0% found this document useful (0 votes)
21 views22 pages

Spe 190315 Ms

The paper discusses the inter-well tracer test conducted in the La Cira Infantas oil field in Colombia, which has a century-long production history and is currently undergoing water flooding for enhanced oil recovery. The study identifies patterns with low areal efficiency and proposes conformance treatments to improve oil recovery by addressing reservoir connectivities and flow distribution. Results indicate that conformance is necessary in several mandrels, highlighting the relationship between facies architecture and water injection flow, and the potential for significant incremental oil production.

Uploaded by

372788554
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
21 views22 pages

Spe 190315 Ms

The paper discusses the inter-well tracer test conducted in the La Cira Infantas oil field in Colombia, which has a century-long production history and is currently undergoing water flooding for enhanced oil recovery. The study identifies patterns with low areal efficiency and proposes conformance treatments to improve oil recovery by addressing reservoir connectivities and flow distribution. Results indicate that conformance is necessary in several mandrels, highlighting the relationship between facies architecture and water injection flow, and the potential for significant incremental oil production.

Uploaded by

372788554
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

SPE-190315-MS

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Inter Well Tracer Test Results in the Mature Oil Field La Cira Infantas

Victor Alfonso Morales, Leyla Kristle Ramirez, Sandy Vanessa Garnica, Luz Adriana Rueda, Vicente Gomez,
Adriana Gomez, and Maria Angelica Bejarano, Occidental de Colombia LLC; G. Michael Shook, Mike Shook &
Associates

Copyright 2018, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Improved Oil Recovery Conference held in Tulsa, Oklahoma, USA, 14-18 April 2018.

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
La Cira Infantas is the oldest oil field in Colombia with approximately 100 years of production history,
located in the Middle Magdalena Valley Basin. The field production comes from the C zone reservoir of
Mugrosa Formation where the depositional environment is a fluvial meandering system. The reservoir has
a high heterogeneity and it is defined as an interbedding of sandstones, shales and siltstones with an average
thickness of 600 ft and a permeability range from 80 mD to 2 Darcy. The field has been under secondary
recovery since the 1960's and in 2005 a redevelopment of the water flooding process began. The field has
approximately 400 patterns and 1,000 active producer wells, 95% of which are under a water flooding
process. Injector wells have a selective string completion, composed of mandrels and packers, independently
injecting in different sand units. Currently, there are patterns with low areal efficiency and consequently
lower than expected recovery factor. An interwell tracer project was executed in a six pattern pilot sector,
composed of 16 distinct mandrels, in order to validate the need of a conformance treatment to improve
current conditions and have a better understanding of the reservoir connectivities. In each selected mandrel
a unique tracer family was used in order to accurately intepret breakthough results.
The workflow in the project starts by using the results of the tracer test to estimate swept volume and
flow geometry in all patterns. The swept zone represents the thief zone in each pattern and provides an
insight of how poor the areal efficiency of the pattern is. Flow geometry is represented in an F- Φ curve
and the tangent is related to the residence time of an arbitrary flow line, which is used to first recognize
the need for a conformance job and then to calculate the fraction of the swept volume needed to treat.
The last step of the workflow is to estimate the incremental oil production rates derived from treating
the thief zone. Two analytic methods were derived for the incremental oil production rate estimates. The
conformance candidates were ranked according to treatment volume vs. incremental oil recovered over a
two-year timeframe. Those results are in process of being analyzed.
The results of the inter-well tracer showed that conformance is needed in 6 individual mandrels and there
is a strong relationship between the facies architecture and the flow distribution of the injected water. Also,
it will improve the definition of the job portfolio for the conformance project which considers 80 candidates
and 2.7 MMBO resources.
2 SPE-190315-MS

The application of conformance treatments is a novelty in multilayer mature oil fields under water
flooding process in Colombia, and the study of inter well tracers is essential for the success of this IOR
technology.

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Introduction
La Cira Infantas field is a water flooded mature oil field located in the Middle Magdalena Valley Basin in
northern Colombia. The field has 100 years of production history, beginning with primary production in
1918 until 1959 when the first water flooding process began.
The producing formation is "Mugrosa", positioned geomorphologically throughout the central valley,
along the Magdalena River, between the Central and West Colombian Andes as shown in figure 1.

Figure 1—Field Location

The Mugrosa Formation is underlined by the Colorado Formation and is overlapped by the Esmeralda
Formation, both deposited during the Early Oligocene (34 m.y.a). The Mugrosa formation is the main
producer of hydrocarbon in La Cira Infantas and it is defined as an interbedding of sandstones, shales and
siltstones. The depositional environment is meandering rivers, making the reservoir an aerial and vertical
heterogeneous formation with an average thickness of 600 ft and a permeability range from 80 mD to 2
Darcy. The vertical sweep is controlled with selective completions as explained below.
Production of the field comes mainly form the lower Mugrosa formation, known as C zone, and it is
subdivided into ten operational units (C1-C, C1Ab-C, C2-C, C2DT-C, Gtb-C, C3-C, C3Cb-C, C4-C, C4Cb-
C and C5-C). The development of the field has also focused production in these units. The sandstones of
C2-C and C2Dt-C units are continuous and have optimal petrophysical characteristics for production and
injection of water in the secondary process; whereas the lower C zone, from C3Cb-C to C5-C, sands are
more discontinuous and have poorest petrophysical characteristics.
SPE-190315-MS 3

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 2—Stratigraphic Column of the basin and type well

In 2005, Oxy and Ecopetrol initiated a joint venture of a new redefined water flooding process currently
ongoing for 13 years. Since the joint venture, the field has expanded to 400 patterns and 1,000 active
producer wells, 95% of which are under a water flooding process. The design of the field considers 20 to 25
acreage area per pattern on average and inverted 5 spot and 7 spot patterns. Since the joint venture, injector
wells have selective strings with mandrels which allows control on the vertical distribution and volume of
water injected per mandrel group. Each mandrel encompasses a group of sand intervals. Injection profiles
are regularly run on injectors to surveil reservoir vertical efficiency.
Since the redevelopment of the water flooding process in 2005 oil recovery has been successful, achieving
an incremental 3% RF in 13 years, compared to the first water flooding process (1959-2005) which
accomplished an incremental RF of 3.8% in 46 years. The good results indicate that the inclusion of new
producer and injector wells, the vertical control on injector wells and the continuous production surveillance
and reservoir surveillance process have positively contributed.
Despite the good results in production, there are injector wells with low vertical efficiency and patterns
areas with low areal efficiency contributing to a low volumetric efficiency and ultimately a lower than
expected recovery factor. Figure 1 below shows a plot of recovery factor versus dimensionless water injected
for an example area. The dotted black line represents the average performance of patterns for the example
area. Patterns above the reference line (green shaded area) have a good performance and patterns below the
reference line (red shaded area) are underperforming. It can be seen that there are many underperforming
patterns (almost half the patterns of the area) that need to be targeted with a solution in order to improve
their secondary recovery factor.
4 SPE-190315-MS

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 3—Secondary recovery factor versus Dimenssionless Water Injected (DWI)

A proposed solution to the underperforming pattern conditions is to apply a deep conformance treatment,
in which the objective is to obstruct the identified thief zone (high recirculation-high heterogeneity zone),
and divert water flow to less permeable sands with higher remanent oil reserves. In this context, the idea is
not to remediate vertical efficiency but rather to significantly improve the reservoir areal sweep efficiency.
Prior to the execution of gel injection, an inter well tracer project was executed over a time frame of six
months in order to validate the need of a gel injection job. The discussion of this paper focuses on the
planning, execution, monitoring & control, and closure of the inter-well tracer application.

Well Candidate Selection


Due to the large volume of patterns in the field, two methodologies were applied in order to narrow down
the selection and correctly identify the most critical patterns that have low areal efficiency – low volumetric
efficiency and hence lower than expected recovery factor.
The first methodology used is the Weighted Parameters Methodology and the second methodology used
is the Efficiency vs. Loss Plot methodology.

Weighted parameters methodology


For this methodology 7 secondary recovery variables were selected. Table 1 below shows a summary of
the selected variables.
SPE-190315-MS 5

Table 1—Weighted paramaters methodology

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
For the application of this methodology the following steps were followed:
1. Perform a statistical analysis with all seven variables per area
2. Select the median of each variable per area
3. For the first six variables, filter patterns if they have values above or below the median. If the pattern
value is above the median, the number 1 is assigned, and on the contrary the number 0 is assigned.
4. For the vertical efficiency variable if the pattern value is below or equal to the median, the number
1 is assigned and on the contrary the number 0 is assigned.
5. Add the values assigned to each variable in one column. The maximum score obtained with the
sum of all variables is 7 points.
6. The patterns that obtained a score between 6 points and 7 points are considered priority 1, the
patterns that obtained a score between 4 points and 5 points are considered priority 2 and the
patterns that obtained a score below 4 points are not considered critical.
7. This process is repeated in each area of the field. It is important to mention that it can also be
extended to a specific sector of an area to obtain more detailed results of the water flooding
performance for a specific number of selected patterns.
The results of each variable for all patterns of an area were added and the patterns with the highest score
were selected for analysis.

Efficiency vs. Loss Plot methodology


The efficiency vs. loss plot methodology considers three additional variables not considered in the first
methodology: volumetric sweep efficiency, dimensionless water injected (DWI) and dimensionless total
water produced (DTP). The dimensionless variables used in this context are discussed in the methodology
created by Simmons and Falls (2005) and further explained by Morales and Ramirez (2018).
This second methodology basis its analysis in a plot with four quadrants as shown in the image below.
6 SPE-190315-MS

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 4—Efficiency vs. Loss Plot methodology

The y-axis is represented by the division of the volumetric efficiency according to material balance and
volumetric efficiency according to fractional flow.
Volumetric efficiency according to material balance.

Equation 1. Volumetric sweep efficiency according to material balance

where

• Np: Information gathered from official production data

• STOOIP: Volumetric information gathered from official area models built in PETREL Software

• Ed: Value of 60% used.

Equation 2. Displacement sweep efficiency

Volumetric efficiency according to fractional flow.

Equation 3. Volumetric sweep efficiency according to fractional flow


SPE-190315-MS 7

where

• So: Current oil saturation, value obtained from the fractional flow curve and looking up at the
saturation value corresponding to the current water cut in the pattern. Value varies per pattern.
• Sorw: Residual oil saturation, value obtained from performed core analysis and imbibition test.

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Value: 30%.
• Soi: Initial oil saturation, value obtained using core analysis and calculated from Archie's method.
Value: 70%
• Ed: Value of 60% used.

The x-axis is represented by the subtraction of dimensionless water injected (DWI) – dimensionless total
water produced (DTP).

• Upper left quadrant: The upper left quadrant on the y-axis represents good conformance. Pattern
that have similar and good volumetric efficiencies using both the fractional flow and the material
balance methods. On the x-axis, it represents patterns that have near 1 voidage or balance ratio. In
other words, water injected is similarly quantified by water produced.
• Upper right quadrant: The upper right quadrant represents patterns with out of zone or out of area
injection. The x-axis loss is higher than 0.2 indicating that more water is being injected compared
to water being produced. Inefficient process of water injected.
• Lower left quadrant: The lower left quadrant represents patterns with thief zones. Efficiencies
calculated with the fractional flow method are much lower than efficiencies calculated with
material balance method. On the x-axis the loss is not very much, indicating that the water injected
is similar to water produced. However it may be through one specific sand unit. This information
can be validated with historical injection profiles.
• Lower right quadrant: The lower right quadrant represents patterns with reservoir and well
conformance issues. Efficiencies are low and losses are high indicating the process is being
inefficient.

Interpretation
Methodology 1 and methodology 2 were compared for each of the six areas. Results showed that there is a
correlation of the critical patterns that outstand for each area using both methodologies. In total, 80 patterns
were identified distributed in all six areas. Figure 5 below shows the distribution of identified candidates
per area:
8 SPE-190315-MS

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 5—Macro analysis candidate selection

Mandrel Candidate Selection


As previously mentioned, injector wells in La Cira Infantas have a selective string completion composed
of packers and mandrels. This type of completion allows to target sand units independently. As a result of
this strategy, which has been very successful in accelerating reserves recovery, there are sand units with
varying levels of maturity. The variation in maturity is attributed mainly to petrophysical properties and to
the mechanical restriction given to each mandrel.
Figure 6 below represents the cumulative water injection per sand unit, for each of the 10 operational
units open in an injector well. In Figure 6 it can be noted that sand unit C2-C, Gtb-C and C3-C are the
sands units that have the highest cumulative volumes with 4.5 MM BW, 2.63 MM BW and 3.28 MM BW
respectively. It is important to keep in mind that some sand units may had been perforated at a different date
and therefore the time frame of injection needs to be taken into account.

Figure 6—Cumulative water injected pattern example

A second consideration for the mandrel selection is to verify that the suspicious sand units have indeed
lateral continuity with neighboring producer wells. In other words, if the gel treatment is focused on a sand
SPE-190315-MS 9

body, the end results must have a collateral impact represented by incremental oil production in the same
sand body. Incremental production from other sand bodies is beneficial and will most definitely add value
to the project but it is not considered as a targeted result. As seen in Figure 7, example injector well A has
been selected and the selected sand unit is open in all neighboring producer wells. Thus, in the best case

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
scenario, it is expected to have an impact in all producer wells improving areal efficiency.

Figure 7—Pattern correlation of targeted sand unit

Pilot Selection
After evaluation of the candidate patterns in each area, 4 patterns in Area B and 2 patterns in Area C were
selected as shown in Figure 8.

Figure 8—Pilot sector map

Area B patterns
The 4 patterns in area B vary in maturity and secondary recovery performance. All patterns have different
secondary recovery factors and even though they are neighboring patterns they each have preferential
injection in different sand units. The differences in each pattern is beneficial in providing an insight of
the results under different circumstances. Additionally, the patterns are relatively confined, separated by
a normal fault southwest. Northeast and northwest the patterns are not confined and possible influences
beyond the selected patterns need to be considered.
Figure 9 below shows the different recovery factors and maturity, represented by DWI, of each pattern.
It can be noted that Pattern 4 has the highest secondary recovery factor (21%) whereas Pattern 1 has the
lowest recovery factor (6%). In terms of DWI, Pattern 1 is the most mature pattern in the selected sand units
(DWI>4) whereas Pattern 2 is the least mature (DWI < 2). For Pattern 1 there is a correlation of highest
10 SPE-190315-MS

cumulative volumes of water injected with least recovery efficiency which translates to poor volumetric
sweep efficiency.

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 9—Area B – Pattern description

Table 2 below shows a summary of the selected mandrels for each of the patterns in Area B.

Table 2—Mandrel and sand unit selection in Area B

Area C patterns
In area C two patterns were selected. The purpose of selecting Pattern 1 was to compare results of an isolated
pattern versus a group of patterns treated together. Figure 10 below shows the characteristics of Pattern 1.
On the other hand, pattern 2 was selected for analysis prior to the execution of a polymer treatment, to verify
or discard the presence of possible thief zones.
SPE-190315-MS 11

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 10—Area C – Pattern description

Table 3 below shows a summary of the selected mandrels for each of the patterns selected in Area C.

Table 3—Mandrel and sand unit selection in Area C

In total, 16 mandrels were selected; 11 in Area B and 5 in Area C.

General Tracer Test Considerations


Chemical tracer used
Fluorobenzoic acid (FBA's) chemical tracers were used for this project. FBA's have been successfully used
as hydrogeological traces and towards the end of the 90's they successfully emerged in the oil industry.
They are compatible in both aqueous fluid as salts and in organic based fluids as acids. These tracers are
made up of a large family or organic molecules that can be measured in one same chromatographic run.
This translates in time optimization, especially for multilayer or multi mandrel tracer projects.
FBA's are detected and quantified through a treatment process in laboratory and chromatography for
concentrations below 1 ppt. The most common species used in the oil industry are 2-FBA, 4-FBA, 2.6-
FBA, 3.4-FBA, 3.5-FBA, 2.4.5-FBA y 2.3.4.5-FBA as shown in Figure 11.
12 SPE-190315-MS

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 11—Fluorobenzoic families

The tracer species for each injector well and mandrel must be selected so that there are no interferences
during the analysis of results of neighboring patterns. For this reason, 10 FBA tracer species were used
and unique tracer species for the injector wells that had more than one mandrel were considered as shown
in Table 4.

Table 4—FBA's families used for each mandrel

Monitoring schedule
In order to detect all possible tracer breakthroughs, a monitoring schedule was implemented in which first
line producer wells were tested at least 30 times and second line producer wells were tested at least 12 times
throughout the six months' time frame of the project. The first test was executed six hours after the first
tracer injection.
Table 5 below summarizes the frequency of the tests for both 1st line and 2nd line producer wells. For
1st line producer wells, it can be noted that beginning the project the frequency was high, testing the wells
almost daily. After the first month the frequency decreased, testing the wells every 5 days. After the third
month the frequency continued to decrease, testing the wells every 10 days. The frequency was high at the
beginning the project expecting an early tracer breakthrough.
SPE-190315-MS 13

Table 5—Monitoring schedule summary

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Analysis schedule
An analysis schedule was designed in accordance to the monitoring schedule. The analysis was performed
in Nalco's local laboratory in Bogota and well samples were analyzed with the use of gas chromatography.
As it can be seen in Table 6, laboratory of the samples began five days after injection of the tracer since an
initial early breakthrough was expected. All first line producer wells were tested with an average frequency
of 10 days. In those wells were breakthrough was evidenced, a higher analysis frequency was used with an
average of five days (highlighted cells in Table 6).
Second line producer wells were analyzed twice during the six-month timeframe of the project. In both
cases no evidence of breakthrough was evidenced for these wells.
In total, 436 samples were analyzed for first line producer wells and 40 samples were analyzed for second
line producer wells, which represents a 43% analysis of all collected samples.

Table 6—Analysis schedule summary

Injection Profiles
In order to monitor water flow through the mandrels were chemical tracer was injected, injection profiles
were run to quantify volume and validate differences over time. Overall, an average of 3 injection profiles
were run for each pattern over the time frame of six months of the inter-well tracer project. Table 6 shows
the water distribution for each pattern and their corresponding mandrel.
In some patterns water flow significantly varies between one injection profile and another one. For
example, in both mandrels of Pattern 2 there is an important difference between the 1st injection profile and
the 2nd injection profile, where in M5 there is a delta of ~420 BWIPD and in M3 & M4 there is a delta of
~680 BWIPD. For this particular pattern, changes were made to the flow regulator valve (FRV) in order to
adjust flow to the recommended volume estimated for that zone. Even though the changes done to the FRV
were small it appears to have had a significant impact on the vertical distribution of the well. In Pattern 3
no modifcations were done to the FRV during the time frame of the project and it is observed that water
14 SPE-190315-MS

flow is maintained relatively constant over time. Patterns 1 and 4 from Area B and Pattern 1 from Area C
had minor FRV modifications.

Table 7—Injection profiles summary

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Results
Results in Area B
Figure 12 shows a schematic of all tracer breakthroughs in the four patterns of Area B.
In Pattern 1 three tracer families were used, each targeting different sand units in M2, M3 and M4.
Tracer breakthrough from all three families was detected primarily in producer well 1 after 45 days, with
high concentration values. After 100 days, tracer breakthough from M3 and M4 was detected with lower
concentration values in producer well 2. Moreover, after 150 days, less critical breakthrough was detected
in producer well 5. In Pattern 2 two tracer families were used, one for zones M3&M4 and a second one
for zone M5. The strongest connectivity was evidenced towards producer well 5 after 33 days. In Pattern
3, the strongest connectivity was evidenced towards producer well 14 and producer well 10, both after 50
days however producer well 14 had higher concentration peaks. Also, in pattern 3 a strong connectivity
was evidenced in producer well 7, however producer well 7 had a workover intervention in which the sand
unit being traced in the injector well was isolated. After the workover intervention, tracer mass was no
longer recovered. In pattern 4, the strongest connectivities was evidenced in producer well 7 after 35 days,
followed by producer well 5 after 43 days. These results confirm the importance of geology on the flow
field and hence areal sweep for all mandrells tested.
SPE-190315-MS 15

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 12—Area B tracer breakthroughs

Figure 13 below shows a schematic of the concentration curve, in parts per billion, versus time for the
four patterns in Area B.

Figure 13—Area B concentration versus time curves

For the tracer accumulation interpretation, the injected water flow is taken into account and the use of
injection profiles to validate consistency in the flow is essential for the interpretation of results.

Results in Area C
Figure 14 below shows a schematic of Pattern 1 and Pattern 2 with the corresponding producer wells and the
most relevant tracer breakthroughs. In Pattern 1, two groups were traced independently: M2/M3 and M3/
16 SPE-190315-MS

M4/M5. No important breakthroughs were detected in the group of M3/M4/M5. In group M2/M3 important
tracer breakthrough were detected in producer well 3 and 5 after 50 days. Moreover, after 93 days, tracer
breakthrough was detected in producer well 4, however this coincides with an increase in injection in M2
and M3. In the last analyzed samples of month 6, tracer breakthrough was detected in producer well 4 from

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
group M3/M4/M5 however the concentration was low and considering the breakthrough delay, it is not
considered a relevant connection. In producer wells 1 and 2 no tracer breakthrough was detected during
the timeframe of the project.
In Pattern 2, one tracer family was used for all sand units. As it can be seen, the most important
breakthrough was towards producer well 10 with a breakthrough time of 42 days. Towards the end of the
sampling phase a small concentration breakthrough was detected in producer well 6 after 150 days. Inter
well tracer in Pattern 2 had a different objective compared to the other five patterns. The main purpose was
to identify if there were any thief zones present prior to the start of a polymer pilot project.

Figure 14—Area C Pattern 1 and Pattern 2 tracer breakthoughs

Figure 15 shows a schematic of the concentration curves in parts per billion versus time for Pattern 1.
Producer wells 5, 3 and 4 have a cumulative tracer mass recovery of 17%, 12% and 11% respectively for a
total accumulation of 41% by the end of the monitoring phase. In group M3/M4/M5 a cumulative of 1.7%
of mass recovered.

Figure 15—Area C Pattern 1 concentration curves


SPE-190315-MS 17

Figure 16 shows a schematic of the concentration curves in parts per billion for Pattern 2. For producer
well 10 there is a cumulative tracer mass recovery of 44% during the timeframe of the project.

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Figure 16—Area C Pattern 2 concentration curves

Interpretation Workflow
Swept volume estimation
Analytic interpretation of tracer data was first developed in the chemical engineering field by Danckwerts
(1953) for chemical reactor beds, and then extended by Deans (1978) to oil reservoirs. The method
of moments is an analytical solution for the swept volume that doesn't depend on (unknown) reservoir
properties in the calculations. The following form of the equations is from Shook et al. (2009) and Shook et
al. (2016). For a conservative tracer injected as a slug with volume VSlug over time ts, the total volume swept
for a given well pair, VSij, is determined from tracer concentration histories at the production well as follows:

Equation 4

Equation 5

In these equations, Cij is concentration of tracer κ injected into well i and observed at well j in units of
mass per volume, mij is the mass of tracer produced, Mκ is the total mass of tracer κ injected and qi and qj
are the injection and production rates, respectively. In general, a tracer injected into well i flows to more
than one production well j. The fraction of volumetric injection rate from well i that flows to a particular
production well j is just the mass fraction of tracer produced at well j (mij/Mκ). The swept pore volume
VSij is the aqueous pore volume between well i and j. The total swept pore volume between well i and j
is the aqueous pore volume divided by the water saturation. The units for volume swept depend on units
for injection rates. When the injection rate fluctuates, we have found it is best to use an average value in
these equations.

Extrapolating the concentration curve


Tracer tests are terminated before the tracer response goes to zero because the tracer concentration decreases
to values less than the detection limit, or the concentration data become too noisy for accurate calculations,
or simply because the sampling is terminated for one reason or another. However, failure to account for the
18 SPE-190315-MS

tracer tail will underestimate the swept pore volume. Tracer concentrations frequently decline exponentially,
such that ln(C) vs. time is linear. In that case, the tracer concentration tail can be described as follows:
Equation 6

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Equation 4 is rewritten as

(4)

The second integral in numerator and denominator can be evaluated in closed form, so the volume swept
is estimated from incomplete tracer recovery as:

(5)

As powerful as exponential extrapolation is, it is difficult to use in cases where multiple peaks occur, or
in cases where exponentially decline isn't established yet. To fill this need, we evaluated curve fitting a log
normal distribution to the tracer dataset. To describe concentration vs. time as a log-normal distribution,
the following form is used.

Equation 7

where σlnt is the standard deviation and μlnt is the mean of the distribution. By defining the mean and standard
deviation, a unique probability distribution function (PDF) can be generated. However, the mean of the
tracer history isn't known until a very late time; we can only observe the mode at early times. The standard
deviation, which is related to the heterogeneity, is also not known until late time. We have had success
in initializing σlnt and μlnt iteratively based on the shape of the tracer history (or some knowledge of the
heterogeneity as shown in Eq 8).
Equation 8
Two changes are required before fitting tracer data: the PDF generated in Excel is normalized, such that

Equation 9

The tracer data has units of μg/l. We use a scaling parameter, Co to match the data. Furthermore, the PDF
should be 0 prior to breakthrough. The final form of the equation we use is
SPE-190315-MS 19

Equation 10

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
where t* = (t-tBT).
The strength of moment analysis cannot be overstated, in that there are no assumptions regarding medium
geometry (e.g., fractures) or other non-ideal conditions. Swept volume is the single most important measure
of flood performance, yet it is a single measure of performance. The residence time distribution of the
produced tracers can be generalized to determine sweep efficiency and flow geometry as described in Flow
Capacity – Storage Capacity diagrams in a simple fashion.

F – PHI Curve
Flow capacity – storage capacity diagrams have appeared in reservoir engineering literature for decades
(Stiles, 1949; Schmalz and Rahme, 1950; Lake, 1989; Gunter et al., 1997). Also known as F-C curves, they
were originally derived for 2-D, vertical cross section, non-communicating, layered reservoirs. Each layer
has different permeability, k, porosity, φ, and thickness, h but the properties are constant in a given layer.
Under those conditions, flow capacity, F, and storage capacity, C are readily calculated.
These properties are also readily calculated from tracer data by the use of the tracer residence time
distribution (Shook et al. 2009). Recall the mean residence time (or volume) is the time-weighted average
residence time of all the flow paths connecting a given injector with a producer. Similarly, the mean
residence volume of flow paths faster than, say, one breaking through at time t is given as:

Equation 11

Normalizing Eq. 14 by the total mean residence volume gives the fraction of the total swept volume that
is completely swept at time t. To emphasize the differences between the static calculation in and this method
based on tracer interpretation, we rename storage Φ:

Equation 12

Flow capacity is likewise straightforward to estimate from tracers. Implicit in tracer interpretation is that
the tracer does not affect the flow paths; it flows with the injected fluid. Fractional recovery of tracer is
therefore directly related to the relative volumetric flow rate of flow paths carrying injected fluid, and so
F is defined as:

Equation 13.
20 SPE-190315-MS

F- Φ curves are used qualitatively and quantitatively, for example "60% of the flow is coming from 12%
of the pore volume." We can also quantify the heterogeneity by defining the Lorenz Coefficient as:

Equation 14

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
A Lorenz coefficient of zero falls along the 45° line on the F- Φ curve, which is a homogeneous
displacement (equal volumetric flow from every incremental pore volume). A value of 1 is referred to as
"infinitely heterogeneous"; it can be interpreted as "all of the flow comes from a small (approaching 0)
portion of the pore volume.
We also show the F- Φ curve is useful in estimating the number of different flow paths in a reservoir
(defined by the same residence time). This is one characteristic property of F- Φ curves that is valuable in
estimating incremental oil recovered after conformance work.
Interwell swept pore volume can also be calculated as a function of time using the Residence Time
Distribution Analysis (RTDA) method (Shook et al. 2009) as

Equation 15

Wu et al. (2008) showed for a case of constant injection and production rates, the slope of the F- Φ curve is

Equation 16

where τ is the residence time of a given streamline and t* is the mean residence time. Of course, if the tracer
only flows through the thief zone the flow capacity curve will reflect only the heterogeneity of the thief
zone itself. In this case, the total pore volume swept will be low relative to expectations from the geologic
model, and so the F- Φ curve, total volume swept and knowledge of the geology is required to determine
the occurrence of fluid cycling.
We can use that information to identify the fraction of the swept volume that requires conformance
treatment. We can, for example, decide to shut off all flow paths with residence time less than five times
the average, find the point on the F- Φ curve with a slope equal to 5, and read Φ at that point; that is the
fraction of the total pore volume swept that requires treatment. The total swept volume is determined from
the mean residence time of the tracer, thus converting Φ to a true volume. Of course, some of the fluid does
not enter the thief zone, and so total treatment volume (VT) is given from Izgec and Shook (2012):

Equation 17

where Vs is the total pore volume swept from RTDA and F and Φ are the points on the F- Φ diagram where
the slope = 5. In cases where the volume swept is consistent with geologic expectations, this approach has
be proven; this was not the case here: in no pattern/mandrel was the volume swept large fraction of the
pore volume expected. In a majority of cases, the volume swept and the Lorenz coefficient were both low,
indicating the entire swept volume was thief zone. In other cases, the volume swept was low and so the F-
Φ curve is unfairly biased towards the thief zone residents times, making the calcutation of the treatment
volume a little harder. We used a fraction of the swept volume for the calcutations shown in Table 8 below
– the fraction is a function of the pattern Lc and volume swept, but it is premature to publish the method
without confirmation. We are waiting for results now, and will modify the method as needed.
SPE-190315-MS 21

Table 8—Interpretation results

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Conclusions
After incorporating a thorough workflow for analysis and interpretation, several key points outstand from
the application of chemical tracers in the pilot sector:

• A strong relationship is evidenced between the sand bodies with the same geological facies and the
flow distribution of injected water. Sand units with same facies are more likely to have a stronger
connectivity compared to same level sand units with different geological facies. Even though same
level sand units may have different geological facies, it does not mean they are not hydraulically
connected, but rather, injection response time is prolongued.
• The pilot sector does not have critical channeling issues. Expected breakthrough times were
approximately 5 days, however, the most rapid breakthrough time occurred 30 days after. Even
though the breakthrough time was delayed from the expected time, the swept volume calculations
indicate that there is an important pore volume not being contacted and therefore expected
incremental recovery oil estimates are attractive.
• For a pattern, the sand units that have lower net sand thickness and same geological facies are more
likely to have channeling issues. This is due to the fact that, in the presence of a thief zone, injected
water has a lower pore volume to flow through (lower breakthrough times) compared to higher net
sand thicknes sand units (higher breakthrough times).

Recommendations
The following recommendations will help in having accurate results for the final interpretation of an inter-
well tracer project:

• Avoid the execution of workover jobs in producer and injector wells during the monitoring and
sampling phase of the inter-well tracer project. Alterating production performance in a pattern will
interfere with the flow distribution and, ultimately, with the treatment volume calculations.
• Increase the sampling and analysis measurements towards the real breakthrough time of tracer. In
some cases, the sampling and analysis schedule is adjusted only to the expected breakthrough time
however it must be re adjusted to have a solid population of measurements towards the high peaks
of tracer mass concentration as well as when a decrease in the peak of tracer mass is evidenced.
• In a project were injector wells are completed with mandrels and packers, as it is the case in
LCI field, it is recommended to trace the identified candidate mandrels independently rather than
combined. This will eliminate doubt for which mandrel was more critical and will aid in having
concrete conclusions regarding the responsible sand unit for the channeling issue.
22 SPE-190315-MS

Acknowledgements
The authors wish to thank the management of Occidental de Colombia for permission to publish this paper.
The authors would also like to recognize the LCI reservoir management team as well as the World Wide
Engineering team who continuously supported and guided the execution of this project: Martin Wolff and

Downloaded from http://onepetro.org/SPEIOR/proceedings-pdf/18IOR/18IOR/D051S028R006/3994193/spe-190315-ms.pdf by Chongqing University of Science & Technology, Chong Qing on 29 December 2024
Ed Dezabala.

References
See attached reference list.

You might also like