You are on page 1of 16

SPE-199122-MS

New Integrated Capillary Pressure and Initial Water Saturation Modeling

Yizhu Liao, Applied Reservoir Technologies

Copyright 2020, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Latin American and Caribbean Petroleum Engineering Conference originally scheduled to be held in Bogota,
Colombia, 17 – 19 March 2020. Due to COVID-19 the physical event was postponed until 27 – 31 July 2020 and was changed to a virtual event. The official proceedings
were published online on 20 July 2020.

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 have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is 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 acknowledgment of SPE copyright.

Abstract
This paper presents a new capillary pressure model, integrated and consistent capillary pressure modeling,
and initial fluid saturation along wells and in reservoirs.
The new capillary pressure model is obtained by fitting Pc curves, end Pc values at end saturations,
and threshold entrance Pc values satisfactorily. Analytical Pc derivative is then obtained. A consistent Pc
derivative value can be used to obtain irreducible water saturation and corresponding maximum capillary
pressure. After Pc curve fitting for multiple cores, correlations can be developed for the parameters in the
new Pc model with different reservoir rock types and reservoir properties.
The new capillary pressure model is then used with free water level and reservoir fluid gravity to obtain
the initial water saturation at any place in the reservoir. The initial water saturation vs. depth obtained
dynamically from the new Pc model is integrated with that obtained statically from logs to achieve the
integrated, reconciled, and consistent initial water saturation vs. depth.
Real examples are presented to show the step by step workflow procedure to obtain the new capillary
pressure model and the integrated initial water saturation vs. depth. These examples include SCAL Pc curve
data from cores and log Swi data vs. depth from wells in two offshore reservoirs.
This paper has three significant technical contributions. The new capillary pressure model is more
versatile than those published and can handle Pc data while other existing Pc models cannot. The analytical
Pc derivative from the new capillary pressure model can provide a consistent criterion for obtaining
irreducible water saturation and maximum capillary pressure values for different rocks. The new capillary
pressure model, when integrated with initial water saturation vs. depth from logs, can provide a consistent
initial fluid saturation distribution in integrated static and dynamic reservoir modeling.

Introduction
It is essential to have equations to calculate initial water saturation in reservoirs. Three Pc models are popular
in literature and industry use: Thomeer, Brooks-Corey, and Lambda models.
Thomeer model:
(1)
2 SPE-199122-MS

or,
(1a)
or,
(1b)
Brooks-Corey model:
(2)
or,
(2a)
or,
(2b)
Lambda model:
(3)
Leverett J function:
(4)
J is a dimensionless Pc function.
Je is an entry J value corresponding to the threshold entry capillary pressure pce. .
Each of the above three models have 3 parameters. They are G, Je, and Swir for Thomeer model; n, Je,
and Swir for Brooks-Corey model; and λ, a, and b for Lambda model. G, n, and λ describe the shape of Pc
curve in each model.
However, these three models require that the entry capillary pressure pce (i.e., entry J function Je) be
greater than zero. If pce (i.e., Je) is zero, none of these models is valid. In addition, another common problem
with these models is that it is usually hard to achieve a good fit for a complete Pc curve that includes the
start (entrance Pc), the shape (curvature), and the end (maximum Pc).
This paper presents a new capillary pressure model that overcomes all these problems. The new Pc model
can fit any capillary pressure data (start, curvature, end) very well even if the entrance capillary pressure
pce is zero. The new Pc model also presents an excellent method to consistently estimate the irreducible
water saturation and the maximum capillary pressure for any samples especially if the end Swir value is not
reached during the lab test, and to define representative Pc curve shapes for different rocks. The new Pc
model, when integrated with initial water saturation vs. depth from logs, provides a consistent initial fluid
saturation distribution in integrated static and dynamic reservoir modeling.
Even if the initial water saturation has been determined very reliably by log interpretations along the
wells, it is still essential to use capillary pressure modeling not only to integrate with logs and other data
but also to populate the initial water saturation in 3D static and dynamic models. A good log interpretation
may have already considered the effects of many factors properly like lithology, mineralogy, cementation,
facies, wettability, reservoir fluids, porosity, permeability, etc. along wellbores, but it is not correct to use
geostatistics to populate a 3D model from initial water saturation results along wellbores. Even if some other
method like interpolation or extrapolation than geostatistics is used to populate a 3D model from initial
water saturation results along wellbores, the initial water saturations populated by any static method cannot
guarantee the initial reservoir equilibrium. Thus, integrated capillary pressure modeling is essential.
SPE-199122-MS 3

New Capillary Pressure Model


The new capillary pressure model for primary drainage is:

(5)

Three parameters a, b, and λ in the new Pc model can be obtained by fitting Pc data.
PcD can be expressed as:

(6)

SwD can be expressed as:

(7)

Eqns. 5, 6 and 7 are the same new Pc model; it is just written differently.
SwD is dimensionless water saturation and PcD is dimensionless capillary pressure.

(8)

(9)

For primary drainage, since Swir≤Sw≤1, 0≤SwD≤1. Since 0≤pce≤pc≤pcmax, 0≤pceD≤pcD≤1.


Note that the entrance capillary pressure pce can be zero or greater than zero.
At SwD=0, PcD=1, Eqn. 5 is automatically satisfied.
At SwD=1, PcD= PceD; PceD (entrance Pc) can be zero or greater than zero.
The new Pc model (Eqn. 5) is different from the Lambda model (Eqn. 3), even though three parameters
a, b, and λ are also used. This new Pc model is a modified hyperbolic equation; the hyperbolic feature can
characterize the Pc curve shape and the modified feature can make it more flexible and more versatile to fit
many shapes and kinds of Pc data. λ can be called a shape factor.
The new Pc model has several advantages. First, it can fit any Pc data very well. It can handle zero
entrance capillary pressure very well while others cannot. Second, it provides an excellent and consistent
way to estimate maximum capillary pressure and irreducible water saturation for any core sample, even if
these endpoint values are not achieved in lab tests. Third, it can be used to define Pc curves for all rocks if
some correlations for a, b, Swir, and pcmax with other rock properties have been found. Fourth, when integrated
with initial water saturation vs. depth from logs, it provides a consistent initial fluid saturation distribution
in integrated static and dynamic reservoir modeling.

Procedure to Obtain the New Capillary Pressure Model


From Eqn. 5, λ can be expressed as

(10)

From PcD= PceD at SwD=1, λ can be expressed as

(11)
4 SPE-199122-MS

If parameters a and b are determined, λ can be calculated by Eqn. 11. Then, the new Pc model is obtained.
From Eqns. 10 and 11, we have

(12)

Define

(13)

(14)

Plot Y vs X. The best "a" and "b" values would give a best unit slope straight line fit of Y vs. × plot.
The "a" and "b" values can be obtained graphically by trial and error, or numerically by optimization, or by
graphically and numerically combined. Here is the procedure.
1°: Tabulate, plot, and visually inspect lab data of pc vs. Sw.
2°: Note the last capillary pressure pcmax at the last minimum Swmin from the lab data.
The last minimum Swmin may not be the "true" Swir. For now, just consider Swmin as Swir.
3°: Calculate dimensionless water saturation SwD with Eqn. 8, and dimensionless capillary pressure
pcD with Eqn. 9.
4°: Decide the entrance capillary pressure pce at Sw=1.0.
The pce value is the intersect pc value of the logical extension of the main Pc curve trend at Sw=1.0.
Thus, the pce value is not necessarily the lab entrance pc value. It can be zero or greater than zero.
5°: Calculate X with Eqn. 13, and Y with Eqn. 14.
6°: Plot Y vs. X. Try different values of a and b until a unit slope straight line is achieved; this is a
trial and error process graphically. Use naked eyes to judge.
Or this can be done numerically below.
Calculate the cumulative square difference between Y and X for all data points as objective
function:

(15)

Use a mathematical "Solver" in excel to minimize the objective function, i.e., to get its value
as close to zero as possible. In this process, if there is any problem in numerical calculation or
convergence, just use a different value of a or b to go through the minimization process.
7°: Calculate λ with Eqn. 11 and the obtained pceD, a, and b.
8°: Calculate pcD vs. SwD with Eqn. 6 and the obtained a, b, and λ.
9°: Plot the lab data pcD and the just obtained model pcD vs. SwD to check the quality of the Pc curve fit.
The plot can also be made in actual variables: the lab data pc and the obtained model pc vs. Sw.
Table 1 presents a real core example to show the Pc curve fitting by graphical method.
SPE-199122-MS 5

Table 1—SCAL Pc Data Curve Fitting by Graphical Method

Fig. 1 presents the graphical search for "a" and "b" values to fit Pc data. The graph is good not only
for finding "a" and "b" values, but also for checking the data quality and reliability, whether graphical or
numerical method is used. The results: pceD =0.007, a=0.005, b=800, λ =1.50971. Fig. 2 presents the Pc data
and the fit with the new Pc model: (i) is pcD vs. SwD plot, and (ii) is pc vs. Sw plot. Note the new Pc model
gives excellent fit to the Pc data.

Figure 1—Search for a and b Values by Graphical Method.

Figure 2—New Pc Model Fits Data by Graphical Method.

The graphical method provides an intuitive way to see the quality and consistency of the raw Pc data.
This way to find "a" and "b" values may be a little subjective and may lead to "sub-optimal" sets of "a" and
"b" values. The numerical method provides a "rigorous and accurate" way to obtain the "optimal" values of
"a" and "b" if the raw Pc data are consistent and have high quality. However, if the raw Pc data quality is
not good or not consistent, the numerical method may lead to some impractical "a" and "b" values.
6 SPE-199122-MS

The recommended method is a combined method of the two. It takes their advantages and overcomes
their disadvantages.
For curve fitting of Pc data, it is recommended to use the following guidelines.
First, always plot the raw Pc data to see its quality and consistency. Plotting can help decide the proper
pce or pceD value, and see the data reliability and the Pc curve features. The less reliable data points can be
ignored or given less weights in curve fitting.
Then, use numerical and/or graphical method any time until a good curve fit is achieved. Thus-obtained
values of pceD, a, b, and λ should be better than those obtained by graphical or numerical method alone.
Finally, always plot Y vs. X to see if a unit slope straight line is achieved. And always plot pcD vs. SwD,
and pc vs. Sw, for raw Pc data and the fitted Pc model to check the quality of the Pc curve fit.

Procedure to Obtain Maximum Capillary Pressure and Irreducible Water


Saturation
It seems that there was no consistent industry standard for maximum capillary pressure (pcmax) and
irreducible water saturation (Swir). Often the last data point was used for pcmax and Swir, which were closely
related to test conditions.
Now with the new Pc model, we can obtain maximum capillary pressure (pcmax) and irreducible water
saturation (Swir) for different cores in a consistent and fair way.
By definition, if an irreducible water saturation (Swir) is reached, any attempt to further reduce the water
saturation will lead to a tremendous (infinite) increase in capillary pressure (pcmax). In practice, a "huge"
increase in Pc from a small Sw reduction, i.e., a big derivative value of dpc/dSw, can be used to obtain "true"
irreducible water saturation (S*wir) and "true" maximum capillary pressure (p*cmax).

(16)

From Eqn. 16, after curve-fitting Pc data, dpc/dSw can be calculated analytically for each data point.
Reversely, for any given (dpc/dSw)* value, the corresponding S*w and p*c can be calculated below.
From Eqn. 16,

(17)

(18)

Thus,
(19)
(20)
Note that in this process, Swir and pcmax are lab data points: pcmax is the maximum capillary pressure at the
minimum water saturation Swir. SwD and pcD are defined based on Swir and pcmax.
The procedure to get "true" irreducible water saturation (S*wir) and maximum capillary pressure (p*cmax)
is below:
1°: choose a "large" value of (dpc/dSw)*
2°: calculate S*wD by Eqn. 17, p*cD by Eqn. 18
SPE-199122-MS 7

If the calculated S*wD>0, then "true" S*wir will be greater than test value Swir (S*wir>Swir), and "true"
p*cmax will be less than test value pcmax (p*cmaxp<cmax)
If the calculated S*wD<0, then "true" S*wir will be less than test value Swir (S*wir<Swir), and "true"
p*cmax will be greater than test value pcmax (p*cmax>pcmax).
3°: calculate S*wir by Eqn. 19, p*cmax by Eqn. 20.
There can be some flexibility in choosing a "large" value of (dpc/dSw)*, but for given SCAL data and
reservoir situations, a reasonable consensus can be chosen in work process. Once a consensus (dpc/dSw)*
value is decided, it can be used for all cores to get S*wir and p*cmax values consistently.
For example, if (dpc/dSw)* = −6000 psi is used, at S*wir capillary pressure pc need to increase by 60 psi

to reduce water saturation Sw by 0.01 (e.g., from 0.19 to 0.18), i.e., . The
negative sign means that capillary pressure increases as water saturation decreases.
For a core presented in Table 1, dpc/dSw is in the last column. For this data set, Swir=0.227, pcmax=107.564
psi. If (dpc/dSw)*=-6000 psi is used, using Eqns. 17 through 20 will lead to S*wD= 0.00442, S*wir=0.2304,
p*cmax=39.17 psi. Note that since S*wD>0, the "true" S*wir is greater than test value Swir (S*wir>Swir), and the
"true" p*cmax is less than the test value pcmax (p*cmax < pcmax.). Inspection of the dpc/dSw column and Fig. 2
also clearly shows this.

Impact of Parameter Values on Capillary Pressure Curves


There are three parameters a, b, and λ in new capillary pressure model Eqn. 18, but λ is not independent
because it can be determined from "a" and "b" by Eqn. 11.
Fig. 3 shows pore size distributions for four cases: 1) less homogeneous ("a" large) and small size ("b"
small) pores, 2) more homogeneous ("a" small) and small size ("b" small) pores, 3) less homogeneous ("a"
large) and large size ("b" large) pores, 4) more homogeneous ("a" small) and large size ("b" large) pores.
Their capillary pressure curves are shown in Fig. 4.

Figure 3—Pore Size Distributions (Pore Homogeneity and Pore Sizes).


8 SPE-199122-MS

Figure 4—Impact of Pore Homogeneity and Pore Sizes on Pc Curves.

Parameter "a" value change moves the Pc curve up and down "vertically". As "a" increases, the Pc curve
moves up and looks steeper. As "a" decreases, the Pc curve moves down and looks flatter.
Parameter "b" value change moves the Pc curve to the left and right "horizontally". As "b" increases, the
Pc curve moves to the left. As "b" decreases, the Pc curve moves to the right.
A more homogeneous pore rock (smaller "a") has a flatter Pc curve, and a less homogeneous pore rock
(larger "a") has a steeper Pc curve. A larger size pore rock (larger "b") has a Pc curve shifted more to the
left, and a smaller size pore rock (smaller "b") has a Pc curve shifted more to the right. The curvature (shape
factor λ) of the Pc curve is a combined result from homogeneity ("a") and pore size ("b") through Eqn. 11.
Reservoir quality is a combined result of pore size ("b" value) and its distribution (homogeneity, i.e.,
sorting) ("a" value). A better reservoir quality reservoir is more homogeneous (smaller "a") and has larger
pore size (larger "b"), while a worse reservoir quality reservoir is less homogeneous (larger "a") and has
smaller pore size (smaller "b"). Of course, a reservoir can be homogeneous (small "a") and has small pore
size (small "b"), or can be heterogeneous (large "a") and has large pore size (large "b").
A more homogeneous, larger pore size rock (smaller "a", larger "b") gives a smaller Swir value and a
smaller Pc value, and the Pc curve is shifted to the left and down. A less homogeneous, smaller pore size
rock (larger "a", smaller "b") gives a larger Swir value and a larger Pc value, and the Pc curve is shifted
to the right and up.
Table 2 summarizes the impact of pore homogeneity (parameter "a") and pore sizes (parameter "b") on
Pc curves.
SPE-199122-MS 9

Table 2—Impact of Pore Homogeneity and Pore Sizes on Capillary Pressure Curves

Capillary Pressure Model Parameter Correlations for Different Rock Types


Correlations for Pc model parameters like a, b, Swir, and pcmax with other rock properties can be developed
for a reservoir if adequate Pc data are available. This section takes Pc data from a well's cores as an example
to show such correlations. Similar correlations can be worked out for other wells and reservoirs.
Once these correlations are obtained, the Pc curves for all rocks for the reservoir can be defined.
Table 3 presents capillary pressure model parameter values of a, b, S*wir, p*cmax for 13 cores from a well.
These parameter values are obtained from graphical method; numerical method also gives similar results
and trends. A big value for (dpc/dSw)* (-6000 psi) was used to get endpoint values S*wir and p*cmax. Note pceD
is an input value for Pc model, not an output result. Centrifuge primary drainage Pc tests were conducted
on all 13 cores.
10 SPE-199122-MS

Table 3—Pc Model Parameter Values from Cores of an Example Well

Figs. 5 and 6 plot Pc model parameter "a" and "b" values vs. FZI respectively. As FZI increases, i.e., as
rock quality increases, "a" will decrease and then stabilize, and "b" will increase and then stabilize. This
is expected as discussed before.

Figure 5—Pc Model Parameter "a" vs. FZI.

Figure 6—Pc Model Parameter "b" vs. FZI.

Figs. 7 and 8 plot S*wir and p*cmax vs. FZI respectively. As FZI increases, i.e., as rock quality increases,
both S*wir and p*cmax will decrease.
SPE-199122-MS 11

Figure 7—S*wir vs. FZI Plot.

Figure 8—p*cmax vs. FZI Plot.

Capillary Pressure Curves and Saturation-Height Relationships


Capillary pressure curves can be defined for different rock types of a reservoir once the correlations for
Pc model parameters "a", "b", S*wir, p*cmax with other rock properties are obtained for the reservoir. The
correlations for the Pc model parameters need to be specific and applicable for the given reservoir.
To obtain the capillary pressure and initial water saturation at any point in a reservoir, in addition to the
Pc model parameters "a", "b", S*wir, p*cmax, it is necessary to know FWL (free water level), the distance from
the specific point to the FWL, the in-situ oil and water gravity in the oil and water zones.
The procedure to determine the capillary pressure and initial water saturation is below. This can be
implemented in any static modeling like Petrel, and in any dynamic modeling like Eclipse.
1°: Set Swi=1 for any point at or below OWC. (OWC=FWL if entry Pc value is zero).
2°: Calculate the vertical distance for any point above FWL. Height (h) = Depth – FWL.
(Note: height above FWL is always positive, meter or feet ssTVD).
3°: For any point above FWL, calculate "a", "b", S*wir, p*cmax from the correlations obtained for the
rock type of the reservoir.

Calculate λ from Eqn. 11, . If pceD is not known, assume it zero.

4°: For any point above FWL, calculate capillary pressure pc.
pc = 1.4223344hΔγ, or pc = 3.28084hΔG, where Δγ is the in-situ gravity difference between water
and oil in the reservoir in g/cm3. ΔG is the pressure gradient difference between water and oil in the
reservoir in psi/ft. h is height above FWL, m ssTVD. pc is the capillary pressure in psi. (Note pressure
gradient and gravity are easily convertible, G = 0.4335275 γ, ΔG = 0.4335275 Δγ; γ = 2.30666 G,
Δγ = 2.30666 ΔG, where γ, Δγ in g/cm3, and G, ΔG in psi/ft.)
Or, pc = 0.4335275 hΔγ, or pc = hΔG, where pc in psi, h in ft ssTVD, Δγ in g/cm3, ΔG in psi/ft.
Or, pc = 9.8 hΔγ, where pc in KPa, h in meter ssTVD, Δγ in g/cm3.
12 SPE-199122-MS

(Note capillary pressure pc and height above FWL (HAFWL, h) are easily convertible.

where pc in psi, Δγ in g/cm3, h in ft. or where pc in KPa, Δγ in g/cm3,


h in meter.)
5°: For any point above FWL, calculate pcD = pc/p*cmax.

If pcD≥1, set Swi = S*wir. Otherwise, use Eqn. 7 to calculate .


Then calculate Swi =(1-S*wir) SwD + S*wir.
Thus, for oil-water system, the initial water saturation at any and every point is determined. Once Swi is
obtained, the initial oil saturation is Soi=1-Swi.

Integration and Reconciliation of Swi from New Pc Model and Logs


Integration, interaction, iteration, and reconciliation of all different data, models, interpretations, and results
is essential and most important to achieve consistent initial water saturation modeling in static and dynamic
modeling. Fig. 9 presents a workflow for integration and reconciliation of initial water saturation from new
Pc model and log interpretation. Even one further step is used, i.e., integration and comparison of original
hydrocarbon in place in 3D static and dynamic models.

Figure 9—Integration and Reconciliation Workflow of Swi and OHCIP from New Pc Model and Log

The above workflow is first used to determine the capillary pressure and initial water saturation at
any point along a well. The Swi from Pc modeling is compared with the Swi from log interpretation.
Reconciliation and iteration are involved to resolve and explain any possible difference in Swi from these
two methods.
Here are the results for the well. The correlations for "a", "b", and pcmax vs. FZI are the same as presented
in Figs. 5, 6, and 8. Fig. 10 presents the correlation of Swir vs. FZI after integration with other data.
SPE-199122-MS 13

Figure 10—Swir vs. FZI Relationship

The base trends (dark lines) of "a", "b", pcmax, Swir in Figs. 5, 6, 8, 10 are used in Pc and Swi calculation.
For the upper zone of the example well, the in-situ oil and water gravity are 0.84 and 1.16 g/cm3
respectively.
Fig. 11 presents the initial water saturation (Swi) from Pc modeling along this well (ssTVD). The
irreducible water saturation (Swir) from Pc modeling is also included in this figure. For comparison, Swi
from log interpretation is also included. The colors are: blue for Swi from log, green for Swi from Pc model,
red for Swir from Pc model.
For this well in this zone, the OWC is 5350 m ssTVD (5277 m MD), and ODT is 5287 m ssTVD (5314
m MD). That is, ODT is 63 m above OWC.
Note from the correlations of "a", "b", pcmax, and Swir vs. FZI, for this well if FZI > 4.0, then a≈0.001,
b≈1000, pcmax≈26 psi, Swir≈0.18. pcmax≈26 psi corresponds to a height about 57 m above OWC. Since a (0.001)
is small and b (1000) is large, the Pc curve will be very flat and low. Practically for FZI > 4.0, if Pc reaches
about 10 psi, the oil-water transition zone will be over. In this case, Pc = 10 psi means about 23 m above
OWC.
Thus, practically the Swi profile won't be sensitive to the location of OWC for this well as long as the
OWC is about 23 m deeper than ODT. That is, any OWC deeper than 5310 m ssTVD would lead to a similar
Swi profile for this well that would look reasonable if there were no other data to invalidate it.
Fig. 11 shows that if the reservoir quality is good (good FZI, good porosity), the Swi from Pc model and
log agrees very well. Some difference can occur if the porosity is low. This difference needs to be reduced
by more efforts from Pc modeling and log analysis and by more integration and reconciliation.
In log analysis, if Archie's equation is used to get Sw, then

(21)

The cementation factor, m, greatly affects the water saturation calculation. Using a constant cementation
factor will lead to erroneous Swi results. For heterogeneous carbonate reservoirs, the cementation factor (m)
varies with rock type and porosity value. The lower the porosity, the lower the cementation factor will be.
In addition, the tortuosity factor (a) and saturation exponent (n) in Archie's equation are not constant either.
Among these three, the cementation factor (m) is the major one that greatly affects Sw calculation.
If log analysis can improve the Swi calculation, it is expected the Swi difference from log analysis and
Pc modeling will be reduced; see References 1–4 for details.
In addition, the integrated Pc modeling and Swi calculation method presented above has been
implemented in reservoir static model and the results are very satisfactory. Fig. 12 presents a comparison
of Swi from log analysis and Pc modeling on the same track. For reference, gamma ray, resistivity,
permeability, and porosity are also shown in Fig. 12. Some difference in Swi occurs when porosity is low:
14 SPE-199122-MS

the Swi from logs is higher than that from Pc modeling, as explained in Fig. 11. In general, the agreement
in Swi from logs (filled with light blue color) and from Pc modeling (red line) is very good.

Figure 11—Swi vs. ssTVD along an Example Well.


SPE-199122-MS 15

Figure 12—Implement Swi from Pc Modeling in 3D Reservoir Static Model.

Conclusions and Recommendations


1. The new capillary pressure model presented in this paper is more versatile than those published and
can handle any Pc data while other existing Pc models cannot. The new Pc model can fit any Pc data
with any curvature or even with zero entrance Pc.
2. The new Pc model can be used to consistently estimate the irreducible water saturation and the
maximum capillary pressure even if a true Swir end value is not reached during the lab test.
3. The new Pc model, when integrated with initial water saturation vs. depth from logs, provides a
consistent initial fluid saturation distribution in integrated static and dynamic reservoir modeling.
16 SPE-199122-MS

4. Real examples in two offshore oil reservoirs are presented to show the step by step procedure to obtain
the integrated capillary pressure model and initial water saturation for cores, along the wells, and in
3D integrated static and dynamic modeling.
5. Even if the initial water saturation is determined reliably by log interpretations along the wells, it is
still essential to use capillary pressure modeling not only to integrate with logs and other data but also
to populate the initial water saturation in 3D static and dynamic models.
6. This paper uses oil-water systems as an example to present the integrated capillary pressure and initial
water saturation modeling. However, the same method and procedures can be equally used for gas-
water systems and gas-oil-water systems.
7. It is recommended that the integrated capillary pressure modeling and initial water saturation methods
presented in this paper be used in integrated reservoir characterization and modeling as a best-practice
workflow and procedure guideline.

Acknowledgement
The author wishes to thank his colleagues Luis Navarrete, Stephen Winstanley, Arnulfo Briceno and Brigida
Fontecha for many helpful discussions, interactions, supports, and applications of the methods to many
reservoirs worldwide, his mentee and colleague Gurpreet Dhillon for working with the author and applying
the methods to various types of reservoirs worldwide, his mentee and colleague Agustin Avellaneda for
working with the author in transforming the author's formulae, algorithms, and procedures to completed
methods, and his former colleague Sebastian Leoni for encouragement in the initial work.
The author also wishes to thank many of his colleagues worldwide for working with him applying the
methods to various reservoirs worldwide and/or providing data and supports; they include Aivones Rivas in
Colombia, Huaila Fonseca in Brazil, Yan Dong, Mirko Hernandez, Jansen Dantas, Bin Liu, et al. in USA,
Alvaro Fernandez, Jorge Pena, Santiago Martinez, et al.

References
1. Focke, J. W. and Munn, D.: "Cementation Exponents in Middle Eastern Carbonate Reservoirs,"
SPE Formation Evaluation, (June 1987) 155–167.
2. Verwer, K., Eberli, G. P., and Weger, R. J.: "Effect of pore structure on electrical resistivity in
carbonates," AAPG Bulletin, (February 2011) 175–190.
3. Hamada, G., M. et al: "Uncertainty analysis of Archie's parameters determination techniques in
carbonate reservoirs," Journal Petroleum Exploration Production Technology, (2013) 3: 1–10.
4. Masoudi, R. et al: "On the Concept and Challenges of Water Saturation Determination and
Modeling in Carbonate Reservoirs," paper IPTC 14588 presented at the 2011 International
Petroleum Technology Conference, Bangkok, Thailand, November 15-17, 2011.

You might also like