SPE-181612-MS

A New Automated Algorithm for Designing the Optimal

Single-Well-Chemical-Tracer SWCT Tests At Various Reservoir Conditions

Rasoul Khaledialidusti and Jon Kleppe, Norwegian University of Science and Technology

Copyright 2016, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Dubai, UAE, 26-28 September 2016.

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
Single-well-chemical-tracer (SWCT) is the most commonly used field method to determine oil or water
saturation in one-spot pilot. The SWCT method overcomes the limitations of the well-to-well tracer method,
being long measuring time and excessive tracer dispersion. However, other limitations are encountered such
as the shorter investigation region and the effect of the product acid due to the hydrolysis reaction on the
reliability of the saturation measurement. The latter one is still considered to be a controversial issue.
Saturation measurements from large investigation regions are more reliable because of larger measured
volume and longer distance away from the wellbore damage and capillary desaturated zone. However, some
criteria prevent the test from reaching the desired investigation region and may thus cause deviations in the
test results from the ideal test. These criteria are the higher fraction of hydrolyzed ester and the detection
limit of ester, hydrolysis reaction during injection and production times, and fluid drift and tracer dispersion
during shut-in time. The main objective of this research is to develop an automated algorithm to design the
optimal SWCT test including the larger investigation region and more ideal tracer profile according to the
reservoir conditions.
The devised algorithm has been programmed based on the six different stages. In the first four stages,
all test design parameters in different investigation regions and retardation factors are calculated. The test
design parameters are sizing the test volume, test timing (i.e., injection, shut-in, and production), tracer
concentration during the test, and the mean residence volume. In the fifth stage, all criteria are taken into
consideration to find the optimal test design. After that, the achieved parameters are applied in the simulation
stage (sixth stage) of the workflow. The objective of the final stage is the investigation of the effect of
the following items on the tracer result in order to develop more ideal (i.e., Gaussian) tracer profiles: 1)
ester bank and concentration, 2) ester properties, 3) shut-in time and adapting the conventional rule to the
reservoir conditions.
To evaluate the workflow methodology, two different field test cases with different reservoir conditions
are employed in order to reflect the influence of different stages of the algorithm. After the numerical
interpretation of these two tests, the algorithm is performed to design the optimal test with all criteria taken
into account. The results show that the proficiency of the algorithm could be noticeable in the design of an
optimal test with larger investigation region yielding more Gaussian-like profiles by the means of modifying
2 SPE-181612-MS

the conventional rule of the shut-in time and selecting an optimal ester bank and concentration fitting to
the reservoir conditions.

Introduction
Determination of residual oil saturation (Sor) is vital in managing and selecting the EOR methods for
economic exploitation of a reservoir. There are many methods to determine Sor including laboratory and
field methods (Kidwell et al. 1980; Donaldson et al. 1981; Blackwell 1985; Chang et al. 1988; Teklu et al.
2013; Khaledialidusti et al. 2014). The laboratory methods are not representative for the large scale of a
reservoir and may not precisely predict Sor at reservoir scale even when performed with extreme accuracy.
Therefore, field methods to determine Sor at reservoir scale are more reliable. Recently, a combination of
the field methods is also proposed in order to determine a more accurate Sor (Khaledialidusti et al. 2015b).
Tracer methods were introduced as the most efficient field methods largely due to (1) the measurement
over large reservoir volume beyond damaged and desaturated regions and (2) non-dependency on porosity
(Khaledialidusti et al. 2015b).
The earliest tracer technique is the well-to-well method (Cooke 1971). This method employs two or
more non-reactive tracers with different partitioning coefficient ( ), where Co and Cw are the tracer
concentrations in oil and water phases at equilibrium. Cooke's method includes the injection of a solution
of the non-reactive tracers (e.g., 1% for each tracer) with different K-values into the reservoir (i.e., "tracer
bank"). Then, this slug pushes through the formation by the volume of water (i.e., "push bank"). Different
K-values lead to traveling speeds of the tracers at different velocities and thus different arriving times to
the production well. The separation between tracer profiles at the production well may be employed to
determine Sor between injection and production wells. The main limitations of this method are: (a) long
measuring time because of large measured volume between two wells and (b) extreme tracer dispersion
especially in layered formations.
An SWCT method, which is implemented only in a one-spot pilot was developed by Deans (1971) to
resolve the barriers of the well-to-well method. Injecting, pushing, and producing back of the two different
non-reactive tracers in a single well lead to simultaneous arrivals to the well due to the reversibility nature
of the non-reactive tracers and it does not lead to Sor measurements. Therefore, the SWCT method utilizes
only one chemical reactive tracer (ester) and hydrolysis reaction in order to cause irreversibility instead of
using two different non-reactive tracers with different K-values.
The SWCT method includes the injection of the ester bank into the target well. Then, the ester bank
displaces away from the wellbore to a radial depth of investigation by the push bank that comprises the
remaining of the SWCT test injection. After the injection step, the well is shut-in for a period of 1 to 10 days,
depending mainly on the formation water salinity, reservoir temperature and ester type. During this period,
the ester hydrolyses via reaction with the formation water to produce a kind of product alcohol tracer that
could be non-partitioning and a small amount of acid (Ester + H2O ↔ Alcohol + Acid). The shut-in time is an
effective parameter in SWCT test designs and must be long enough to form a detectable amount of alcohol.
At the end of this period, the unreacted (remaining) ester and the product alcohol are located together away
from the wellbore before the production step. During the production step, samples of the produced water are
collected at the wellhead and analyzed for the tracer concentrations. The effect of different K-values leads
to an observable separation between the product alcohol and unreacted ester profiles, as illustrated in Fig. 1.
SPE-181612-MS 3

Figure 1—Hypothetical SWCT test tracer concentration profiles to determine Sor (Data from Khaledialidusti et al. 2015b).

Although the SWCT method overcomes the limitations of the well-to-well method, other limitations are
encountered. These are: (a) a shorter depth of the investigation region (R-value) that is typically in the
range of 10-50 (ft) (Deans et al. 2007), and (b) the effect of the product acid on the reliability of the Sor
measurement that is still controversial.
Larger R-values are more promising for the determination of Sor because of the larger measured volume
and being further away from the wellbore damage and capillary desaturated zone. In practice, however,
reaching the desired R-value is limited by some restrictions such as fluid drift, physical dispersion,
hydrolysis reaction during injection and production (transient) time, and the high detection limit of tracers.
These limitations cause that the test results deviate more from the ideal test (i.e, Gaussian profiles) for large
R-values. As much as this deviation is higher, the probability of failure and error in Sor measurement is
higher and the interpretation of the tracer data is more complex.
The other limitation is the effect of the product acid on the reliability of the Sor measurement that this effect
still has a room for more investigation. Wellington et al. (1994) believed that the product acid affects the Sor
measurement. However, Deans et al. (1994) and Ghosh (1994) contradicted that belief. We believe that the
effect of the product acid on the reliability of the test results depends on many parameters such as reservoir
temperature and pH, formation brine components, rock compositions, ester bank and concentration, and
shut-in time. Therefore, in order to reach a conclusion in regard to this controversial issue, all of these
effective parameters and all reactions that could affect the pH of the reservoir during the SWCT test should
be taken into account for each field test.
The main objective of this research is to design an optimal SWCT test including (1) larger R-value and
(2) more conformity with the ideal test design.
Thus, a new algorithm has been devised in order to take into consideration simultaneously these principles
with regards to all limitations and parameters that affect the SWCT test design. Note that the algorithm has
been devised based on the assumption of the ideal SWCT test and the effect of the non-idealities on the
deviation from the ideal test have been considered. These assumptions are:

• Product alcohol is soluble only in water phase (K-valuealcohol = 0),

• Ester injection as an impulse,

• No dispersion and mixing effects,

• No reaction during transient (injection and production) time,

• No fluid drift (ester is stationary during shut-in time),

• Flow is reversible.

The main parameters of the optimal test design are:

4 SPE-181612-MS

• Selecting the optimal ester and the properties of the selected ester according to the reservoir
conditions.
• Sizing the test volume including the ester and push banks, the injection, shut-in, and production
times.
• Fraction of hydrolyzed ester and tracer concentrations during the test.

• Acceptable deviation from the ideal test results.

• Dispersion.

Optimal SWCT Test Design

After making sure that the oil cut of the test well is at or near Sor, some required properties of the target
formation must be specified before starting the test design. All of these effective parameters in the optimal
SWCT test design are linked together as a chain and they also influence each other considerably, as shown
in Fig. 2.

Figure 2—Effective parameters on optimal SWCT test design

• Reservoir lithology: Many SWCT tests were performed in both sandstone and carbonate
formations (Carlisle et al. 2014; Khaledialidusti et al. 2015a). The results showed that the test
design in carbonate formations is much more complex because of the dispersed nature of the
tracer profiles and the non-validity of the local equilibrium assumption; however, the results
from sandstone formations were more satisfying (Deans et al. 2007; Skrettingland et al. 2011).
The algorithm is only reliable for sandstone formations because it is devised based on the local
equilibrium assumption.
• Reservoir temperature and salinity: These two parameters have a significant effect on the
test design by affecting the ester properties including partitioning coefficient (K) and hydrolysis
reaction rate (KH). These two properties increase with reservoir temperature and salinity. Both very
high and low K- and KH-values make restrictions to reach larger R-values.
SPE-181612-MS 5

K-values are usually measured in the laboratory at reservoir conditions where the ester is located during
the test. In order to avoid chemical incompatibility and to insure a condition similar to the laboratory
measurements, the produced water from the target pilot with the reservoir temperature should be used
to displace the ester. A pre-flushing with injection cold water before ester injection to provide a lower
temperature region for the ester is not recommended because of the probability of the slower propagation
of the thermal front velocity than the ester bank velocity (Deans et al. 1980).
In the ideal SWCT test, no alcohol produces during transient time and all hydrolysis reaction takes place
only during shut-in time. In reality, however, some alcohol produces during transient time depending on the
KH-value and transient time. This could be one of the reasons of tailing in the tracer profiles and makes the
interpretation of the tracer profiles more complex. Since the KH-value is strongly dependent on the reservoir
temperature and salinity, more reaction occurs at high temperature and high salinity (HT/HS) reservoirs.
Terefore, the deviation from the ideal test and the probability of complete reaction of ester which leads to
failure in the test is higher in HT/HS reservoirs.
On the other hand, a lower KH-value leads to the design of a SWCT test with longer shut-in time in order
to form a detectable amount of the second tracer. Longer shut-in time intensifies the unfavorable phenomena
such as physical dispersion, fluid drift, and cross flow that could affect the SWCT test results by causing
more deviation from the ideal test.
Higher K-values lead to more solubility of the tracer in oil, and then, more injection time needs to reach
the specific R-value. The notion of more injection time is consistent with our discussion of higher KH-values.
A lower K-value leads to smaller differences in mean residence time of unreacted ester and alcohol
profiles and make the interpretation of the results more difficult, especially at lower Sor.
Therefore, the selection of the proper ester among the available esters in the industry with optimal K-
and KH-values is fundamental to design the optimal test.

• Test interval size and average porosity: In target wells with higher interval size and porosity, a
larger ester bank should be injected in order to reach larger R-values. It could increase cost and the
probability of getting non-Gaussian profiles. Therefore, the requirement for finding the optimal test
design is stiffer at this condition, although Sor determinations are not dependent on these parameters.
• Injection/Production rate: The production rate should be able to produce back the ester bank
in the range of one day to two days. The injection and production rates of the SWCT tests are
approximately the same and the effect of these rates on the fracturing of formation and fluid drift
must be considered during the test implementation. The total volume of the test injection is chosen
to ensure that the test can be injected in one day or less. In practice, the total production time
required is at least twice the total injection time to make sure of the complete production of tracers.
• Fluid drift: In ideal tests, all chemicals are stationary in the formation during the shut-in period.
In practice, however, fluids relocate to different extents during shut-in time due to the pressure
gradients across the formation or pressure differences between different layers in the interval size
(Wellington et al. 1994). Designing the test with shorter shut-in time could facilitate the adverse
effect of the fluid drift on the tracer profiles while it could decrease the ratio of product alcohol in
shut-in time to transient time and more deviation from the ideal test in larger R-values.
• Physical dispersion: This is another adverse phenomenon that affects the tracer profiles more
severely at larger R-values.

Workflow of finding optimal SWCT test

The algorithm developed here has been programmed based on the six different stages (Fig. 3). In the first
four stages, all possible test designs are determined. The design parameters are sizing the test volume,
test timing, tracer concentrations during the test, the mean residence volume, and fluid drift and physical
6 SPE-181612-MS

dispersion effects. In the fifth stage, all criteria are considered to find the optimal test design. Then, in the
sixth stage, the obtained parameters are applied in the simulation stage of the workflow in order to study
the test in more details and improve the test design.

Figure 3—The algorithm for designing optimal SWCT test

SPE-181612-MS 7

All required parameters to design the optimal test are quantified via an input file. These are the reservoir
temperature, brine salinity, average porosity, permeability, interval size, wellbore radius, estimated Sor, water
viscosity, drift velocity, ester concentration (%) and density, laboratory K- and KH-values.
A "Mini" test, including only pushing and pulling of the ester bank without shut-in time, should be
implemented in the target well before doing the "main" test in order to investigate formation parameters
such as drift velocity and KH-value in case of lack of these parameters.

Stage 1 - Ester Selection

The first step after getting input data is ester selection and the calculation of K- and KH-values.

• K- and KH-values: The K-value is a crucial parameter for havnig an appropriate retardation factor
(β-value). This factor is related to the distance between the mean residence volumes of tracer
profiles. The KH-value is also a determinative parameter in test timing.
In practice, after the selection of the most proper ester among the available esters, these two parameters
must be measured in laboratory using formation water and reservoir temperature. In the algorithm, in
case of the lack of the laboratory measurements, the proposed correlations by Deans et al. (1980) and
Carlisle et al. (1982) to calculate these values for two applicable esters (i.e., Propyl format (PrFr) and
Etyle Acetate (EtAc)) are used. These correlations were achieved using many laboratory measurements in
different temperatures and salinities that are given in Appendix A.
The ester with lower KH-value among the available esters should be selected in order to perform at higher
reservoir temperatures. Since PrFr reacts too fast above 130 °F (KH(PrFr) ~ 50 KH(EtAc)), it is not proper
for a high temperature reservoir. Therefore, if the temperature is higher than 130 °F, EtAc is considered
automatically as the ester. Otherwise, PrFr is considered by the program.
Retardation factor (β-value). This is derived from chromatography theory and is a function of K-value
and Sor as follows:

(1)

The β-value increases with Sor and leads to a larger distance between the peaks of the tracer profiles. In
the case in which no residual oil phase is present in the pore space, the ester and product alcohol travel back
to the wellbore at the same velocity and the peaks superimpose in the production plot. The suitable range
of β-value was proposed by some authors as follows:

Deans et al. (1980): 0.5 < β < 2.0

Deans et al. (1994, and 2007): 0.5 < β < 1.5
Shook et al. (2004): 0.2 < β < 3.0

In the algorithm, the β-values in the range of 0.2 and 3.0 with the incremental change of 0.2 are considered.
According to Eq. 1, different K-values are calculated for each β-value at a given Sor. In reality, however,
the selected ester has only a unique K-value at the reservoir condition. Therefore, the algorithm has the
capability to design the SWCT test using all different calculated K-values. It helps us to highlight the effect
of the K-value for designing more ideal tracer profiles.
Investigation region (R-value). Designing the most ideal SWCT test with the largest possible R-value is
the objective of the algorithm. The suitable R-value was suggested by Deans et al. (2007): 10 ft < R < 50 ft
8 SPE-181612-MS

In the algorithm, the proposed range of R-values with the incremental change of 1.64 ft is used. For each
R-value, the tests with β-values from 0.2 till 3.0 with the incremental change of 0.2 are designed, as it is
mentioned.

Stage 2 - Volume of the Test Designs

In this stage, first, the highest possible injection/production rates based on some criteria are determined.
Second, total required volume (ester and push banks) and the minimum and maximum ester banks are
calculated for all possible R- and β-values. Then, injection, shut-in, and production times are calculated for
all cases based on the conventional rule. Finally, in order to get more interpretable tracer profiles from the
test, the rule for shut-in time is adapted to the reservoir conditions.

• Injection/Production rate (qinj/qprod). The highest possible qinj and qprod need to reach the largest
possible R-value. Transient time decreases with qinj and qprod in order to reach a specific R-value.
Therefore, lower alcohol is being produced during this period and tracer profiles are more ideal.
However, several parameters prvent the test from having higher rates such as: a) facility of the
target well and wellbore damage, b) oil saturation, and c) fluid drift. In the algorithm, all of these
effects are taken into account to find the highest possible qinj and qprod. These rates are initially
specified in the input file.
a. Facility of the target well and wellbore damage: The upper limit of these rates is considered
based on these factors in the input file.
b. Oil saturation: Since oil saturation in the test interval must be at or near Sor before SWCT test
implementation, the effect of the rate on the oil saturation must be controlled. The Capillary
Desaturation Curve (CDC), that is the measurement of Sor verses capillary number (Nc), is
used to make sure that the rates are safe and do not lead to production of oil by viscous forces.
The details of the Nc calculation and applied CDCs in the algorithm are given in Appendix B.
c. Fluid drift: This unfavorable phenomenon increases with the injection rate and deviate the
mean residence volume of the tracers from the ideal ones. Here, the degree of this deviation
is calculated and if it is significant, the rates are reduced.
• Total required volume (VTotal). VTotal is calculated for all possible R- and β-values using the
following equation:
(2)
Here Ø and H are porosity and interval size of the target well, respectively, and are given in the input
file. The ratio of the ester bank (VSlug) to VTotal could have a great influence on the tracer profiles and is
defined as follows:

(3)

The range of efficient F-value was suggested by Deans et al. (1980): 0.25 < F < 0.5
Using this range, the minimum and maximum possible VSlug is determined. The following question is
answered in order to study the effect of this ratio on the optimal test design and improve the reliability of
the method. "Could a smaller ester bank and/or lower ester concentration improve the test data?"

• Injection, Shut-in, and Production Times. Test timing is the main part for designing the SWCT
test. Tracer profiles are more Gaussian (ideal) with decreasing the total test time. On the other
hand, more time needs for reaching the larger R-values. Therefore, all mentioned barriers cause
deviation from the ideal test are taken into account for linking these two principles together in
order to design an optimal test.
SPE-181612-MS 9

After determining qinj and qprod and calculating VTotal for all possible range of R- and β-values, the total
injection and production times (tinj and tprod) are calculated by the following equations:

(4)

(5)

Note that tinj and tprod are equal if qinj and qprod be equal. In practice, however, the fluid must produce back
at least twice as much as the fluid is injected to make sure that complete tracer profile will be achieved.
Provided that the SWCT test performs with less amount of the hydrolysis reaction during transient time
compared to the shut-in time (tsoak), the test can fulfill closer to the ideal condition. To meet this criterion,
the following equation is used to determine the optimal shut-in time.
(6)
Here m is the ratio of the injection rate to the production rate (m = qinj/qprod = tprod/tinj) and C is a constant
coefficient to specify the ratio of tsoak to the total transient time. Deans et al. (1980) proposed the "two-times
rule" by using C = 2, which means tsoak is twice as long as the total transient time. Since hydrolysis reaction
rate varies with the reservoir conditions, the "two-times rule" is adapted to the reservoir conditions in the
workflow to design the optimal test.
Deans et al. (2007) showed that tsoak could be reduced to about the same as transient time. The California
turbidite tests are examples of tests with a relatively short tsoak. Shorter tsoak could facilitate the adverse effect
of the fluid drift. On the other hand, shorter tsoak leads to increasing the ratio of reaction during transient to
shut-in time. As much as the ratio of product alcohol during the transient time to shut-in time increases, the
tailing of the alcohol profile growths and it raises the uncertainty of Sor values (Mechergui et al. 2012).
The total test time (ttot) is calculated by the following equation:
(7)

Stage 3 - Tracer Transport through Porous Media

In this stage, tracer velocity and concentration through porous media are calculated as follows:

• Tracer Velocity. The tracer velocity across the target formation is calculated with the assumptions
of the ideal SWCT test, incompressible flow, pseudosteady state, single-phase, no pressure gradient
across the reservoir, and radial flow from the wellbore radius (rw) until the R-value with a
homogeneous layer of the interval size (H) and porosity (Ø).
The distance between rw and the R-value is divided by N cells with the same properties. Then, ester
velocity (ue) is calculated in each cell as follows:

(8)

Since the relation between ue and reaching time of ester (te) is not linear, the mid-point rule is performed
for the te calculation of the impulsive ester to each cell using Eq. 9. The number of cells (N) should be large
enough to minimize the error of this integration rule. For checking the calculations, (te)N must be equal to tinj.

(9)
10 SPE-181612-MS

The second tracer (product alcohol) is soluble just in water and travel with Darcy velocity
( ).

• Tracer Concentration. In order to calculate the ester and product alcohol concentrations (i.e., CA
and CB), first, the fraction of hydrolyzed ester (Xh) is calculated in accordance with the obtained
(te)N using the following equations:

(10)
Then, CA and CB are calculated using Eqs. 11 and 12.
(11)
(12)
The initial condition for the CA and CB calculations may be determined from the assumption of the
impulsive ester injection as follows:
(13)
where (CA)ini is injected ester concentration (mg/L) and may be calculated using the following equation:
(14)
where ρA is ester density (kg/m3) and δ is the ratio of ester solution to ester bank (wt %) that are given in
the input file.

Stage 4 - Distance from the Ideal SWCT Test

In this stage, the mean residence volume with considering all non-idealities is calculated and compared with
the ideal test design.

• Mean Residence Volume. The mean residence volume for ideal SWCT test is given as follows:

(15)

where and are the volumes at which the statistical peak concentrations of the ester and product alcohol
tracers are produced, respectively.
However, in reality, ester reaction during transient time is inevitable. In some reservoirs, fluid drift and
dispersion phenomena have also a significant influence on the mean residence volume of the ideal SWCT
test (Eq. 15).
In the algorithm, the effect of the hydrolysis reaction during transient time on the mean residence volume
is studied by comparing the given analytical solution (Deans et al. 1980) in Eq. 16 with Eq. 15.

(16)

In the analytical solution, the effects of fluid drift and dispersion have been ignored.
Deans et al. (1980) proposed a simplified equation (Eq. 18) to determine the mean residence volume
under the following condition:
(17)
SPE-181612-MS 11

(18)

where . From Eqs. 18 and 15, it may be inferred that the M-value may be negligible (tinj << tsoak;
M → 0) in the design of a more ideal test.

• Fluid Drift. The assumption of radial flow is not necessarily accurate in many reservoirs under
active waterflooding because of the unbalanced fluid withdrawals and fluid drift. However, it has
not been taken into consideration in the given mean residence volumes (Eqs. 15, 16, and 18).
In the algorithm, for studying the effect of fluid drift during shut-in time on the ideal mean residence
volume the dimensionless parameter "D" is determined as follows:

(19)

"D" is the square of the ratio of the total drift distance during the test to the R-value. UD is interstitial
linear "drift" velocity which is given in the input file. Then, the mean residence volume when taking fluid
drift into account under the condition of (Xh)N < 0.5 may be calculated as the following empirical equation
(Deans et al. 1980):

(20)

where P1 and P2 are be calculated as follows:

(21)
(22)
Eqs. 20 and 18 give the same result if D = 0.
The upper limit of "D" was observed at 0.66 for most SWCT tests until the present time (Deans et al.
1980). In the few cases with D > 0.66, distortion of the tracer profiles affected the test results considerably.
Therefore, minimizing this parameter is required by choosing the ester with higher K-value and/or the test
design with shorter R-value and tsoak.
Using the sensitivity analysis that is given in Appendix C, the optimal β-value was determined to 1 + P1.
Nonetheless of the M-value in Eq. 18, the optimal β-value was determined to 1.0 if the fluid drift effect is
negligible. So, the optimal β-value increases with the fluid drift.

• Physical Dispersion. Dispersivity is the other undesired phenomenon that affects the quality of the
tracer profiles to some extent. It could be more severe and be deviated the test results more from
the Gaussian tracer profiles in larger R-values. In the algorithm, physical dispersion parallel to the
flow direction is calculated based on the tracer and water properties such as molecular weight of
water, temperature, water viscosity, and molar weight of the tracer as follows:

(23)

In Eq. 23, the Peclet number ( ) is calculated using the interstitial velocity ( ) at the R-value,
the grain size (b), and the molecular diffusion coefficient (Dm). In porous media, the molecular diffusion
must be replaced by dispersion (Adler 1992). Dm of the tracer with different K-values could be estimated
12 SPE-181612-MS

using different mathematical expressions that are given in Appendix D. In the algorithm, this coefficient is
estimated by the Wilke-Chang expression (Perry et al. 1997).
The typical grain size of 2×10−2 (cm) is considered as a default value which is adjustable in the input file.
The formation factor, F, was estimated 2.84 from Archie's law for unconsolidated sand (Bear 1988). The
parameters α11 and β11 are dependent on the media and were estimated at 0.59 and 1.2, respectively, from
the experimental data (Dullien 1992).

Stage 5 – Criteria
After designing all possible SWCT tests with different R- and β-values, the following criteria are taken into
account to find the optimal SWCT test design (i.e., larger R-value and more in conformity with the ideal
test design) for the target well:

• Criterion #1: The lower and upper limits of Xh should be in the following range (Deans et al. 1980):

• Criterion #2: Detectability limit of the tracers is considered to be 200 (mg/L); however, different
esters have different detection limits.
• Criterion #3: tinj should be one day or less in order to have negligible hydrolysis reaction during
the transient time.
• Criterion #4: Hydrolysis reaction effects during the transient time on the mean residence volume
that is determined by comparing Eqs. 15 and 16.
• Criterion #5: Fluid drift effect during the shut-in time on the mean residence volume that is
determined by comparing Eqs. 18 and 20.
• Criterion #6: Ester dispersion at the end of the injection time that is determined by Eq. 23.

After taking all criteria into account and deciding on the optimal test design, the obtained parameters are
applied for the simulation stage of the workflow in order to improve the optimal design by studying the
injected ester concentration, ester bank, and tsoak effects.

Stage 6 – Simulation
In the previous stages, the optimal test design with the larger R-value and lower non-ideal effects is
determined. However, there are still some parameters for improvement that may be studied using the
simulation stage. These parameters are: 1) injected ester concentration and the ester bank size, 2) modifying
the conventional rule using adapting tsoak to different reservoir conditions. In order to adapt tsoak, the algorithm
repeat again from time calculation section in stage #2 with different "C" coefficient and results are compared
to find the optimal test design with the optimal tsoak. 3) Comparing the optimal test design with the obtained
β-value and the optimal β-value of 1 + P1. The optimal β-value for each test may not be applicable in the
field test due to lack of the availability of different esters.
The simulation stage of SWCT tests is performed using UTCHEM which computes the total tracer
concentrations from species conservation equations including a reaction term for the ester. UTCHEM
calculate the concentration of tracers with the assumptions of incompressibility of fluid flow, pseudosteady
state, single phase, and radial model (Deans et al. 2007). In addition, tracers do not occupy any volume and
also have no effect on the physical properties (UTCHEM Technical Documentation 2000).

Could a smaller ester bank and/or lower ester concentration improve the
test data?
Answering to this question could improve the SWCT test designs in several ways. On the one hand, a smaller
ester bank and/or lower ester concentration excludes the test to reach larger R-values especially for HT/HS
SPE-181612-MS 13

reservoirs because of the greater fraction of hydrolyzed ester and the difficulty for the tracer detection at
the wellhead. On the other hand, a larger ester bank and/or higher ester concentration causes some other
difficulties as follows:

• Greater impact on the pH-alteration during the shut-in time and inconsequence probable variable
KH-value.
• Requiring more space for the preservation. Since the existing esters are highly flammable, it could
be more unsafe.
• Only one ester could be used in each test; however, a smaller ester bank allows us to inject several
esters with different K-values in the same target well in order to improve the accuracy of the SWCT
test.
• Not being cost-effective.

Therefore, lower ester concentration and/or smaller ester bank might be preferred as far as the tracer
detection limit does not cause problem. The schematic view of different ester concentration and ester bank
is shown in Fig. 4.

Figure 4—Schematic view of different ester concentration and ester bank.

PH-variation during the SWCT test

Hydrolysis reaction during the SWCT test leads to the production of some acid in addition to the second
tracer. Therefore, the pH of the reservoir in the ester bank could change probably. It is generally accepted
that the KH-value are largely dependent on pH and this rate will vary by about a factor of 10 by a change
of 1 unit in pH value (Wellington et al. 1994). The assumption of constant KH-value in the numerical
interpretation of SWCT tests could cause some error in Sor measurements if the KH-value varies considerably.
The measured pH at the wellhead during the production is the minimum amount of the pH-variation because
of the dispersive mixing in the reservoir and could not be a reliable reference for making sure about being
pH-constant. Therefore, more analysis is needed to design a SWCT test with lowest pH-variation in the
ester bank.
The product acid during transient time has an insignificant effect on the KH-value due to (1) the short
time of this step leads to negligible acid production. (2) the pH-front moves with Darcy velocity and ahead
of the ester bank that has a partially oil solubility.
Unlike the transient time, the product acid during shut-in time might be considerable and also remains in
the ester bank. It might reduce the pH and KH-value. The ester concentration, ester bank size, and modifying
"two-times rule" are taken into consideration to lessen the probability of pH-change during shut-in time.
The effect of pH-variation during SWCT tests on the Sor measurement is still controversial. Some
researchers believe that the pH-variation is significant; however, others believe that this variation is
negligible. Wellington et al. (1994) stated that "The pH-driven rate dependence of ester hydrolysis can cause
transient-generated tracer to dominate the response and make the test insensitive to oil saturation." On the
14 SPE-181612-MS

other hand, Deans et al. (2007) also stated that "The acid formed during the reaction is not observed because
it is neutralized by the natural base components of the reservoir."
Deans et al. (1994) studied the effect of pH-drop and the variation of KH-value in California Turbidite
reservoirs with almost 1000 (mg/L) bicarbonate level, which is a very extensive amount. They showed that
the difference between constant and variable KH-values is only on the tail of the tracer profiles, as shown in
Fig. 5, and the assumption of constant KH-value could not affect the Sor measurements in this SWCT test.
Therefore, a SWCT test with lowest pH-variation leads to more Gaussian tracer profiles and more accurate
Sor measurement.

Figure 5—Simulated ethanol concentration using constant and variable

reaction rate in California Turbidite Test (Data from Deans et al. 1994).

Ghosh (1994) simulated several SWCT tests at various temperatures, formation brine compositions and
buffering action, initial pH-values, and reservoir rock contents to study the effect of these parameters on
the amount of the pH-variation. It was showed that (1) natural buffering (i.e., carbonates and bicarbonates)
in almost all formation brines prevent substantial pH-variation at the normal ester concentrations (1 wt%)
used. (2) Even in the poorly buffered formation brines, the most part of the second tracer produces in the
high pH-zone during shut-in time.
The effect of pH-drop during shut-in time on the KH-value depends strongly on some other parameters
such as ester bank and concentration and tsoak in addition to the parameters that Ghosh (1994) took into
account. We believe that KH-variation and the amount of error in Sor determination might be significant in
some SWCT tests. All of these parameters should be analyzed together in order to decide on the optimal
test design with the lowest probability of KH-variation.
Different formation brine contents and rock compositions could have a different effect on the
buffering of the system. The dissolution of solid calcite leads to releasing calcium and carbonate ions
( ). Then, the produced carbonate ions shifts the homogeneous aqueous phase reactions
(e.g., and ) to the left side and improve the buffering of the system
+
by the reduction of the hydrogen ion (H ) concentration and resisting the pH-variation.
Initial pH of the reservoir is the other effective parameter on the KH-variation. Fig. 6 shows the KH-
variation with pH reported by Wellington et al. (1994) and extracted data from Int'l Critical Tables for EtAc
at 195 °F. As it can be seen, the KH-values are more or less in the same order for the range of pH1 and
pH2 with the minimum of pHm. If the initial pH of the reservoir and pH-variation during the shut-in time
occurs in this range, the KH-variation could be negligible; however, the reservoir with either higher initial
pH than pH2 or lower than pH1, KH-variation could be more considerable during shut-in time. In this case,
the assumption of the constant the KH-value in the simulators could probably affect the Sor determination.
SPE-181612-MS 15

Figure 6—Effect of pH on the hydrolysis rate of EtAc (Data from Deans et al. 1994).

Results and Discussion

To evaluate the workflow methodology, two field SWCT tests (Test #11 and #25) which were given by
Deans et al. (1980) have been selected because (1) they could be represented by a simple one-layer reservoir
simulation model and (2) they have different test and formation properties (Table 1) that could highlight the
influence of different stages of the algorithm on the SWCT test design.

Table 1—Field SWCT Test Parameters

Reservoir Properties Ester Properties

Test #11 Test #25 Test #11 Test #25

Interval Size, ft 20 32 EtAc Concentration, vol% 1.074 0.9

Porosity 0.325 0.25 MeOH Concentration, vol% 0.5 0.5

Permeability, md 400 1000 Slug Volume, ft 3 842 1684

Reservoir Temperature, °F 170 172 Push Volume, ft3 2526 5052

Salinity, mg/L 54400 100000 Injection Rate, ft3/day 4029–4029 4808–3558

Production Rate, ft3/day 3014 3465

Shut-in Time, days 2.92 10.5

Drift Velocity, ft/day 0.5 0

K-value 4.68 8.17

Sor, % 33 12

The field Tests #11 and #25 have been simulated using UTCHEM to match the modeled tracer response
with the measured tracer data. The procedures of finding the best-fit parameters were explained by Deans
et al. (2007) and Khaledialidusti et al. (2015a). The final best-fits are illustrated in Fig. 7.
16 SPE-181612-MS

Figure 7—Measured (dotted) and modeled (solid lines) tracer profiles in Tests #11 and #25.

First stage
Ester is selected based on the reservoir temperature after getting the input parameters. In both tests, EtAc
was injected into the formation and K-value was measured in the laboratory with the reservoir conditions
(Table 1). Likewise, EtAc has also been selected by the algorithm due to the temperature more than 130 °F.
The R- and β-values are calculated using Eqs. 1 and 2 with regards to the given input data in Table 1. The
results of this stage for these two field tests are given in Table 2. The calculated R-value is 8.63 ft and 12 ft
for Tests #11 and #25, respectively. The measured K-value is used in the algorithm. The K-value has also
been calculated (Table 2) using the proposed correlations that are given in Appendix A in order to evaluate
the efficiency of the first stage in the absence of the laboratory measurements. While there is a significant
difference in the measured and calculated K-value in Test #11, the difference is not considerable in Test
#25, especially in Eq. A1. There is also a great difference in the obtained KH-value using the numerical
interpretation (i.e, matching the measured and modeled tracer profiles) and the calculate KH-value (Eq. A2)
in both tests. Therefore, the proposed correlations are not reliable for all tests; however, these could be
trustworthy enough for finding the optimal test design in lack of the laboratory measurements.

Table 2—Ester Properties

K-value (Laboratory β-value R-value, ft Calculated K-value KH-value

measurement)
Eq. A1 Eq. A2 Numerical Eq. A2
interpretation

Test #11 4.68 2.31 8.63 6.01 10.98 0.01 0.039

Test #25 8.17 1.11 12 8.19 8.96 0.018 0.041

Second stage
VTotall for all possible test designs (i.e, the R-value of 10 ft to 50 ft and β-value of 0.2 to 3 with incremental
change of 1.64 ft and 0.2, respectively) is determined using Eq. 2. This stage starts with the examination of
qinj and qprod using Nc calculation (Appendix B). The algorithm run using the given rates in Tests #11 and
#25 and the results of Nc calculation show that these injection rates could increase in order to reach larger
R-values as shown in Fig. 8. The distance closer than 3.28 ft to the wellbore is considered as a desaturated
zone. In practical SWCT test design, the decision on the flow rates should make based on the wellbore
damage, facility, and fluid drift in addition to Ncc criterion. Upper possible limit of the rates in the target
well should specify in the input file. Then, the maximum possible rates based on Ncc criterion are selected.
Here, the rates are increased to 11300.7 ft3/day and 18363.6 ft3/day with regards to the Ncc criterion. Higher
rates intensify the fluid drift value that is unknown for the new rates.
SPE-181612-MS 17

Figure 8—Nc calculation for Tests #11 and #25.

Next, the timing of all possible test designs using "two-times rule" are calculated using Eqs. 4–7. Due to
the large number of data, the results only for the R-value of 32.8 ft are shown in Fig. 9.

Figure 9—VTotall and test timing for Tests #11 and #25 at R-value of 32.8 ft.

From Fig. 9, the ester with an upper K-value is required in order to have higher β-values with the same
Sor (Eq. 1). The β-value has a great impact on the optimal test design; however, the lack of the variety of
the esters could deprive us to perform the test with the optimal β-values. tinj and tprod is the same for both
tests because of the equal qinj and qprod.
The K-values for having the optimal β-value of 1.0 based on the sensitivity analysis (Appendix C) for
Tests #11 and #25 are 2.03 and 7.33, respectively, as shown in Fig. 9. However, the measured K-values
for these tests were 4.68 and 8.17, respectively (Table 2). It shows that these injected esters could not lead
to performing the test at optimal conditions. Therefore, other kinds of esters must be used for achieving
the optimal test.

Third stage
In order to determine the tracer concentrations during the SWCT test using Eqs. 8 thorough 14, the distance
between wellbore and R-value in all cases are divided to N cells, as described earlier. First, the velocity
and reaching time of the impulsive ester bank in each cell ((ue)n and (te)n) are calculated for all possible
R- and β-values. Then, the fraction of hydrolyzed ester in each cell ((Xh)n) is calculated. Due to the large
number of data, the results only for alcohol concentration with three different β-values at R-value of 32.8
ft are shown in Fig. 10.
18 SPE-181612-MS

Figure 10—Ethanol Concentration for Tests #11 and #25 at R-value of 32.8 ft.

Fig 10 shows that most of the alcohol is produced during shut-in time in both tests as expected. Second,
larger K-values have led to production of more alcohol and it might lead to failour of the test. This is
because tinj increases with K-values to reach a specific R-value and tsoak also increases based on "two-times
rule". Therefore, reduction of tsoak could be an ideal solution to eliminate this difficulty in larger K-values
especially at HT/HS reservoirs. On the other hand, the higher ratio of product alcohol during shut-in time
to transient time is more desirable as mentioned earlier. The effect of tsoak is evaluated in the simulation part
of the workflow (stage 6).

Fourth stage
The ideal mean residence volume in all possible cases is calculated using Eq. 15. Then, the effect of the
hydrolysis reaction during transient time and fluid drift during shut-in time on the deviation of the mean
residence volume from the ideal case are determined. Finally, the amount of the physical dispersion in all
cases is calculated using Eq. 23.
Fig. 11 shows that the deviation between the ideal mean residence volume (Eq. 15) and the one with
reaction consideration during transient time (Eqs. 16 and 18) increases with β-values. Fluid drift also affects
the mean residence volume and the deviation between the mean residence volume with considering fluid
drift (Eq. 20) and the ideal one (Eq. 15) increases with β-values. However, fluid drift has no effect in Test
#25 because of negligible amount of the fluid drift.

Figure 11—Mean Residence Volume for Tests #11 and #25 at R-value of 32.8 ft.

Designing the SWCT test with smaller β-values is more in conformity with the ideal test. However, it
causes smaller difference between the mean residence times of the ester and alcohol profiles, especially in
lower Sor. The availability of different esters to have the optimal β-value is another barrier to implement
SPE-181612-MS 19

the optimal SWCT test. Therefore, all aspects should take into account in different reservoir conditions to
find the optimal test.
At the end, physical dispersion ( ) in all cases is calculated using Eq. 23. Fig. 12 shows and NPe
only for three different R-values for Tests #11 and # 25. As it can be seen, growths with both β- and R-
values. As it is mentioned earlier, makes limitation to reach larger R-values, especially in high dispersive
reservoirs. There is a negligible difference between Npe for Tests #11 and # 25 because of having almost
the same Nc in both tests as shown in Fig. 8.

Figure 12—Physical dispersion for Tests #11 and #25 at R-value of 32.8 ft.

Fifth stage
All the mentioned criteria are taken into account to find the optimal SWCT test design (i.e., larger R-value
and more in conformity with the ideal test design) among all possible test designs. Part of the results of the
designed SWCT Tests #11 and #25 with the new determined qinj are listed in Table 3 based on the largest
R-value, optimal β-value, and applied ester.

Table 3—Designed SWCT Test

Test #11 Ester Con.: 9687.48 (mg/L) Vslug ft3 Time days Mean Residence volume

R-
β-value K-value Min. Max. inj soak Eq. 15 Eq. 16 Eq. 18 Eq. 20
Value ft

#11-1 24.6 0.2 0.41 2485.15 4970.29 0.88 3.52 1.2 1.16 1.16 1.12

#11-2 19.7 1 2.03 2650.82 5301.65 0.94 3.75 2 1.71 1.71 1.6

#11-3 14.8 2.4 4.87 2534.85 5069.7 0.90 3.59 3.40 2.43 2.43 2.27

Test # 25 Ester Con.: 8118 (mg/L) Vslug ft 3

Time days Mean Residence volume

R-
β-value K-value min max inj soak Eq. 15 Eq. 16 Eq. 18 Eq. 20
Value ft

#25-1 26.25 0.2 1.47 4570.81 9141.63 1 3.98 1.2 1.16 1.16 1.16

#25-2 19.68 1 7.33 4285.15 8570.27 0.93 3.73 2 1.71 1.71 1.71

#25-3 18.04 1.2 8.8 3960.78 7921.57 0.86 3.45 2.2 1.83 1.83 1.83

Largest R-value has been calculated to be 24.6 and 26.25 (ft) for Tests #11 and #25, respectively.
Performing these two designed tests is required the smallest β-value of 0.2. According to Eq. 1, esters
with K-values of 0.41 and 1.47 at reservoir conditions are needed for Tests #11-1 and #25-1, respectively.
The selection of these designed tests is restricted by lack of the diversity of esters with theses K-values at
20 SPE-181612-MS

reservoir conditions in industry. In addition, these small β-values causes the minor distance between the
peaks of the tracer profiles especially for Test #25 including Sor=0.12.
As mentioned earlier, the optimal β-value was determined to be 1 + P1 based on the sensitivity analysis
where P1 is a dependent parameter to fluid drift (Eq. 21). Here, the effect of fluid drift on the optimal β-
value has been ignored and the optimal β-value of 1 is considered for both field tests. However, the optimal
β-value only for Test #11 is a little bit more than 1 due to fluid drift of 0.5 ft/day. For designing the test with
the optimal β-value of 1, applying the ester with K-value of 2.03 and 7.33 at reservoir conditions is required
for Tests #11-2 and #25-2, respectively. There is still the difficulty of the diversity of esters in industry for
performing the test with the optimal β-value of 1. Applying these tests lead to decrease R-values to 19.7 (ft)
and 19.68 for Tests #11-2 and #25-2, respectively (Table 3).
The applied ester (i.e., EtAc) for Tests #11-3 and #25-3 does not let these tests reach R-values larger than
14.8 and 18.04 (ft), respectively. This is because of the high K-values of EtAc at the reservoir conditions in
these two tests, which leads to getting β-values 2.4 and 1.2 for Tests #11-3 and #25-3, respectively. Since
the difference of β-values between Tests #11-3 and #11-2 is significant compared to Tests #25-3 and #25-2,
the deviation of the mean residence volume from the test with the optimal β-value is more significant.
Therefore, the selected ester strongly affects the optimal design of an SWCT test and the importance of the
ester shortage in industry is serious.
The difference between the calculated mean residence volume (Eq. 20) and the ideal one (Eq. 15) is
shown in Fig. 13 (a). From this figure, this difference in Test #11-3 is twice Test #25-3. Generally, this
difference for Test #11 is greater than for Test #25. One of the reasons for this could be the fluid drift in Test
#11; whereas fluid drift is insignificant in Test #25. Therefore, the necessity of the test design is more critical
for the test including fluid drift. Fig. 13 (b) shows the effect of fluid drift on the mean residence volume.
As it can be seen, tsoak has a great impact on the effect of fluid drift in the SWCT test and the deviation of
the mean residence volume from the ideal one increase with tsoak. Designing the SWCT test with shorter
tsoak than "two-times rule" is the only remedy to reduce the undesirable effect of the fluid drift on the test
results. This effect is evaluated in the following.

Figure 13—(a): Deviation from the ideal mean residence volume, (b): effect of fluid drift on the mean residence volume.

Sixth stage
Designed SWCT tests in Table 3 have been modeled using minimum VSlug (F=0.25) in order to evaluate
the tracer response in the target well more precisely as shown in Fig. 14. As it can be seen, there is an
insignificant difference between the arrival times of EtAc profiles due to minor difference between tinj for
the test designs; whereas this difference in EtOH profiles is considerable due to great difference between
the β-values of the test designs. In tests with β-value of 0.2 (Tests #11-1 and #25-1), there is a short distance
between the ester and alcohol peaks as expected. It could make difficulty for the measurements at the
wellhead while this design is more in conformity with the ideal test as shown in Fig. 13 (a). In addition, it
could be difficult to satisfy β-value of 0.2 at the reservoir conditions due to lack of the ester diversity. The
modeled tests with the optimal β-value of 1.0 (Tests #11-2 and #25-2) show that the difficulty of the short
SPE-181612-MS 21

distance between the peaks of the tracers is eliminated. However, the deviation from the ideal test design
is similar to the former design with β-value of 0.2 as shown in Fig. 13 (a). There is still the difficulty of
finding the ester to satisfy β-value of 1.0 at the reservoir conditions. In the modeled tests with the applied β-
value (Tests #11-3 and #25-3), it is illustrated that there is a greater deviation from the optimal modeled test
for Test #11-3 in compare to test #25-3. This is because of the greater difference between the applied and
optimal β-value in Test #11. Therefore, the issue of the ester shortage in industry is more serious for Test #11.
From Fig. 14, the maximum EtOH concentration in Tests #11 is less than 0.02 vol%; which does not
hold criterion #2 regarding the minimum alcohol concentration of 200 (mg/L). This difficulty applies also
to the designed Tests #25-3 because of the assumption of impulsive ester injection in the workflow that is
not satisfied in modeling. As a result, the minimum ester bank (VSlug) is not be sufficient for these designs
because of the detectability problem and should be increased; however, a larger VSlug causes some other
operational difficulties as mentioned earlier. In order to study the effect of VSlug, the result of the maximum
VSlug (F=0.5) with the same ester concentration for Tests #11-2 and #25-2 is shown in Fig. 15.

Figure 14—Modeled tracer profiles for three different test designs.

From Fig. 15, it can be observed that the difficulty regarding the tracer detectability is reduced by
performing the SWCT test with larger VSlug. On the other hand, the probability of getting more non-Gaussian
and long tail profiles could be increased, which makes the test interpretation more complex. In addition,
larger VSlug is not desirable due to flammability of the esters and cost. Shipping larger VSlug is more risky
especially for offshore pilots.

Figure 15—Ester bank volume effect in Tests #11 and #25.

22 SPE-181612-MS

The effect of VSlug on the tracer profiles is evaluated using the difference between median and mode (Δ
= |median – mode|) of the modeled tracer profiles. Mode (i.e., the peak of the profiles) and the median of
the profiles are calculated using the peak median method that is given in Appendix E.
For symmetric profiles, the mode and median are equal, as it is shown in Fig. E1. Differences between
these two parameters occur with skewed profiles. As much as this difference is greater, the profile is more
skew towards higher produced volume such as tracer profiles with longer tail and non-Gaussian distribution.
Mode and median have been calculated for the tracer profiles shown in Fig. 15. The results are given
in Table 4 and Fig. 16. Results show that (1) the mode and median in the modeled profiles with minimum
VSlug is ahead of the modeled profiles with maximum VSlug. This is because VTotal and tinj is constant and push
bank is more for the smaller VSlug. Mode of the ester bank travels deeper using larger push bank. (2) Δ has
increased in the profiles with larger VSlug. The latter observation shows that the skew of profile with larger
VSlug and the probability of getting longer tail profiles increase.

Table 4—Mode and median of the tracer profiles

EtAc EtOH

VSlug Mode Median ΔEtAc Mode Median ΔEtOH

Max 7345.46 7866.9 521.44 3955.24 4522.19 566.94

Test #11-2
Min 9040.56 9402.72 362.15 5085.31 5426.62 341.30

Max 11936.34 12725.73 789.39 6327.26 7063.94 736.68

Test #25-2
Min 14690.88 15204.41 513.53 7345.44 7877.17 531.73

Figure 16—Mode and Median difference

The effective remedy to facilitate this difficulty could be using smaller VSlug with higher concentration.
However, the probability of pH-variation could be increased due to higher product acid in smaller volume.
PH-variation dependent on the formation brine compositions and buffering action, initial pH-values, and
reservoir rock contents. Study of the effect of the injected ester concentration on the pH-variation is the
objective of future work.
The best SWCT test could be designed with minimum VSlug and minimum tracer concentration if the
detectability of the tracers does not make restriction. Otherwise, the decision on VSlug and ester concentration
is dependent on other mentioned parameters. If pH-variation is negligible at higher concentration, the test
design with minimum VSlug and higher concentration could be the best design. Or else, it could be design
of a test with larger VSlug and lower ester concentration.

Effect of shut-in time ("C" coefficient)

Since reaction rate is dependent on the reservoir conditions, "two-times rule" as a general rule for all SWCT
test designs and all reservoir conditions could not be always optimal. In this section, the effect of different
SPE-181612-MS 23

"C" coefficient in tsoak is evaluated in order to adapt "two-times rule" based on the reservoir conditions for
Tests #11 and #25. Results are shown in Table 5 and Figs. 17 and 18.

Table 5—Effect of soak time

Test #11-2 (β-value = 1.0, K-value = 2.03) Mean Residence volume

"C" coefficient soak Eq. 15 Eq. 16 Eq. 18 Eq. 20

2.4 4.50 2.0 1.74 1.74 1.6

2 "two-times rule" 3.75 2.0 1.71 1.71 1.6

1.6 3.0 2.0 1.67 1.67 1.6

Test # 25-2 (β-value = 1.0, K-value = 7.33) Mean Residence volume

"C" coefficient soak Eq. 15 Eq. 16 Eq. 18 Eq. 20

2.4 4.48 2.0 1.74 1.74 1.74

2 "two-times rule" 3.73 2.0 1.71 1.71 1.71

1.6 2.98 2.0 1.67 1.67 1.67

Figure 17—Shut-in time effect in Tests #11 and #25.

Figure 18—(a): Deviation from the ideal mean residence volume, (b): effect of fluid drift on the mean residence volume.

Fig. 17 shows that tsoak affects the height of the peaks, but not their positions. While the height of the
EtOH peak affected strongly with tsoak, there is a negligible change in the height of the EtAc because of
the relatively small amount of the hydrolysis in these cases (less than 10%). Therefore, in cases where the
detectability of the product alcohol is a difficulty, designing the test using larger tsoak could be a good solution
if fluid drift is not considerable.
Fluid drift is also another effective concern that increases with tsoak and makes more deviation from the
ideal test. Shorter tsoak could facilitate the strength of fluid drift on the tracer profiles and the reliability of
24 SPE-181612-MS

the Sor determination as calculated in Table 5. Shorter tsoak ("C"=1.6) has the lowest dislocation of the ester
profile for Test #11 while there is no dislocation for Test #25 because of negligible fluid drift (Fig. 17).
In addition, shorter tsoak leads to less product acid during shut-in time and the probability of pH-variation
declines. The effect of the tsoak on the pH-variation is the objective of the future work.
On the other hand, the deviation from the ideal mean residence volume increases with shorter tsoak as
calculated in Table 5 and shown in Fig. 18 (a) in Test #25. This is because the larger ratio of the product
alcohol during transient time to shut-in time and it could increase the tailing of the tracer profiles. In Test
#11, this deviation has not changed in different tsoak because of the fluid drift effect that compensated the
desirable effect of the product alcohol ratio during transient time to shut-in time.
Decision on the "C" coefficient and "two-times rule" is dependent on fluid drift, the detectability criteria
of the tracers, ratio of the product alcohol during transient time to shut-in time, and the effect of the product
acid on the pH-variation. Larger tsoak could intensify the undesirable effect of the fluid drift and product
acid during shut-in time, whereas it could improve the ratio of the product alcohol during shut-in time to
transient time and the detectability of the product alcohol. Therefore, all of these effects should take into
consideration in order to adapt tsoak.

Conclusion
A new algorithm has been developed in an effort to design the optimal SWCT test (i.e., larger investigation
region and more conformity with the ideal test design) on the basis of the reservoir properties. The algorithm
has been programmed in six different stages. The weight of almost all effective parameters on the SWCT
test design and the deviation from the ideal test has been taken into account. All these parameters are linked
together and the decision for one of them is dependent to others.
Two sets of field data have been used to evaluate the efficiency of the uatomated algorithm for designing
the optimal SWCT test. It has been shown that "two-times rule" for test timing should be adapted based
on the fluid drift and the reservoir properties. It might be increased if the fluid drift is insignificant. The
decision on the ester bank and ester concentration should be made according to the probability of the pH-
variation and the detectability limit of tracers. Many parameters such as natural buffering in the formation
brine, rock composition, and reservoir temperature could affect this notion.
The results show that the automated algorithm is an efficient tool in order to desing the optimal SWCT
test with considering all effective parameters.

Acknowledgments
The authors would like to acknowledge financial support for this research from the Research Council
of Norway, Institute for Energy Technology (IFE) (http://www.ife.no), and the resources and staff of the
Department of Petroleum Engineering and Applied Geophysics at the Norwegian University of Science
and Technology.

References
Adler P.M. (1992). Porous media: geometry and transports. Butterworth/Heinemann.
Anton, L., & Hilfer, R. (1999). Trapping and mobilization of residual fluid during capillary desaturation in porous media,
Phys. Rev. E, 59(9), 6819–6823.
Bear J. (1988). Dynamics of fluids in porous media. Courier Dover Publications.
Blackwell, R. J. (1985). An Overiew of In-Situ Methods for Determining Remaining Oil Saturations. Society of Petroleum
Engineers. doi:10.2118/13702-MS.
Carlisle, C.T., & Kapoor, S. (1982). Development of a Rapid and Accurate Method for Determining Partition Coefficients
of Chemical Tracers Between Oils and Waters (for Single-Well Tracer Tests). Contract No. DOE/BC/10100-4.
SPE-181612-MS 25

Carlisle, C., Al-Maraghi, E., Al-Saad, B., Britton, C., Fortenberry, R., & Pope, G. (2014). One-Spot Pilot Results in the
Sabriyah-Mauddud Carbonate Formation in Kuwait Using a Novel Surfactant Formulation. Society of Petroleum
Engineers. doi:10.2118/169153-MS.
Chang, M. M., Maerefat, N. L., Tomutsa, L., & Honarpour, M. M. (1988). Evaluation and Comparison of Residual Oil
Saturation Determination Techniques. Society of Petroleum Engineers. doi:10.2118/14887-PA.
Cooke, C. E., 1971. Method of determining fluid saturations in reservoirs. United States Patent No. 3590923A.
Deans, H. A. (1971). Method of determining Fluid Saturations in Reservoirs. U.S. Patent No. 3, 623, 842.
Deans, H.A., & Carlisle, C.T. (2007). The Single-Well Chemical Tracer Test – A Method for Measuring Reservoir Fluid
Saturations In situ. Petroleum Engineering Handbook, Larry W. Lake, Editor, SPE, Volume V, Reservoir Engineering
and Petrophysics, pp 615–649.
Deans, H.A., & Majoros, S. (1980). The Single-Well Chemical Tracer Method for Measuring Residual Oil saturation.
Contract No. DOE/BC/20006-18.
Deans, H. A., & Ghosh, R. (1994). pH and Reaction Rate Changes During Single-Well Chemical Tracer Tests. Society
of Petroleum Engineers. doi:10.2118/27801-MS.
Deans, H.A., & Carlisle, C. T. (2007). The Single-Well Chemical Tracer Test - A Method for Measuring Reservoir Fluid
Saturations In Situ. Petroleum Engineering Handbook, Larry W. Lake, Editor, SPE, Volume V, Reservoir Engineering
and Petrophysics, pp 615–649.
Donaldson, E. C., & Staub, H. L. (1981). Comparison of Methods for Measurement of Oil Saturation. Society of Petroleum
Engineers. doi:10.2118/10298-MS.
Dullien F.A.L. (1992). Porous media: fluid transport and pore structure. Academic Press, London.
Ghosh, R. (1994). Effect of pH on Tracer Reaction (Ester Hydrolysis) Rate in Single Well Chemical Tracer Tests for
Residual Oil Saturation – A Quantitative Approach:, M.S. Thesis, University of Wyoming, May, 1994.
Huseby, O. K., Sagen, J., & Dugstad, O. (2012). Single Well Chemical Tracer Tests - Fast and Correct Simulations.
Society of Petroleum Engineers. doi:10.2118/155608-MS.
Khaledialidusti, R., Kleppe, J., & Enayatpour, S. (2014). Evaluation and Comparison of Available Tracer Methods for
Determining Residual Oil Saturation and Developing an Innovative Single Well Tracer Technique: Dual Salinity
Tracer. International Petroleum Technology Conference. doi:10.2523/17990-MS.
Khaledialidusti, R., Kleppe, J., & Skrettingland, K. (2015a). Numerical Interpretation of Single Well Chemical
Tracer (SWCT) Tests to Determine Residual Oil Saturation in Snorre Reservoir. Society of Petroleum Engineers.
doi:10.2118/174378-MS.
Khaledialidusti, R., Enayatpour, S., Badham, S.J., Carlisle, C.T., Kleppe, J. (2015b). An innovative technique for
determining residual and current oil saturations using a combination of Log-Inject-Log and SWCT test methods: LIL-
SWCT. J. Pet. Science and Engineering, pp. 618–625. doi:10.1016/j.petrol.2015.10.023.
Kidwell, C. M., & Guillory, A. J. (1980). A Recipe for Residual Oil Saturation Determination. Society of Petroleum
Engineers. doi:10.2118/8451-PA.
Mechergui, A., Romero, C., & Ganzo, A. (2012). Design of SWTT at High Salinity and High Temperature: Where are the
Limits? Society of Petroleum Engineers. doi:10.2118/155515-MS.
Mohanty, K. K. (1983). Multiphase Flow in Porous Media: III. Oil Mobilization, Transverse Dispersion, and Wettability.
Society of Petroleum Engineers. doi:10.2118/12127-MS.
Perry R.H., Green D.W., and Maloney, J.O. (1997). Perry's chemical engineers' handbook, 7th ed., McGraw-Hill, New
York.
Shook, G.M., & Ansley, S.L. (2004). Tracers and Tracer Testing: Design, Implementation, and Interpretation Methods.
INEEL/EXT-03-011466, Idaho National Engineering and Environmental Laboratory Bechtel BWXT Idaho, LLC.
Skrettingland, K., Holt, T., Tweheyo, M. T., & Skjevrak, I. (2011). Snorre Low-Salinity-Water Injection—Coreflooding
Experiments and Single-Well Field Pilot. Society of Petroleum Engineers. doi:10.2118/129877-PA.
Teklu, T. W., Brown, J. S., Kazemi, H., Graves, R. M., & AlSumaiti, A. M. (2013). A Critical Literature Review
of Laboratory and Field Scale Determination of Residual Oil Saturation. Society of Petroleum Engineers.
doi:10.2118/164483-MS.
UTCHEM Technical Documentation (2000). Center for Petroleum and Geosystems Engineering, The University of Texas
at Austin.
Wellington, S. L., & Richardson, E. A. (1994). Redesigned Ester Single-Well Tracer Test That Incorporates pH-Driven
Hydrolysis Rate Changes. Society of Petroleum Engineers. doi:10.2118/24135-PA.
Wellington, S. L., Simmons, J. F., Hara, S. K., & Richardson, E. A. (1994). A Single Well Test Method That Yields a
Dispersion-Free Measure of Reservoir Fluid Drift Rate. Society of Petroleum Engineers. doi:10.2118/27757-MS.
26 SPE-181612-MS

Appendix A

Correlations for calculation of K- and KH-value

K(EtAc) and KH(EtAc)
The following correlation is implemented to calculate K(EtAc) (Deans et al. 1980):

(A1)

Where Ss (mg/L) and T (°F) are formation brine salinity and reservoir temperature, respectively. The
average deviation of the correlation from the laboratory measurements is within 20% and the laboratory
measurements were done in the following temperature and salinity range:

• Temperature Range: 80 – 221 (°F)

• Salinity Range: 0 – 200000 (mg/L)

Since the other formation parameters, such as oil composition and tracer concentration, influence the
K value, the following correlations which took more effective parameters into consideration is also used
(Carlisle et al. 1982):

(A2)

Above coefficients in low salinity brine (0 < Ss < 20000 (mg/L)) and high salinity brine (20000 < Ss <
100000 (mg/L)) are listed in Table A-1:

Table A-1—Values of the constants

Deviation
from
A 2
B (10 ) C D E (10 )−4
F G (10 ) 4 −4
H (10 ) Q X (10 )−4
laboratory
measurements

0 < Ss <
1.6 5 0.2 0.7 18 0.8 18 5 1871 34 10%
20000

20000 < Ss
1.85 5 0.3 0.7 18 0.785 36 3 1871 34 5%
< 100000

Laboratory measurements were carried on under the following conditions:

• Temperature Range: 76 – 200 (°F)

• EtAc Concentration, vol%: 0.15% – 1.4%

• Gas-to-Oil Ratio (GOR): 0 – 600 (SCF/bbl)

• Oil Composition: Highly paraffinic to Highly naphthenic

In order to calculate KH(EtAc), since the proposed correlation by Deans et al. (1980) does not work
properly, the revised correlation which obtained by linear fit of the laboratory data is performed. The
hydrolysis rate in the implemented correlation mainly is a function of temperature.
(A3)
SPE-181612-MS 27

K(PrFr) and KH(PrFr)

The following correlation is implemented to calculate K(EtAc) (Deans et al. 1980):

(A4)

The average deviation of the correlation from the laboratory measurements is within 30% and the
laboratory measurements were done in the following temperature and salinity range:

• Temperature Range: 80 – 130 (°F)

• Salinity Range: 0 – 200000 (mg/L)

This correlation for K(PrFr) is just a function of salinity and Deans et al. (1980) showed that if the
dependency on salinity is considered, the dependency on temperature is negligible; however, other reservoir
parameters, such as tracer concentration and oil composition, have an effect on partitioning coefficient of
tracers.
KH(PrFr) just have dependency on temperature, non-dependence on salinity, Similar to KH(EtAc)
(A5)
28 SPE-181612-MS

Appendix B
Nc Calculation and Applied CDCs
Nc is a dimensionless term used to represent the ratio of viscous forces to capillary forces.
(B1)

Where u, μ, and σ are Darcy velocity (cm/sec), viscosity (dyne-sec/cm2), and interfacial tension (dyne/
cm), respectively. The Darcy velocity is defined as:

(B2)

Where q, r, and H are injection/production rate (cc/sec), distance from the wellbore (cm), and height of
injection interval (cm), respectively.
In this study, the CDCs were measured for sandstone systems by Mohanty (1983) and Anton et al. (1999)
is taken into consideration, as shown in Fig. B1. The key observation from CDC is the region where Sor is
constant (plateau region) and critical capillary number (Ncc) where the Sor begins to drop. It means that in
Nc lower than Ncc, viscous force is not able to change the trapped oil. The results show that Ncc is almost
10−5. The distance less than 3.28 ft (1 m) from the wellbore is considered as a capillary desaturated zone.

Figure B1—Applied CDCs for Sandstone Formations

SPE-181612-MS 29

Appendix C
Theory of the Sensitivity Analysis
In this theory, it is supposed that the total injected volume ( ) is fixed and the mean retention volume of
the alcohol (product tracer), , will vary with Sor. Then, sensitivity is as follows:

(C1)
30 SPE-181612-MS

Appendix D
Estimation of the Molecular Diffusion Coefficients
There are some mathematical correlations to estimate the diffusivity of different tracers in water at
a particular temperature. These expressions are dependent to tracer properties such as molar volume,
molecular weight and liquid viscosity. The most extensively expressions are:
1. Wilke-Chang Correlation (Perry et al. 1977):

(D1)

Where φB, MB, T, μB, and VA are the factor that accounts for hydrogen bonding (equals 2.6 for water),
the molecular weight of water, absolute temperature (K), the water viscosity (kg/m s), and specific
molar volume of the tracer at the normal boiling point (m3/kg mol).
2. Stokes-Einstein Correlation:

(D2)
3. Polson Correlation:

(D3)

MA is the molecular weight of tracer.

SPE-181612-MS 31

Appendix E
Mode and Median: Peak Median Method
Mode and median of the profiles are calculated using the peak median method which is an acceptable
approximation for the median calculation as shown below:

(E1)

Where QU and QL are the produced volume at as shown in Fig. E1.

Figure E1—Mode and Median