 5, 734-759.
16. Al-Fossail, K.A., Saner, S., Asar, H.K., and Hossain, M.: SI Metric Conversion Factors
“Factor Affecting Mercury Capillary Pressure Behavior of
Saudi Arabian Carbonate Reservoir Rocks,” (1991),
acre X 4.046 873 E+03 = m2
Proceedings of the 7th Middle Eastern Oil Show, ft X 3.048 E-01 =m
Bahrain, 815-820 ft2 X 9.290 304 E-02 = m2
17. Porras, J.C., and Campos, O.: “Rock Typing: A Key for psi X 6.894 757 E+00 = kPa
Petrophysical Characterization and Definition of Flow
Units, Santa Barbara Field, Eastern Venezuela Basin,” paper
SPE 69458 presented at the 2001 SPE Latin American and
Caribbean Petroleum Engineering Conference held in
Buenos Aires, 25-28 March.
18. Ausbrooks, R., Hurley, N.F., May, A., and Neese, D.G.:
“Pore-Size Distributions in Vuggy Carbonates from Core
Images, NMR and Capillary Pressure,” paper SPE 56506
presented at the 1999 SPE Annual Technical Conference
and Exhibition held in Houston, Texas, 3-6 October.
19. Leverett, M.C.: Capillary Pressure in Porous Solids, Trans.,
AIME, Vol. 142 (1941) 341-358.
20. Collins, R. E.: Flow of Fluids through Porous Materials,
Research & Engineering Consultants, Inc., Englewood,
Colorado (1990).
21. Leverett, M.C.: Flow of Oil-Water Mixtures through
Unconsolidated Sands,” Trans. AIME, Vol 132 (1939) 149
22. Ma, S., Jiang, M-X., and Morrow, N.R.: “Correlation of
Capillary Pressure Relationships and Calculations of
Permeability,” paper SPE 22685 presented at the 1991 SPE
Annual Technical Conference and Exhibition of the Society
of Petroleum Engineers, Dallas, TX, October 6-9.
23. Thomeer, J.H.M., “Introduction of a Pore Geometrical
Factor Defined by the Capillary Pressure Curve,” Petroleum
Transactions, AIME (1960) Vol. 219, 354-358.
24. Swanson, B. F.: “A Simple Correlation between Air
Permeabilities and Stressed Brine Permeabilities with
Mercury Capillary Pressures,” paper SPE 8234 presented at
the 1978 Annual Fall Technical Conference and Exhibition
of the Society of Petroleum Engineers of AIME, Houston,
TX, October 1-3.
25. Brooks, R. H., and Corey, A. T.: “Hydraulic Properties of
Porous Media,” Hydrology Papers No. 3, Colorado State
University, Ft. Collins, CO (1964).

You might also like