Sunteți pe pagina 1din 23

SPE-185716-MS

Simulation Study of In-Depth Gel Treatment in Heterogeneous Reservoirs


with Sensitivity Analyses

T. Khamees, R. E. Flori, and M. Wei, Missouri University of Science & Technology

Copyright 2017, Society of Petroleum Engineers

This paper was prepared for presentation at the 2017 SPE Western Regional Meeting held in Bakersfield, California, USA, 23 April 2017.

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
In-depth gel treatment has become an attractive and optimum technology for remedying any problems
that cause poor sweep efficiency, such as heterogeneity of the reservoir and unfavorable mobility ratio
(due to high oil viscosity). The low recovery factor resulted from the difference in reservoir properties
especially reservoir permeability, would lead eventually to use of the chemical materials such as polymer
and gel to correct this condition. A comprehensive simulation study of deep placement of weak gel in thick
heterogeneous reservoir is presented in this paper. A conceptual model with quarter of inverted nine spot
pattern using CMG-STARS commercial simulator was built, to demonstrate the effectiveness of in-depth
gel treatment in correcting the heterogeneity in this thick reservoir. The model consists of one injector and
three producers with three layers of different permeabilities and thicknesses. These wells are perforated in
all layers of the model (i.e., layers 1, 2, & 3).
The results showed that injecting even a low PV of gel into high permeability layers could make a
remarkable increase in oil recovery factor and incremental oil over the base case water flooding. Polymer
and gel degradation rate have a significant impact on the reservoir performance after the treatment. Three
different scenarios for both polymer and gel degradation are considered: no degradation, 2 years and 4 years
degradation. The results were showed that always the runs without degradation yielded higher recovery
factor regardless the injected PV. In addition, when gel is injected into only high permeability layers a
higher incremental oil and higher oil recovery factor were obtained in comparison with runs when gel is
injected into all three layers. Moreover, the more homogeneous the reservoir is, the higher recovery factor
could be obtained. However, in this study, changing the crossflow (kv/kh) value has no effect on oil recovery
values over the ranges selected (i.e., 0.001, 0.005, 0.01, and 0.02). The best time to start gel injection is
also investigated, it was revealed that it is better to start gel treatment when water cut is about 80%, because
before that time the high permeability layers may still contain a moveable oil that can be recovered. Another
set of runs were carried out to show the importance of injecting polymer and gel together. The results
showed that when gel is injected into all three layers with 4 years gel degradation and 2 years polymer
degradation, polymer flooding then gel treatment yielded better results than gel treatment then polymer
flooding. Finally, a sensitivity analyses, by CMG-CMOST, was carried out to show the importance of in-
situ parameters (i.e., k1, k2, k3, kv/kh, RRFT, and thicknesses of the layers) and operating parameters (i.e.,
polymer properties, injection pressure, and injection time) on cumulative oil production and oil recovery
2 SPE-185716-MS

factor. An optimum values have been obtained that yielded the highest cumulative oil production and highest
oil recovery factor. All gel treatment runs had been compared with water and polymer flooding to evaluate
the diversion potential of in-depth gel treatment.

Introduction
In-depth gel treatment has several advantages compared with near-wellbore gel treatment. If there is
communication between layers by crossflow, near-wellbore gel treatment will not be effective, because
injected water will fairly quickly find a way around the placed blockage. Additionally, placing gel away
from the wellbore, where a high pressure gradient exist, will result in less injectivity problem (Fletcher et
al., 1992).
The use of these chemicals, which considered as an enhanced oil recovery methods, are varied according
to the existing problem in the reservoir and the source of water production. Conformance problem which
caused by the existence of different layers with different permeability (i.e., highly heterogeneous or naturally
fractured reservoirs) is one of the main source of excess water production leading to unevenness in the
flood front. Accordingly, an extensive experimental works supported by pilot tests were carried out for more
than three decades to select the best strategy to attack excess water production. In addition, a huge efforts
and money were spent in order to select the best combination of chemicals' (i.e., polymer and crosslinker)
concentration and type, which are suitable for different reservoir conditions.
As an alternative way to design and study a given combination of polymer and crosslinker that can be
used in the field application, is the use of numerical modeling. There are different commercial and inhouse
simulators such as CMG-STARS, ECLIPSE, UTCHEM, and UTGEL (University of Texas at Austin in
house simulators), which can be used to model chemical enhanced oil recovery. The uses of these simulators
are ranging from validate a given experimental work, build a conceptual model, and/or build a pilot model
depending on real field data supported by experimental work. Numerical modeling of the weak gel as an
in-depth fluid diversion in porous media had been carried out by using POL-GEL simulator (Shiyi et al.,
2000). The effect of polymer concentration, the gelation time, the kinetic of reaction on the formation of
LCFG (low concentration flowing gel), the influence of transition pressure of LCFG on the mobility of
LCFG in porous media, and the effect of viscosity and permeability reduction on plugging function were
considered in this model. They concluded that the optimum polymer concentration is 400 ppm and the
optimum crosslinker concentration is 20 ppm while maintaining the injection rate at 0.2 PV.
A numerical study using the ECLIPSE Black Oil Polymer option to model a polymer gel treatment
pilot test without mechanical isolation had been developed for HPHT reservoir conditions (Herbas et al.,
2004). They used MARCIT (Marathon Conformance Improvement Technology), which is a low MW
polyacrylamide crosslinked with Cr(III)-carboxylate. The model consist of five hydraulic units separated
by a thick shale layers between each one of them. They modeled five radially induced fractures in two of
these five units, which have the highest permeability. These fractures are extended 30 feet radially. The
results from the simulation suggested that induced fractures in sandstones, as they were modeled with the
available information, are not sensitive to the recovery after the gel treatment.
A mathematical model had been developed to simulate selective flow behavior of polymer gel into
high permeability layers (Hofner et al., 1992). The authors demonstrated by laboratory coreflood data that
certain xanthan-based profile control gel exhibit the property of "selective penetration", thus these gel flow
preferentially into higher permeability zones and eliminate the need for mechanical zone isolation. An
experimental and numerical study to investigate the competition for chromium between xanthan biopolymer
and resident clays in sandstones had been carried out (Garver et al., 1989). The authors aimed to explain the
reasons behind the slow transport of the Cr(III), which rendered the gel treatment as an ineffective process.
They concluded that, numerical model of in-situ gelation system could help to give a better idea on the
amount of chromium that is necessary to make the gelation possible and to produce strong gel.
SPE-185716-MS 3

An in-depth fluid diversion process is different from conventional gel treatment (Sorbie and Seright,
1992). In conventional gel treatment, there is no postflush prior to gelation. On the contrary, in their method
they used water to displace the gelant into predetermined distance in high permeability layer. The gelant of
low viscosity will penetrate to small distance in low permeability layer. The rear of the gelant in the most-
permeable zone outruns the front of gelant bank in an adjacent less-permeable zone. However, one important
limitation to this process; once beyond the gel bank in the most-permeable zone, fluids can crossflow back
into the high permeability channel.
A gelation model that describes retention mechanisms and gelation strength using UTCHEM simulator
had been developed (Stavland et al. 1994). They used xanthan/Cr3+ as a gel system and the chemical
reactions are dominated by the pH and dissolution rates of carbonates (i.e., the retention is to depend strongly
on the pH level). They concluded that gelation strength increases by increasing polymer concentration,
whereas increased injection rates decreases the gelation strength.
A comprehensive study of deep placement of a polymer gel system in layered reservoirs had been
developed (Ghahfaroski et al. 2016). The controlled release of crosslinkers for delayed gelation is modeled
by long half-life of the gelants assuming a simple model of reaction kinetics. Results showed that
permeability modification of the thief zone and the long gelation time played an important role in in-depth
gel treatment.
To define the reservoir characteristics which lead to successful gel treatment, several cases had been run
(Tsau et al. 1985). The results showed that a low level of crossflow is conducive to a successful vertical
conformance treatment, and the contrast in the permeability-thickness product is a useful parameters in
identifying treatment candidates
To improve sweep efficiency, the viscosity of the displacing fluid (i.e., water) should be increased or the
thief zones should be blocked off. Gel treatment technology combines the function of lowering the mobility
ratio by polymer flooding, and correcting the heterogeneity by crosslinking polymer with crosslinker to
distribute post-water injection evenly after the treatment. CMG-STARS software was used to model the
effect of gel injection with and without gel degradation, injection of gel into all layers in the model in
comparison with gel injection into thief zones only were modeled. In addition, the effect of formation
heterogeneity, crossflow, gel injection start time, and a combination injection of polymer and gel also
investigated. Finally, sensitivity analysis using CMG-CMOST also utilized to find out the best combination
of in-situ parameters and operating parameters that yields the higher cumulative oil production and higher
oil recovery factor.

Description of the Simulation Model


The reservoir model of this study is one-quarter of inverted nine spot pattern with one injector and three
producers. Two reasons behind selection of this pattern. The first reason, this pattern will maximize the
production rate, because it has high ratio of producers to injectors (three producers and one injector). The
second reason, in most cases the directional permeability in not quantified, the inverted nine spot provides
the flexibility to re-align the pattern as directional permeability becomes apparent from performance. The
selected well distance is 1170 ft, and the total net thickness is 164 ft. the four wells are completed through
all three layers. The size of the model is 1170 ft × 1170 ft × 164 ft with 30 × 30 × 3 gridblocks, the size of
each gridblock is 39 ft with different layer thickness as shown in Table 1.

Table 1—Basic parameters of the model (Bai et al., 2004)

Layer No. Thickness (ft) Porosity (%) Permeability (mD)


1 16 32 10000
2 82 30 2000
3 66 28 780
4 SPE-185716-MS

Two rock types were used; rock type 2 for layers 1 and 2, while rock type 1 for layer 3 as shown in
Figure 1. The reason of selecting two rock types, that we consider layers 1 & 2 is like a coduit, while
layer 3 is a matrix with very high permeability. Table 2 shows basic parameters of the reservoir, which
includes reservoir temperature and presuure, initial water saturation, and kv/kh. The reservoir is considered
as a thick heterogeneous reservoir as shown in Table 1, combined with high oil viscosity (i.e., 78 cP). Thus,
two important factors that lower oil recovery and yield a poor sweep efficiency are exist in this model.
Figure 2 shows a 3D visualization of the model while Table 3 shows the fluid properties. In this study, the
polymer and the crosslinker concentration are 1000 ppm and 20 ppm, respectively as shown in Table 3.
These concentrations were selected to form a weak gel that capable of penetrating deep into the formation.
The injection rate is selected to be 1070 STB/D for all runs, and was considered as a constraint. On the other
hand, the production constraint is selected to be bottomhole pressure. The model was run for 25 years (i.e.,
from 01/01/1995 to 01/01/2020) for all scenarios.

Figure 1—Relative-permeability curves for the two rock types in the model (Scott et al., 1987)

Table 2—Basic parameters of the reservoir

Reservoir temperature 149 °F


Initial reservoir pressure 2031 psi
kv/kh (base case) 0.01
Initial water saturation 0.22
Rock type 1 For layer 3
Rock type 2 For layers 1 & 2
SPE-185716-MS 5

Figure 2—3D visualization of the model

Table 3—Fluid properties used in the model

Water viscosity & density 0.6 cp & 62.4 lb/ft3


Oil viscosity & density 78.0 cp, 59.31 lb/ft3
Polymer viscosity & density 30.0 cp, 62.4 lb/ft3
Crosslinker viscosity & density 30.0 cp, 62.4 lb/ft3
Gel viscosity & density 30.0 cp, 62.4 lb/ft3
Polymer mW 10000 lb/lbmole
Crosslinker MW 206 lb/lbmole
Gel MW 10206 lb/lbmole
Polymer concentration 1000 ppm
Crosslinker concentration 20 ppm

Numerical Simulation Runs


Water Flooding Forecast
The initial water flooding case that is considered to be "base case" is associated with kv/kh equal to 0.01
and layer 1 permeability is 10000 mD. Six water flooding cases as shown in Table 4 were considered.
Theses scenarios will be use as a comparison with polymer flooding and gel treatment, however, base case
(i.e., WF2) will be use the most of the time throught this study. As can be seen from Table 4, the higher
the crossflow value, the higher the recovery factor. These recovery factors are corresponds to water cut
ranging from 93 - 95 % depends on crossflow value and layer 1 permeability. It is obvious that lowering the
heterogeneity (i.e., decreasing layer 1 permeabililty), from 10000 mD with kv/kh equal to 0.01 (i.e., WF2) to
5000 mD with kv/kh equal to 0.01 (i.e., WF5) is yielded a higher recovery factor as can be seen from Figure 3
which shows the recovery factor and the water cut for WF2 and WF5 runs, respectively. On the other hand,
increasing layer 1 permeability to 20000 mD with kv/kh equal to 0.01 (i.e., WF6) will lower the recovery
factor. The lower recovery factor in water flooding can be attributed to two main reasons, the higher the
viscosity of oil which will make unfavorable mobility ratio, and the heterogeneity of the reservoir. Thus, an
attempt to correct these reasons will eventually improve the reservoir performance.
6 SPE-185716-MS

Table 4—Water flooding results

Scheme kv/kh Perm of layer 1, mD Recovery factor, %

WF1 0.02 1 0000 28.24


WF2 (base case) 0.01 10000 27.80
WF3 0.005 1 0000 27.77
WF4 0.001 1 0000 27.38
WF5 0.01 5000 30.00
WF6 0.01 20000 25.98

Figure 3—Recovery factor and water cut comparison for WF2 (base case) and WF5

Polymer Flooding Forecast


15 runs of polymer flooding are performed, each 5 runs refers to different degradation time. Thus, runs
PFND1-PFND5 refers to polymer flooding without polymer degradation, runs PF4YD1-PF4YD5 refers to
polymer flooding with 4 years polymer degradation, and runs PF2YD1-PF2YD5 refers to polymer flooding
with 2 years polymer degradation as shown in Table 5. Different pore volumes of polymer were injected
which corresponds to different polymer injection period. In all of these runs, the polymer is injected when
the water cut in the model reached 85% and are compared with base case water flooding (i.e., WF2). Polymer
degradation time refers to decreasing the efficiency of polymer inside the reservoir, thus 4 years degradation
is better and gives higher results than 2 years degradation as can be noticed in Table 5. However, all polymer
flooding scenarios, whether with or without degradation gives higher recovery factor and good performance
in comparison with water flooding due to improving of water viscosity which yield favorable mobility ratio.
SPE-185716-MS 7

Table 5—Polymer flooding results

Scheme Degradation time, PV, injected Recovery factor, Incremental oil, Incremental Oil, Bbl of polymer/
years polymer % Bbls % Bbl of oil
WF2 27.80

PFND1 - 0.05 29.91 1 92570 7.6 3.05


PFND2 - 0.10 32.15 39831 3 15.6 2.95
PFND3 - 0.15 34.24 58951 0 23.1 2.99
PFND4 - 0.20 36.31 780036 30.6 3.01
PFND5 - 0.25 38.00 934506 36.7 3.14

PF4YD1 4 0.05 29.19 126729 5.0 4.64


PF4YD2 4 0.10 30.47 244585 9.6 4.8
PF4YD3 4 0.15 31.54 342761 13.4 5.14
PF4YD4 4 0.20 32.48 4291 17 16.8 5.48
PF4YD5 4 0.25 33.26 499780 19.6 5.88

PF2YD1 2 0.05 28.97 107381 4.2 5.47


PF2YD2 2 0.10 29.93 195041 7.7 6.02
PF2YD3 2 0.15 30.69 264664 10.4 6.66
PF2YD4 2 0.20 31.30 320269 12.6 7.34
PF2YD5 2 0.25 31.80 366661 14.4 8.01

It is obvious from the above table, that polymer flooding runs without polymer degradation (i.e., polymer
will keep its viscosity and residual resistance factor for entire simulation time) gives higher results than
4 years degradation and the least recovery factor obtained with 2 years polymer degradation. In addition,
degradation affect the injection process economically; more barrels of polymer are needed for each barrel
of oil when there is polymer degradation in comparison with no degradation runs. Moreover, the more
PV injected, the higher the recovery factor for the same degradation time. Incremental oil is calculated by
subtracting the cumulative oil production after polymer flooding run from the cumulative oil production
after waterflood run (in this case WF2 run). Figure 4 shows recovery factor and water cut comparison
between WF2 and PF4YD5.

Figure 4—Oil recovery factor and water cut comparison between WF2 (base case) and PF4YD5
8 SPE-185716-MS

Gel Treatments Simulation & Forecast


In this study, in order to inject gel deep into high permeability layers (thief zones) and divert the subsequent
water flow into lower permeability layers, a low concentration is selected; 1000 ppm and 20 ppm for the
polymer and the crosslinker, respectively. The weak gel is fromed by three ways (Shiyi et al., 2000):
1. The weak gel is formed prior to injection (preformed gel).
2. Crosslinker is injected following polymer, and the weak gel (or low concentration flowing gel, LCFG)
is formed while they met in the reservoir.
3. Polymer and crosslinker are injected at the same time and the gel is formed at a fixed time. The
concentration of polymer and crosslinker is quite low and the gelation time at laboratory could be
several days to a week. The gelation time and the gelation rate should be known precisely in order
to model this simulate this well in the model.
In this study, the reaction frequency factor (FREQFAC keyword in CMG-STARS), which is related to
the amount of time molecules will colloids each other to cause a reaction, that control the formation of
gel is 4.6512*106 day−1 (i.e., 3230 min−1). This value will give an enough time for the polymer and the
crosslinker to travel deep into the thief zones. The gel residual resistance factor considered in this study
is 20 for rock type 1 and 10 for rock type 2. Moreover, the maximum adsorption capacity for gel is equal
to the residual adsorption level, which implies that the adsorption is completely irreversible.Two methods
of gel injection is selected in this paper: bullhead and zone isolation. In bullhead method, gel is injected
into all three layers without isolation and these runs will be refered as "GA", while zone isolation method
means that gel is injected into only layers 1 & 2 (i.e., highest permeability layers) and these runs will be
refered as "OG". Thus, 30 runs were conducted using CMG-STARS, 15 "GA" runs and another 15 "OG"
runs. For each category, a subdivision of runs were done in order to study the effect of gel degradation and
the effect of gel injection method on the cumulative oil production and the recovery factor. In all of these
30 gel treatments, gel is injected when water cut in the model reached 85% and compared with WF2 (base
case) and the crossflow (i.e., kv/kh) is 0.01.
Bullhead Injection Runs. As with polymer flooding, 15 scenarios were conducted with different type of
gel degradation. The first five runs from GAND1-GAND5 without gel degradation and different PV of gel is
injected; injected gel PV increase from 0.01–0.1. Another 5 runs from GA4YD1-GA4YD5 were performed
with 4 years of gel degradation. The last 5 runs from GA2YD1-GA2YD5 were carried out considering 2
years of gel degradation. Gel degradation refers to the time that gel will continue to be active in the reservoir.
Table 6 shows the detailed results of these fifteen runs. It is apparent from that table, increasing PV of
injected gel will increase the recovery factor and ultimate production. In addition, for same PV injected up
to 0.05, the number of barrels of gel required to produce one barrel of oil is not changing and it is almsot the
same. As an example, for 0.03 PV of gel injected, with different gel degradation (i.e., GAND3, GA4YD3,
and GA2YD3 respectively), we have to inject almost the same amount of gel to produce one barrel of oil.
However, with 0.1 PV of injected gel, the number of barrel of gel required to produce one barrel of oil will
increase with decreasing the efficiency of gel inside the reservoir (i.e., gel degradation). Thus, 0.1 PV of
injected gel may be ideal but it could cost the project additional money especially if we encounter a shorter
degradation time (i.e., 2 years or less).
SPE-185716-MS 9

Table 6—Gel injection into all layers

Scheme Degradation time, PV, injected gel Recovery factor, Incremental oil, Incremental Oil, Bbl of gel / Bbl of
years % Bbls % oil
WF2 27.80

GAND1 - 0.01 30.18 21 7496 8.5 0.536


GAND2 - 0.02 31.57 345278 13.5 0.680
GAND3 - 0.03 33.05 481231 18.9 0.730
GAND4 - 0.05 34.98 657865 25.8 0.893
GAND5 - 0.10 39.08 1033530 40.6 1.137

GA4YD1 4 0.01 29.98 199700 7.8 0.584


GA4YD2 4 0.02 31.46 334969 13.1 0.670
GA4YD3 4 0.03 32.67 446022 17.5 0.789
GA4YD4 4 0.05 34.33 598225 23.5 0.982
GA4YD5 4 0.10 37.47 886409 34.8 1.325

GA2YD1 2 0.01 29.87 189138 7.4 0.617


GA2YD2 2 0.02 31.60 347621 13.6 0.674
GA2YD3 2 0.03 32.88 464961 18.2 0.757
GA2YD4 2 0.05 33.99 567579 22.3 1.035
GA2YD5 2 0.10 36.77 8221 35 32.3 1.429

Figure 5 shows a comparison between water flooding, polymer flooding, and gel treatment. The selected
runs is as follows: WF2 (base case) for water flooding, PF4YD2 which represent injecting 0.1 PV of polymer
with 4 years of polymer degradation, and GA4YD5 which represent injecting 0.1 PV of gel with 4 years
gel degradation. It is very clear the tremendous increase in oil recovery factor which is associated with a
big reduction in water cut in case of gel treatment (green curve) in comparison with polymer and water
flooding. The polymer and gel injection started at the same time when water cut reached 85%. Figure 6
shows bottomhole pressure in the injector, it is evident how much gel treatment increases water residual
resistance factor. The weak gel and polymer flooding can share different characteristic, but the gel treatment
effects last longer and affect more pronouncly than the effect of polymer flooding and can alleviate the
effect of heterogeneity.

Figure 5—Oil recovery factor and water cut comparison between WF2 (base case), PF4YD2, and GA4YD5
10 SPE-185716-MS

Figure 6—Injector BHP comparison between WF2 (base case), PF4YD2, and GA4YD5

Zone Isolation Runs. Another fifteen runs were considered to test the effect of injecting the gel into only
high permeability zones (i.e., layers 1 & 2). As done previously with the bullhead method, five of these
runs from OGND1- OGND5 without gel degradation, another five runs were performed with 4 years gel
degradation, and the last five runs with 2 years gel degradation. Table 7 whows the results of these runs.

Table 7—Gel injection into high permeability layers

Scheme Degradation time, PV, injected gel Recovery factor, Incremental oil, Incremental Oil, Bbl of gel / Bbl of
years % Bbls % oil
WF2 27.80

OGND1 - 0.01 31.22 31 3003 12.3 0.373


OGND2 - 0.02 33.21 495972 19.5 0.472
OGND3 - 0.03 34.73 635326 24.9 0.554
OGND4 - 0.05 36.71 81 6299 32.0 0.720
OGND5 - 0.10 39.11 1 035890 40.6 1.134

OG4YD1 4 0.01 30.94 287568 11.3 0.406


OG4YD2 4 0.02 32.98 474271 18.6 0.494
OG4YD3 4 0.03 34.16 582858 22.9 0.604
OG4YD4 4 0.05 35.85 738020 29.0 0.796
OG4YD5 4 0.10 38.62 991151 38.9 1.185

OG2YD1 2 0.01 30.53 249970 9.8 0.467


OG2YD2 2 0.02 32.58 437861 17.2 0.535
OG2YD3 2 0.03 33.86 555631 21.8 0.634
OG2YD4 2 0.05 35.38 694322 27.2 0.846
OG2YD5 2 0.10 37.72 908767 35.7 1.293

Same conclusions could be obtained from the above table as with the bullhead injection, however,
comparing two runs (for example GA4YD5 with OG4YD5) can tell us readily that injecting gel into only
high permeability layers without damaging the other layer gives better results than injecting gel into all
three layers as can be seen in Figure 7 below.
SPE-185716-MS 11

Figure 7—Oil recovery factor and water cut comparison between WF2 (base case), GA4YD5, and OG4YD5

The above results (i.e., both bulkhead and zonal isolation), revealed the importance of protecting
hydrocarbon zones in wells with radial flow, in order to achieve good gel placement (Seright, 1996). Seright
added that "extreme reservoir heterogeneity does not eliminate the need to protect hydrocarbon productive
zones during gelant placement in unfractured injection wells (where flow is radial)". Therefore, zonal
isolation is more effective and gives higher recovery factor than bullhead injection.
Effect of Changing Layer 1 Permeability. To test to what extent could changing layer 1 permeability affect
the results, another set of runs were performed which depict increasing and decreasing the heterogeneity of
the model. Four values of layer 1 permeability were selected: 3000, 5000, 10000, and 20000 mD and keeping
other variables like the crossflow at 0.01 and the permeability of the other two layers (i.e., 2000 & 780 mD).
These runs are done with polymer flooding and gel treatment with 0.1 PV injected, and compared with it
suitable water flooding runs. Similarly, both polymer flooding and gel treatment started when water cut in
the model reached 85% as before. Names of these runs are followed by the value of layer 1 permeability
value, for example PF3000 and GA3000 is for polymer flooding and gel treatment when layer 1 permeability
is 3000 mD, respectively. In addition, gel is injected into all three layers with 2 years degradation for both
polymer and gel. Table 8 shows the results of these runs, which demonstrate the profound effect of increasing
layer 1 permeability from 3000 mD to 20000 mD. The more heterogeneous the reservoir, the lower the
recovery factor for both polymer flooding and gel treatment. However, increasing the heterogeneity of the
model will decrease the number of polymer required to produce one barrel of oil but will increase the number
of barrel of gel required to produce one barrel of oil.
12 SPE-185716-MS

Table 8—Effect of formation heterogeneity

Scheme Layer1 PV, injected Recovery factor, Incremental oil, Incremental Oil, Bbl of polymer or
permeability, mD polymer or gel % Bbls % gel / Bbl of oil
WF1 3000 31.13
PF3000 3000 0.1 32.52 127015 4.5 9.25
GA3000 3000 0.1 39.61 777504 27.2 1.511

WF2 5000 29.94


PF5000 5000 0.1 31.48 140800 5.1 8.345
GA5000 5000 0.1 38.34 769367 28 1.527

WF3 10000 27.77


PF10000 10000 0.1 29.43 151746 6.0 7.742
GA10000 10000 0.1 35.38 697776 27.4 1.684

WF4 20000 25.38


PF20000 20000 0.1 27.35 183796 7.9 6.392
GA20000 20000 0.1 31.86 597435 25.7 1.967

As can be seen from the above table, the effectiveness of using gel in remedying the heterogeneity of
the reservoir, which can clearly noticed from the high incremental oil associated with gel treatment for all
four cases.
Effect of Crossflow (Kv/Kh). Crossflow is a common problem in almost all reservoirs, which can occur
when there is no barrier between the adjacent beds or layers. In-depth gel placement should consider this
problem, otherwise the injected water after the treatment will bypass again into high permeability zone
which make the treatment uneffective. Four vaules of kv/kh were selected; 0.02, 0.01, 0.005, and 0.001. In
this section, the gel is injected into all three layers (GA runs) and the name of the run followed by the value
of the crossflow (for example GA0.02), all other parameters kept constant. In addition, the gel injection
started when water cut in the model reached 85%, with 4 years degradation time. Finally, the gel treatment
runs compared with its water flooding counterpart as shown in Table 9.

Table 9—Effect of crossflow

Scheme kv/kh PV, injected Recovery factor, Incremental oil, Incremental Oil, Bbl of gel / Bbl of
polymer or gel % Bbls % oil
WF1 0.02 28.24
GA0.02 0.02 0.1 36.98 801 81 9 31.0 1.465

WF2 0.01 27.80


GA0.01 0.01 0.1 36.74 819487 32.2 1.434

WF3 0.005 27.77


GA0.005 0.005 0.1 36.50 800150 31.4 1.468

WF4 0.001 27.38


GA0.001 0.001 0.1 36.32 819298 32.6 1.434

Theoretically, the higher the crossflow value, the less effectiveness the gel treatment. However, in this
study, no specific relationship between the crossflow and the gel treatment could be drawn. Possible reason
for this discrepancy is that gel is injected into all three layers without isolation. Kim, 1995 noticed that
the performance of in-depth gel treatment was much better than the near-wellbore gel treatment when
crossflow was dominant. We could see clearly from Table 9, the recovery factor, the barrels of incremental
SPE-185716-MS 13

oil, the percentage of the incremental oil are almost the same with small differences. Figure 8 shows the
water residual resistance factor at 01/01/2020 (i.e., end of the simulation time) for layers 1, 2, & 3 for run
GA0.005. It is clear from that figure below how deep the gel is invading layer 1 (permeability 10000 mD)
in comparison with layer 2 (permeability 2000 mD) and layer 3 (permeability 780 mD). The water residual
resistance factor is higher in layer 1 than layer 2, whereas there is no change in layer 3.

Figure 8—Water residual resistance factor for layers 1,2, & 3, run GA0.005

Effect of Water Cut at Time of Gel Injection. In order to investigate the best time to start the gel injection,
three values of water cut were selected. The selected values are shown in Table 10, which also shows the
results of these runs. Gel treatments were compared with WF2 (base case), the gel being injected in all
layers is 0.02 PV without degradation.

Table 10—Effect of crossflow

Scheme Water cut, % PV, injected Recovery factor, Incremental oil, Incremental Oil, Bbl of gel / Bbl of
polymer or gel % Bbls % oil
WF2 27.80
GA65%WC 65 0.02 32.61 440732 17.3 0.534
GA80%WC 80 0.02 32.28 41 0532 16.1 0.571
GA90%WC 90 0.02 30.68 263979 10.4 0.888
14 SPE-185716-MS

The table above revealed that, for severe heterogeneous reservoir (as the case in this study), the sooner the
treatment, the better the results and the lower volume of gel needed to achieve higher recovery. However,
it can be seen that not a big difference is noticed between 65% and 80% water cut runs in term of recovery
factor. As a result and based on the above findings, we can conclude that the best time to start gel treatment
when the water cut in the reservoir reached 80%. Figure 9 shows a comparison between oil recovery, water
cut, and cumulative oil production for these three runs, which compared with WF2 run.

Figure 9—Oil recovery factor, water cut, and cumulative oil


production comparison between WF2 (base case), GA65WC, GA80WC

Effect of Injecting Polymer and Gel Together. Six different runs were conducted to show the importance
of injecting polymer then gel or gel then polymer. These runs were compared with WF2 (base case), and
for different degradation time. The gel is injected into all three layers (GA runs) and into only layers 1 and
2 (OG runs) without degradation and with 4 years degradation. Whereas 2 years polymer degradation were
considered for polymer flooding. Small PV of gel and polymer were selected: 0.01 PV and 0.09 PV for gel
and polymer, respectively. Table 11 shows the result of these six runs with a comparison with WF2, and for
all six runs the treatment started when water cut in the model reached 85%.

Table 11—Effect of together injection of polymer & gel

Scheme Degradation time, years Injected PV Recovery Incremental Incremental Bbl of


factor, % oil, Bbls Oil, % polymer or
gel / Bbl of oil
Gel Polymer Gel Polymer
WF2 27.80

GAP-1 4 2 0.01 0.09 30.46 2431 48 9.5 4.829


GAP-2 - 2 0.01 0.09 32.70 449096 17.6 2.615

PGA-1 4 2 0.01 0.09 30.76 271121 10.6 4.331


PGA-2 - 2 0.01 0.09 31.22 31 3525 12.3 3.745

OGP-1 4 2 0.01 0.09 30.83 277072 10.9 4.238


OGP-2 - 2 0.01 0.09 32.57 437250 17.2 2.685

Table 11 revealed an interesting results regarding combination of polymer flooding and gel treatment. The
results showed that, for 4 years gel degradation and 2 years polymer degradation, polymer flooding followed
be gel treatment (PGA-1 run) yield better result than gel treatment followed by polymer flooding (GAP-1
SPE-185716-MS 15

run). In addition, for no gel degradation and 2 years polymer degradation, gel treatment then polymer
flooding ( GAP-2 run) yield better result than polymer flooding then gel treatment (PGA-2 run). Moreover,
For 4 years gel degradation and 2 years polymer degradation, gel injected into layers 1 & 2 (OGP-1 run)
resulted in a subtle difference in comparison with gel injected in all three layers (GAP-1 run). Finally, for
no gel degradation and 2 years polymer degradation, gel injected in all three layers (GAP-1 run) showed a
good performance than gel injected into layers 1 & 2 (OGP-2 run). Figure 10 shows a comparison of some
of these runs, which is between WF2, GAP-2, and PGA-2. As mentioned previously, GAP-2 is better than
PGA-2 in terms of oil recovery factor and water cut reduction.

Figure 10—Oil recovery factor and water cut comparison between WF2 (base case), GAP-2, and PGA-2

Sensitivity Analyses Study


Sensitivity Analysis is used to determine how sensitive an objective functions is to different parameters
and their ranges in values. In another word, sensitivity analysis allows the identification of the parameter
or set of parameters that have greatest influence on the model output (Zhang et al., 2015). CMOST is
CMG's history matching, optimization, sensivity analysis, and uncertainty assessment tool. It can launch
multiple simulation jobs to multiple computers using different input parameter values. To make the best use
of CMOST, the user should have a good understanding of the reservoir model they are working with. For
Sensitivity Analysis and Uncertainty Assessment using classical experimental design or Latin hypercube
design, a response surface methodology is applied. Response surface methodology (RSM) explores the
relationships between input variables (parameters) and responses (objective functions). The main idea of
RSM is to use a set of designed experiments to build a proxy (approximation) model to represent the original
complicated simulation model. The most common proxy models take either a linear form or quadratic form
of a polynomial function. After a proxy model is built, Tornado plots displaying a sequence of parameter
estimates can be used to assess the sensitivity of parameters (CMOST user's guide version 2010, CMG).
In this section, CMOST will be used as a sensitivity analysis tool for determining the overall variation of
simulation results under different parameters values and/or which parameters would have the greatest effect
on simulation results. However, knowing how to create base dataset and CMOST master dataset is out of
the scope of this paper, and one can refer to CMOST user's guide manual by CMG. By default, CMOST
selects the unit system that has been used in the dataset.
16 SPE-185716-MS

Parameters
In this study, the parameters are recognized to be of two categories, which can affect the objective functions:

• Reservoir parameters: which represent the parameters that we do not have physical control over
on location, such as reservoir and oil properties as shown in Table 12.
• Operating parameters: which represent the parameters that we do have physical control over on
location, which include, but not limited to, polymer properties, injection rate, and injection pressure
as shown in Table 13

Table 12—Reservoir parameters ranges used in CMOST

Parameters Default value Lower limit Upper limit


Perm of layer 1, mD 10000 3000 20000
Perm of layer 2, mD 2000 500 2500
Perm of layer 3, mD 780 250 1000
kv/kh 0.01 0.0005 0.2
Oil viscosity, cP 78 20 97.5
RRFT1 20 10 100
RRFT2 10 5 50
Thick. of layer 1, ft 16 10 20
Thick. of layer 2, ft 82 50 150
Thick. of layer 3, ft 66 35 125

Table 13—Operating parameters ranges used in CMOST

Parameters Default value Lower limit Upper limit


Injection Pressure, psi 2000 1500 2500
Injection rate, bbl/day 1070 500 3000
Polymer mole fraction 1.8033e-6 1.35248e-6 2.25413e-6
Polymer MW, lb/lbmole 10000 7500 12500
Polymer viscosity, cP 30 22.5 37.5

Ten reservoir (in-situ) parameters and four operating parameters were selected separately, which represent
the most influential parameters on the simulation runs. Normally, data ranges (i.e., upper and lower limit)
are automatically set to ±25% of the default value for each parameter by CMOST, but can be adjusted if
necessary. The created parameters can be selected directly from the input data file for specific run, or can
be selected manually and entered the default, the lower, and the upper values.

Objective Functions
The objective functions determine which results CMOST will analyze. Two sets of objective functions were
selected, one group with reservoir parameters as shown in Table 14 and the other group with operating
parameters as shown in Table 15.
SPE-185716-MS 17

Table 14—Objective functions used in CMOST with reservoir parametrs

Name Origin type Origin name Property


Oil_Prod_Cum_SCTR Sectors Entire field Oil Prod Cum SCTR
Oil_Rec_Fact_SCTR Sectors Entire field Oil Recovery Factor SCTR
Cum_Oil_Prod Wells Producer1 Cumulative Oil SC
Cum_Wtr_Prod Wells Producer1 Cumulative Water SC
BHP Inj Wells Injector Well Bottom-hole Pressure
Oil_Vol_SCTR Sectors Entire field Oil Volume SC SCTR

Table 15—Objective functions used in CMOST with operating parametrs

Name Origin type Origin name Property


Cum_Oil_Prod Sectors Entire field Oil Prod Cum SCTR
Oil_Rec_Fact_SCTR Sectors Entire field Oil Recovery Factor SCTR
Water_Cut Sectors Entire field Water Cut SCTR-%

Results and Anlayses of Reservoir Parameters


In this section we will overview the effect of selected parameters on the oil recovery factor as an objective
function. This section will includes:

• Proxy analysis- Model QC.

• Proxy analysis- Effect Estimate (Tornado Chart).

• Sobol analysis.

Proxy Analysis- Model OC. In Figure 11, the true simulation results are plotted against the proxy estimated
results. The distance from the 45 degree line indicates how much the proxy results deviate from the actual
simulation results. In a perfect match, all points should fall on the 45 degree line. The points that fall on
the 45 degree line are those that are perfectly predicted. The points that are far away from the 45 degree
line are outliners (CMOST user's guide version 2010, CMG). To better understand the matching between
the simulated and proxy results, we should navigate to the statistics tab in CMOST. Check the R-Square, R-
Square Adjusted, and R-Square Prediction. If they are all above 0.5, it indicates a sufficient match. In our
study, these values are 0.9397, 0.9132, and 0.8493 which indicate a good match. In addition, if prob>F is
≤ 0.05, it shows a good agreement between the simulated and proxy results. In our study, this value match
this criteria (less than 0.00001).
18 SPE-185716-MS

Figure 11—Simulated oil recovery factor vs. proxy predicted oil recovery factor

Proxy Analysis- Effect Estimate (Tornado Chart).. In this study, a second-degree (Quadratic) polynomial
model is used. For this type of proxy model, parameters interaction effects (cross terms xixj) and quadratic
effects (x2j) can be extracted in addition to linear effects (xj).

Here a0 is the intercept; a1, a2,...., ak are the coefficients of linear terms; ajj are the coefficients of quadratic
terms; and aij are the coefficients of cross (interaction) terms (CMOST user's guide version 2010, CMG).
Tornado plot gives a visual representation of the parameter sensitivity. Parameters with high negative
or high positive values on the plot can be considered to have more effects on the objective functions. The
maximum and minimum bars represent the maximum and minimum objective function values from all
experiments as shown in Figure 12. Note that KvKh is crossflow and not the product of horizontal and
vertical permeability (i.e., kv/kh), because using division symbol in creating parameter in CMOST is not
acceptable.
SPE-185716-MS 19

Figure 12—Tornado chart showing the effect of in-situ parameters on recovery factor

The analysis of the Tornado plot suggests the following conclusions regarding the sensitivities of these
parameters on oil recovery factor:
1. Permeability of layers 2 & 3, crossflow, & RRFT2 are the main positive linear effects.
2. Linear effect of Oil_Vis is negatively impact oil recovery factor.
3. The non-linear (quadratic) effect of Oil_Vis*Oil_Vis is also important.
4. Another linear effect, such as the permeability of layer 1, and the thicknesses of all three layers affects
negatively oil recovery factor.
5. The quadratic effect of crossflow (kv/kh) has a negative effect on the designated objective function.
For example, changing layer 2 permeability from 500–2500 mD will increase the recovery factor
by 7.793%, because the heterogeneity between layer 1 and layer 2 will decrease. On the other hand,
considereing a permeability range for the thief zone (layer 1) from 3000–20000 mD will have a negative
impact on the recovery factor (i.e., will decrease oil recovery factor by 4.826%). Therefore, the positive
values will have a positive impact, while the negative values will have a negative impact.
Proxy Analysis- Sobol Anlysis. The Sobol method (Sobol 1993), is a type of variance-based sensitivity
analysis approach. The main idea of variance-based method is to quantify the amount of variance that each
input factor Xi contributes to the unconditional variance of output V(Y). In this way, the corresponding
impact of the input data to output model response are quantified from the analysis of these variances.
The result of weighting was listed in Figure 13 to emphasize the contributions of different variables to oil
recovery factor. The most influential parameter is oil viscosity with 48% impact on recovery factor, followed
by the permeability of layer 2 with 15% effect. The least effect is thickness of layer 1 then RRFT.
20 SPE-185716-MS

Figure 13—Sobol result analysis, oil recovery factor

Results and Anlayses of Operating Parameters


Proxy Analysis- Effect Estimate (Tornado Chart). Although we could control the operating parameters,
however, the effect of these parametres is not as strong as with reservoir parametres and that evident from
Figure 14. In this case, the minimum and maximum recovery factor from all experiments are different from
the previous case. As can be seen from Figure 13, changing the polymer concentration, injection rate, and
injection pressure will have a small effect on the results. Therefore, changing polymer concentration and
injection rate will increase the recovery factor, while changing the injection pressure will have a small
negative impact on the recovery factor.

Figure 14—Tornado chart showing the effect of operating parameters on recovery factor

Detecting the Optimum Parameters


In the sensitivity analyses study of the reservoir parameters, more than 40 experiments were run by
CMOST. In each one of these experiment, an optimum parameter values were detected which yielded the
highest results. Table 16 shows the base case values and the optimum values that could give a 49.06% oil
recovery factor. This study could be beneficial when we want to compare different reservoirs with different
characteristics.
SPE-185716-MS 21

Table 16—Comparison between base case values and optimum values obtained by CMOST

Parameters Base case value Optimum value obtained by


CMOST
Perm of layer 1, mD 10000 8100
Perm of layer 2, mD 2000 2100
Perm of layer 3, mD 780 550
kv/kh 0.01 0.06035
Oil viscosity, cP 78 20
RRFT1 20 20
RRFT2 10 18
Layer 1 thickness, ft 16 18
Layer 2 thickness, ft 82 60
Layer 3 thickness, ft 66 35

Conclusions
1. For both polymer flooding and gel treatment, increasing pore volume injected, has a direct effect on
recovery factor and incremental oil recovery. However, the longer the degradation time, the higher
the recovery factor for the same PV injected.
2. For polymer flooding, when there is no degradation, increasing injected PV has a little effect on
number of barrels of polymer required to produce one barrel of oil. However, for 2 & 4 years polymer
degradation, increasing injected PVs yield an increase of number of barrels of polymer required to
produce one barrel of oil.
3. For gel treatment, whether there is no degradation or there is a degradation, increasing injected PVs
will increase barrels of gel required to produce one barrel of oil.
4. For gel treatment, injecting gel into high permeability layers only (i.e., zone isolation method) have
better results than injecting gel into all three layers (i.e., bullhead method).
5. For gel treatment, increasing the permeability of high permeability layer (i.e., layer 1) will decrease
the recovery factor using both polymer flooding and gel treatment.
6. For the range of crossflow taken into account in this study (i.e., kv/kh= 0.001, 0.005, 0.01, & 0.02),
there is no substantial difference in oil recovery for gel treatment. Therefore, a sensitivity analysis run
is performed using CMOST software to test wide range of kv/kh values.
7. This study revealed that the optimum time to start gel treatment is when the water cut reach 80%,
because prior to that date, high volume of oil is still available in the high permeability layer. Therefore,
it is not desirable to start gel treatment earlier.
8. With 4 years gel degradation and 2 years polymer degradation, polymer flooding then gel treatment
(i.e., PGA-1 run) yielded better result than gel treatment then polymer flooding (i.e., GAP-1 run).
Note that the treatment started when water cut reached 85% for all cases.
9. Without gel degradation and 2 years polymer degradation, gel treatment then polymer flooding (i.e.,
GAP-2 run) yielded better result than polymer flooding then gel treatment (i.e., PGA-2 run).
10. With 4 years gel degradation and 2 years polymer degradation, gel injected into layers 1 & 2 (i.e.,
OGP-1 run) resulted in slight difference in factor than gel injected into all layers (i.e., GAP-1 run).
11. Without gel degradation and 2 years polymer degradation, gel injected into all layers (i.e., GAP-2 run)
showed a good performance than gel injected into layers 1 & 2 (i.e., OGP-2 run).
12. Response surface methodology is applied for sensitivity analysis to explore the relationships between
input variables (parameters) and responses (objective functions).
22 SPE-185716-MS

13. Two catergories of parameters could be recognized, reservoir parameters and operating parameters.
We do not have any control on reservoir parameters, however, it has more influence on the objective
functions than the parameters than we do have control (i.e., operating parameters)
14. Thickness and permeability of layers 2 & 3, crossflow, and RRFT2 has a positive impact on cumulative
oil production. Whereas, permeability of layer 1, & oil viscosity have a negative impact on cumulative
oil production.
15. Increasing polymer concentration not necessarily will increase the recovery factor, on the contrary, it
may have a negative impact on the project by increasing the expenses.

Acknowledgments
The author would like to acknowledge the Ministry of Oil of Iraq & the ExxonMobile Iraq Training
Technology & Scholars Program (ExxonMobile Iraq) for granting him a PhD scholarship and financial
support.

Nomenclature
ADMAX maximum adsorption capacity, lbmole/ft3
ADRT residual adsorption level, lbmole/ft3
G gel treatment
kv/kh ratio of vertical to horizontal permeability
kl layer 1 permeability, mD
k2 layer 2 permeability, mD
k3 layer 3 permeability, mD
Perm permeability
PF polymer flooding
Polcomp polymer mole fraction
PV pore volume
RRFT1 residual resistance factor for the adsorbing component in rock type 1
RRFT2 residual resistance factor for the adsorbing component in rock type 2
RSM response surface methodology
SCTR sector
WF water flooding

References
Bai, B., Wang, Q., Du, Y., & Liu, Y.Z. 2004. Factors Affecting In-Depth Gel Treatment for Reservoirs with Thick
Heterogeneous Oil Layers. Presented at the 55th Annual Technical Meeting, Clagary, Alberta, Canada, 8–10 June,
Paper 2004-140. https://doi.org/10.2118/2004-140.
Computer Modeling Group 1987-2010. User's guide CMOST Studio Version 2010. Calgary, Alberta, Canada.
Fletcher, A.J.P., Flew, S., Forsdyke, I.N., Morgan, J.C., Rogers, C., & Suttles, D. 1992. Deep diverting gels for very
cost-effective waterflood control. Journal of Petroleum Science and Engineering, Volume 7, Issues 1–2, April, Pages
33–43. http://dx.doi.org/10.1016/0920-4105(92)90006-M.
Garver, F.J., Sharma, M.M., & Pope, G.A. 1989. The Competition for Chromium Between Xanthan Biopolymer and
Resident Clays in Sandstones. Presented at the SPE Technical Conference and Exhibition, San Antonio, 8–11 October.
SPE-19632-MS. http://dx.doi.org/10.2118/SPE-19632-MS.
Ghahfarokhi, A.J., Kleppe, J., & Torsaeter, O. 2016. Simulation Study of Application of a Water Diverting Gel in Enhanced
Oil Recovery. Presented at EAGE Conference and Exhibition, Vienna, Austria, 30 May-2 June. SPE-180190-MS.
http://dx.doi.org/10.2118/180190-MS.
Herbas, J., Kumar, S., Moreno, R., Romero, M.F., & Avendano, H. 2004. Reservoirs Simulations of Gel Treatments
to Control Water Production, Improve the Sweep Efficiency and the Conformance Factor in Eastern Venezulean
HPHT Fractured Reservoirs. Presented at SPE International Petroleum Conference, Puebla, Mexico, 8–9 November.
SPE-92003-MS. http://dx.doi.org/10.2118/92003-MS.
SPE-185716-MS 23

Hoefner, M.L., Seetharam, R.V., Shu, P., & Phelps, C.H. 1992. Selective penetration of biopolymer profile-control gels:
Experiment and model. Journal of Petroleum Science and Engineering, Volume 7, Issues 1–2, April, Pages 53–66.
http://dx.doi.org/10.1016/0920-4105(92)90008-Q.
Kim, H.S., 1995. Simulation Study of Gel Conformance Treatments. Ph.D. Dissertation, the University of Texas at Austin.
Scott, T., Roberts, L.J., Sharpe, S.R., Clifford, P.J., & Sorbie, K.S. 1987. In-situ Gel Calculations in Complex Reservoir
Systems using a New Chemical Flood Simulator. SPE Reservoir Engineering Vol. 2, Issue 4, Pages 634–646.
SPE-14234-PA. http://dx.doi.org/10.2118/14234-PA.
Seright, R.S., 1996. Improved Methods For Water Shutoff. Semi-Annual Technical Progress Report, Submitted to BDM-
Oklahoma, US DOE, October 1995-September 1996. http://baervan.nmt.edu/groups/res-sweep/media/pdf/annual-
reports/A1996.pdf
Shiyi, Y., Dong, H., Qiang, W., & Hua, Y. 2000. Numerical Simulation on Weak Gel Injection. Presented at the SPE
Asia Pacific Oil and Gas Conference and Exhibition, Brisbane, Australia, 16-18 October. SPE-64291-MS. http://
dx.doi.org/10.2118/64291-MS.
Sobol, I.M. 1990. Sensitivity analysis for non-linear mathematical models. Presented at Mathematical modeling and
computational experiment, Edition 1, Pages 407–414, New York, John Wiley & Sons. http://www.worldcat.org/title/
mathematical-modeling-and-computational-experiment/oclc/25442694?referer=di&ht=edition.
Sorbie, K.S., & Seright, R.S. 1992. Gel Placement in Heterogeneous Systems With Crossflow. Presented at
SPE/DOE Eighth Symposium on Enhanced Oil Recovery, Tulsa, 22–24 April. SPE/DOE 24192-MS. http://
dx.doi.org/10.2118/24192-MS.
Stavland, A., Kvanvik, B.A., & Lohne, A. 1994. Simulation Model for Pridicting Placement of Gels. Presented
at Annual Technical Conference and Exhibition, New Orleans, 25–26 September. SPE-28600-MS. http://
dx.doi.org/10.2118/28600-MS.
Tsau, J.S., Hill, A.D., & Sepehrnoori, K. 1985. Modeling of Permeability Reducing Vertical Conformance Treatments.
Presented at International Symposium on Oilfield and Geothermal Chemistry, Phoenix, 9–11 April. SPE-13569-MS.
http://dx.doi.org/10.2118/13569-MS.
Zhang, X-Y, Trame, M.N., Lesko, L.J., Schmidt, S. 2015 Sobol Sensitivity Analysis: A Tool to Guide the Development
and Evaluation of Systems Pharmacology Models. Presented in CPT: Pharmacometrics & Systems Pharmacology,
Volume 4, Issue 2, Feb., Pages 69–79. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5006244/.

S-ar putea să vă placă și