SlideShare a Scribd company logo
1 of 9
Download to read offline
Estimation of underground interwell connectivity: A data-driven
technology
Ehsan Jafari Dastgerdia
, Ali Shabanib,
*, Davood Zivarc,
*, Hamid Reza Jahangirid
a
Department of Petroleum Engineering, Khazar University, Baku, Azerbaijan
b
Department of Chemical and Petroleum Engineering, Sharif University of Technology, Tehran, Iran
c
Special Core Analysis Lab, Petro Makhzan Kav (PMK), Tehran, Iran
d
Department of Chemical and Petroleum Engineering, Iran University of Science and Technology, Tehran, Iran
A R T I C L E I N F O
Article History:
Received 21 August 2020
Revised 17 October 2020
Accepted 9 November 2020
Available online xxx
A B S T R A C T
Water injection into petroleum reservoirs is widely performed around the world for enhancing oil recovery.
Understanding the underground fluid path is an important factor in improving reservoir performance under
waterflooding operation. This may be used to optimize subsequent oil recovery by changing injection pat-
terns, assignment of well priorities in operations, recompletion of wells, targeting infill drilling, and reduce
the need for expensive surveillance activities. Most of the hydrocarbon reservoirs are equipped with sensors
that measure the flow rate, pressure, and temperature in the wellbores continuously. Valuable and useful
information about the interwell connections can be obtained from the measured data. In the present paper, a
novel data-driven approach is developed that contributes to estimating the underground interwell connec-
tions through analyzing and processing the injection and production data of a petroleum reservoir. This novel
data-driven technology is named Detection of Events (DoE) and combines with the Capacitance Resistance
Model (CRM) to quantify the connections between injectors and producers. The proposed approach has been
applied to both synthetic and field cases. The results of the method show a good agreement between CRM
and DoE for the field case (75%), and present a good insight into the better understanding of the underground
connections.
© 2020 Published by Elsevier B.V. on behalf of Taiwan Institute of Chemical Engineers.
Keywords:
Data-driven approach
Capacitance resistance model (CRM)
Detection of events (DoE)
Interwell connections
1. Introduction
Data acquisition systems are used in the petroleum industry to
monitor the field operation continuously. The gathered data can be
used for the well-log prediction, risk analysis, reservoir management,
drilling engineering, and uncertainty analysis through applying dif-
ferent data-driven methods [1,3,4,24,25]. Injection of water into the
oil reservoirs for enhancing oil recovery is widely used around the
world since the early days of the petroleum industry. Increasing the
oil recovery by waterflooding can prolong the life of a petroleum res-
ervoir. Waterflooding needs detailed management and continuous
monitoring. During a waterflooding operation, understanding the
underground fluid paths is necessary information for the efficient oil
displacement, determination of infill drilling, and injection-produc-
tion pattern. This information depends on the quality of the connec-
tion between wells of the reservoir. There are different methods to
quantify the underground connection among the wells including tra-
ditional simulators, tracers, and data-driven approaches
[2,8,12,15,17,30,31,33]. Traditional simulators have been used by the
petroleum companies as a standard reservoir evaluation method.
However, due to the limitations of the modeling techniques, reservoir
simulators might not fully represent the true reservoir performance.
Besides, maintaining a truly representative full reservoir simulation
model for the routine and frequent analysis of waterflood perfor-
mance prediction is a challenging process. Nowadays, using tracers is
not common in the petroleum industry, because it takes a long time
for the tracers to breakthrough, which makes the procedure time and
money consuming.
Over the past few years, several data-driven approaches based on
available reservoir data have been introduced and developed for pre-
diction of the production rates and the connections between wells
[5,23,37]. Some researchers used a data-driven interwell numerical
simulation model (INSIM) for modeling waterflooding [9,39]. Other
researchers used spearman rank correlations to find the connectivi-
ties between injectors and producers of a field [10,27,28,32].
Recently, capacitance resistance models were introduced for water-
flooding studies and were used widely by companies and researchers
for estimating the connections between the wells of a reservoir
[6,7,16,1821,23,29,36,38]. This intuitive and numerically fast tech-
nique utilizes only the injection and production data. It exhibits the
* Corresponding authors.
E-mail addresses: alishabani.sharif@gmail.com (A. Shabani), Davood_zivar@yahoo.
com (D. Zivar).
https://doi.org/10.1016/j.jtice.2020.11.008
1876-1070/© 2020 Published by Elsevier B.V. on behalf of Taiwan Institute of Chemical Engineers.
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
Journal of the Taiwan Institute of Chemical Engineers 000 (2020) 19
Contents lists available at ScienceDirect
Journal of the Taiwan Institute of Chemical Engineers
journal homepage: www.elsevier.com/locate/jtice
strengths of the injector-producer connections without geological
information [35]. Izgec and Kabir [11] introduced the CRM as a reli-
able method that its results have a good agreement with traditional
reservoir simulations [11]. Moreno [22] and Zhao et al. [39] tried to
find the connectivity in multi-layer systems [22,39]. Shabani et al.
[31] used CRM for understanding the injectors’ efficiency and analyz-
ing the interwell connections. The neural network also can be used
for prediction of the production rate and the interwell connections
[5,26,34]. Artun [5] used weight factors achieved from the neural net-
work to define the quality of connections between injectors and pro-
ducers of a synthetic reservoir [5]. As the neural networks are not the
physics based methods, they cannot estimate the interwell connec-
tions as efficient as CRMs [13,14,26]. Jahangiri et al. [12] developed
and applied Top-Down WaterFlood (TDWF) diagnostics and optimi-
zation method for the North Sea and used it as a “virtual” interwell
tracer. As demonstrated for the North Sea field, the efficiency of the
most recent injector in the field will be understood without the need
to await the results of a tracer program.
In this paper, we present a novel approach named Detection of
Events (DoE), based on relating the injection and production data of a
real reservoir. The proposed approach complements CRM by data
mining techniques to quantify and validate the CRM injector-pro-
ducer connection estimates. This has implications when it comes to
prioritizing water injectors when optimizing injection under various
water constraints (e.g. individual well, well pad and/or total vol-
umes). The proposed method is tested for the synthetic and real res-
ervoirs, then the results are compared with the results of CRM. The
structure of the paper is as follows: Section two presents the mathe-
matical methodology. Results and discussions of the proposed
method are presented in section three, and section four presents the
conclusions.
2. Mathematical methodology
Understanding the underground interwell connections presents
valuable information about the performance of a waterflooding
operation. For estimating the interwell connections in the present
study, both analytical capacitance resistance model (CRM) and detec-
tion of events (DoE) approach are used. These two complementary
data-driven methods, which do not require any geological or
dynamic reservoir models, quantify the underground connections
between the injectors and the producers.
2.1. Capacitance resistance model (CRM)
A detailed methodology of the capacitance resistance model is
presented in this section. Capacitance resistance models are used in
most engineering fields including electrical, chemical, petroleum, etc.
In the petroleum industry, CRM enables a reservoir engineer to esti-
mate production rates of producers, and understand the under-
ground connection between each pair of injector-producer through
history matching of the production data [22,29].
By assuming different control volumes, different forms of the CRM
can be developed, including CRMT, CRMP, and CRMIP [29]. Among
different forms of CRM, the CRMP (CRM Producer) can be used for a
system of multi-well. It can estimate the production rate of producers
and the quality of the underground connection between each pair of
injector-producer. By assuming a constant productivity index (J), and
a step variation of the injection rate and production pressure in the
time interval of Dtk, the production rate of each producer can be esti-
mated by the underlying equation [29]:
qjðtkÞ ¼ qjðtk1Þe

Dtk
tj þ 1e

Dtk
tj
  X
Ninj
i¼1
fijIi
k
Jjtj
DPk
w;j
Dtk
 #
ð1Þ
In this equation, qj, tj, and Jj are the production rate, the time con-
stant, and the productivity index of the jth producer respectively. fij is
the connection factor between the ith injector and the jth producer, Ii
is the injection rate of the ith injector, and Njnj is the number of active
injectors. The connection factor (fij), time constant (tj), productivity
index (Jj), and the primary production rate of the producer (q0) are
estimated using the optimization process. It is obvious that the sum
of the connection factors between an injector and other producers
must be less than one as presented below (Npro is the number of
active producers in the reservoir):
X
Npro
j¼1
fij1 ð2Þ
The calculated connection factors (fij) create a matrix and are used
for identification of the quality of the connection between each pair
of injector-producer. The high value of the connection factor means
that there is a stronger underground connection between the injector
and producer. According to the magnitude of the connection factor
between each pair of injector-producer, the quality of the connection
is defined as presented in Table 1.
2.2. Detection of events (DoE)
DoE is a quantitative method for identifying the flow path
between injectors and producers of a reservoir. This method is based
on finding relations between injection and production events. A flow
rate or a pressure event is defined as the certain changes in that flow
rate or pressure in a certain period of time. For determination of the
Nomenclature
AP Agreement parameter
fij Fraction of the rate of injector i that is towards to the
producer j
Ii Injection rate of the ith injector (STB/Day)
Jj Productivity index of the jth producer (STB/Day.psi)
Ninj Number of injectors
Npro Number of producers
ng Number of the elements in the green squares of the
Boston matrix
no Number of the elements in the orange squares of the
Boston matrix
nr Number of the elements in the red squares of the
Boston matrix
Pw Wellbore pressure (psi)
Q Flow rate (STB/Day)
qj Production rate of the jth producer (STB/Day)
Scoreij Score of the connection between the injector i and
the producer j in DoE
Tmin Minimum time lag for an event of an injector to be
seen in a producer (Days)
Tmax Maximum time lag for an event of an injector to be
seen in a producer (Days)
T Time (Days)
a Constant variation derivate (STB/Day2
or psi/Day)
b Minimum event variation
Table 1
Quality of the connections according to the connection factor
matrix in CRM.
Amount of the connection factor Quality of the connection
fij  0.2 Weak
0.2  fij  0.5 Medium
0.5  fij Strong
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
2 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
quality of connection between each pair of injector-producer, DoE
detects injector and producer events, then it identifies that every one
of the producer events is due to changes in producer choke valve or
caused by the certain injector event. According to the number of evi-
dences that represent the connection between each pair of injector-
producer, an interwell connection will be assumed for these wells.
In order to discriminate between non-connection events (trig-
gered by an operator intervention at the surface) and connection
events (due to the reservoir response to other inputs), evidence for
operator action is required. When the operator changes the choke
valve size, the flow rate and wellbore pressure change simulta-
neously. These variations lead to non-connection events that are not
representative of a connection between wells. DoE requires injection
rates of the injectors for identification of event start date and dura-
tion. But in producers, DoE requires flow rate and wellbore pressure
in order to check whether the events are made by a change in the
choke valve or not.
DoE is implemented in 3 steps including, 1) detection and evalua-
tion of the events in the injection rates of the injection wells, 2)
detection of the events in the production rates and bottom-hole pres-
sure (BHP) of the production wells, and 3) comparison and evaluation
of the event in order to find the connectivity and association between
each pair of injector-producer. The steps of a DoE algorithm are pre-
sented below:
2.2.1. Detection and evaluation of the events in the injectors
In the first step, the events of injection rates are identified. The
event of an injector rate means a sharp change in injection rate, shut-
in of the injector, or its constant flow rate (during a certain period of
time). For an injection flow rate, four types of events exist, including
Increases, Decreases, No flow, and Smooth. In Table 2 and Fig. 1, four
types of these events have been presented.
No flow events are detected easily from the injection data. In
order to detect other types of events, it is required to define a crite-
rion (based on derivate of flow rate), to find out whether the flow
rate is increasing, decreasing, or constant. For this purpose, it is
required to define a parameter named constant variation derivate
(a). A constant variation derivate is a constant number (derivate) for
determining types of changes in the wellbore flow rate. The slope of
the flow rate diagram versus time (@Q
@T ) is compared with the value of
a in order to recognize the type of events in each day, as presented in
Table 3. In Table 3,Q is the flow rate and T is the time.
After finding the type of events in each day, the same and sequen-
tial daily events are merged. For example, 2 sequential increasing
events in each day are merged to have an increasing event that lasts
for 2 days. Then, in order to find the considerable Increasing and
Decreasing events, the absolute change of flow rate to its primary
value is compared with a parameter named minimum event variation
(b) as presented in Table 4. Those events that their absolute variation
is less than b are excluded from the Increasing and Decreasing events
and allocated to the Smooth events.
2.2.2. Detection of the events in the production rate and pressure of the
production wells
For each producer, daily events of production rate and wellbore
pressure are detected using the method presented in the previous
section (same procedure as the injection flow rate). For daily events
of wellbore pressures, the value of constant variation derivate (a) is
different from the one for flow rates. No merging will be done for the
events of production wells, and there is only daily events for the pro-
duction rate and wellbore pressure. This is because these daily events
are required to detect the events which are created by the operator
(changing choke valve size). For instance, if the operator is opening
the choke valve (Increasing event of the production rate in the pro-
ducer), the pressure of the producer is decreasing so rapidly
(Decreasing event of the pressure in the producer). This escalation in
the production rate of the producer is not representative of an inter-
well connection because it is not created by increasing the injection
rate of other injectors.
Fig. 2 shows an example of a change in choke valve size. The rate
of the injector i is decreasing on day 62, and it seems that with 5 days
lag time the flow rate of producer j is decreasing too (due to the injec-
tor i). But, the pressure of the producer j is increasing. This means that
the reduction in the producer rate is due to the choke valve closing by
the operator, and these events are not representative of the connec-
tion between injector i and producer j.
2.2.3. Comparison and evaluation of the events to find the connection
between each pair of injector-producer
As discussed in the previous section, although some events in the
production wells are created by the operator, there are some other
events in the production wells that have been created by the injec-
tion wells as presented in Table 5.
For instance, as can be observed in Fig. 3, the production rate
event in the producer j is increasing (on day 110), meanwhile the
pressure event of the producer is constant. If this increment in the
producer rate was due to the opening of the choke valve, the pressure
of the producer would be decreasing so rapidly which is not the case
here. This means that an external factor is supporting the producer.
In this situation, if the injection rate event in the injector (with a
proper lag time) is increasing too, this pair of injector-producer will
be assigned one score. As shown in Fig. 3, there is an increasing event
in the injector i on the 100th day, so this pair of injector-producer
will be assigned one score in DoE. The minimum lag time (Tmin) and
the maximum lag time (Tmax) after the beginning of the injector event
are selected as the boundaries for evaluation of producer events. Only
the production rate and pressure events that their start time is in this
timespan will be considered for scoring. The values of Tmin and Tmin
depends on the size of the reservoir, and are as the input parameters
determined by the user. In Fig. 4, two sequences of events that show
the connection and non-connection between one pair of injector-
producer (for a real case), have been presented.
After surveying all of the events between each pair of injector-
producer, a score matrix with Ninj columns and Npro rows is produced.
Then, according to the 10 percentile and 50 percentile of the score
matrix, the quality of the connection is determined as presented in
Table 6.
3. Results and discussions
For analysis the correctness of the proposed method, both syn-
thetic and real waterflooding cases were studied in this section. For
each case, CRM and DoE are run to find the connection between
wells. Then, the estimated interwell connections of both methods are
compared and discussed.
3.1. Synthetic case
A simple synthetic waterflooding case was used in this section to
perform a detailed analysis of the results of the DoE. The synthetic
case is a homogeneous 1D reservoir that has been flooded for 100
Days. It consists of only one injection well and two production wells
as presented in Fig. 5. For simplicity of the synthetic case, the
Table 2
Events of the flow rate.
Event name Definition
Increasing An increase in flow rate
Decreasing A decrease in flow rate
No flow Shut-in well
Smooth Flow rate is almost constant
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 3
bottom-hole pressure of both producers is constant during water-
flood operation, so the last term of Eq. (1) in CRM and the pressure
events in DoE can be neglected.
3.1.1. Results of CRM for the synthetic case
As was discussed in Section 2.1, the connection factors of CRMP
(producer base representation of the CRM) can be used to find the
quality of the connection between injectors and producers. Hence,
this model is run for finding the interwell connections in the syn-
thetic reservoir. History matching of injection and production data
leads to a connection factor matrix (fij) with Ninj columns and Npro
rows as presented in Table 7. Fig. 6 presents the simulated (from
CRMP) and observed total (oil+water) production rate of the syn-
thetic reservoir. According to Table 7, the connection factor between
injector 1 and producer 1 (I1  P1) is 0.65, and for injector 1 and pro-
ducer 2 (I1  P2) is 0.35. It means that there is a strong connection for
the first pair and a medium connection for the second one (according
to Table 1). This is because of the producer 1 closeness to the injector
1 in comparison with the producer 2.
Fig. 1. Different types of events for the flow rate.
Table 3
Daily event detection.
Criterion Event type
@Q
@T  a Increasing event
@Q
@T  a Decreasing event
a @Q
@T a Smooth event
Table 4
Detection of Smooth events.
Criterion Event type
j Q2Q1
Q1
jb Increasing or Decreasing event
j Q2Q1
Q1
j  b Smooth event
Fig. 2. Events those are not representative of the connection between injector i and
producer j.
Table 5
Events that present the connection between a pair of injector-producer.
Production pressure event Production rate event Injection rate event
Increasing Increasing Increasing
Smooth Increasing Increasing
Increasing Smooth Increasing
Decreasing Smooth Decreasing
Decreasing Decreasing Decreasing
Smooth Increasing Smooth
Decreasing Decreasing No flow
Decreasing Smooth No flow
Fig. 3. Events that are representative of the connection between injector i and pro-
ducer j.
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
4 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
3.1.2. Results of DoE for the synthetic case
According to the discussions in the DoE methodology, some con-
stants are needed for running a DoE. For the synthetic case, a is
selected 50 (STB/Day2
), b is 0.2. Tmin and Tmax are selected according
to the size of the injectors events (Tmin = 0 Days and Tmax = 5 Days).
Fig. 7 presents the events of the injector 1 versus the events of pro-
ducers 1 and 2. Also, the details of these events have been presented
in Table 8 and Table 9. Note that for the simplicity smooth events
have not been reported here, because they don’t affect the connec-
tions in this case. For the big data containing a huge number of
events, a code can be developed for the detecting and scoring the
events as described in Section 2.2.
As presented in Fig. 7, there are 2 Increasing-Decreasing events in
the injector. The magnitude of the first Increasing-Decreasing events
are too large, so they create Increasing-Decreasing events in both
producers. Hence, according to the size of these events (duration),
the score of the first events are allocated to both pair of injector-pro-
ducer (see Table 9). The second Increasing-Decreasing events are
small, so they cannot be detected in producer 2 (a  50), and the sec-
ond producer receives no score from these events.
The score matrix (achieved from DoE) for the synthetic case is pre-
sented in Table 10. According to the 50 percentile of the score matrix
(see Table 6), producer 1 shows a strong connection with injector 1,
while a medium connection is observed between producer 2 and the
injector. Therefore, both of CRMP and DoE are in agreement and pres-
ent the same results about the connections of the synthetic case.
Fig. 8 presents the connection map of DoE and CRMP for the synthetic
reservoir. In this figure, different connection strengths are shown by
the line width. A thick line represents a relatively strong connection,
and a medium line represents a medium connection.
3.2. Real case (field case)
In this section, the injection and production data of a real res-
ervoir have been studied. Workflows are implemented in this res-
ervoir to continuously improve their performance. In the real
case, it is important to highlight key assumptions to ensure
higher confidence in results, especially in the application of CRM.
One requirement is the judicious selection of the time interval for
the study. The selected data should include a minimum of
120 days of injection and production (to ensure that the wells
have affected each other), with a relatively constant number of
the active injector and producer wells.
The studied reservoir consists of several production and injec-
tion wells, but only for 250 days, the number of active wells in
this field is constant (no new well is drilled or perforated, and
there are 6 active producers and 4 active injectors). So, we use
the injection and production data of this specified timespan for
DoE and CRM analyses. The schematic of the real reservoir has
been presented in Fig. 9.
3.2.1. Results of CRM for the real case
Similar to the synthetic case, CRMP is used for finding the inter-
well connections in the real reservoir, and then the results are com-
pared with the results of the DoE method. Fig. 10 shows the
simulated (from CRMP) and observed total (oil+water) production
rate of the entire reservoir. According to the high value of R-squared
(0.916) presented in Fig. 10, the results of the simulated production
rate (CRM) show a good agreement with observation data. Fig. 11
shows the connectivity map of the field achieved by CRM (according
to Table 1). As presented in Fig. 11, CRMP estimates that there is a
strong underground connection between injector 1 and producer 5
(I1  P5), and injector 3 and producer 3 (I3  P3).
Fig. 4. Two sequences of connection and non-connection events in a real case.
Table 6
Quality of the connections according to the score
matrix created by DoE.
The value of the score Quality of the connection
Scoreij  P10 Weak
P10  Scoreij  P50 Medium
P50  Scoreij Strong
Fig. 5. Schematic of the synthetic reservoir (red circles are the producers and the blue
triangle is the injector).(For interpretation of the references to color in this figure leg-
end, the reader is referred to the web version of this article.)
Table 7
Connection fac-
tor matrix for
the synthetic
case.
fij I1
P1 0.65
P2 0.35
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 5
3.2.2. Results of DoE for the real case
Sensible values must be selected for each one of the DoE parame-
ters. In the real case, a is 300 STB/Day2
for the flow rate events, and
50 psi/Day for the pressure events. b is 0.1, Tmin is 2 Days and Tmax is 8
Days. In Fig. 12, the results of the connectivity map of the DoE for the
entire field have been shown. As presented in Fig. 12, DoE estimates
that there is a strong underground connection between injector 2
and producer 3 (I2  P3), injector 3 and producer 3 (I3  P3), and injec-
tor 3 and producer 4 (I3  P4). Injector 3 and producer 3 (I3  P3), was
reported to have a strong connection by both DoE and CRM. DoE
Fig. 6. Total production rate of the synthetic reservoir.
Fig. 7. Events of the injector 1 versus the events of producers 1and 2 in the synthetic case.
Table 8
Events of the injector 1 in the synthetic case.
No. Event type Duration (Days) a b
E1 Increasing 914 600 3
E2 Decreasing 1419 600 3
E3 Increasing 4951 200 1
E4 Decreasing 5156 200 1
Table 9
Events of the producers in the synthetic case.
Well No. Event type Duration a b Connection Score Total score
P1
E5 Increasing 915 195 1.74 Yes 6
28
E6 Decreasing 1526 103.7 0.61 Yes 11
E7 Increasing 4955 67.3 0.62 Yes 6
E8 Decreasing 5661 55.9 0.26 Yes 5
P2
E9 Increasing 1015 104.5 1.11 Yes 5
11
E10 Decreasing 1622 77 0.46 Yes 6
E11 Increasing 5055 37.7 0.51 No 0
E12 Decreasing 5661 26.4 0.23 No 0
Table 10
Score matrix for
the synthetic case.
Scoreij I1
P1 28
P2 11
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
6 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
considers the connection between injector 2 and producer 3 (I2  P3)
as a strong connection, while CRM expresses that this connection is
medium. CRM estimates that there is a strong connection between
injector 1 and producer 5 (I1  P5), while DoE estimates that this is a
weak connection.
It is time-consuming for analyzing one by one of the estimated
connections for a real reservoir. Hence, the Boston matrix has been
used in order to have a quick and fair comparison between the con-
nections calculated by CRM and DoE (see Table 11). In the Boston
matrix, as the higher number of elements are on the main diagonal,
the results are in a better agreement. The green squares show the
number of pairs that have been estimated similarly by CRM and DoE
(for example, 14 pairs with a weak connection are recognized by
both CRM and DoE). The red squares show where the two models dis-
agree with each other (for example, there is a pair that CRM considers
Fig. 8. Connection map of the synthetic reservoir (estimated by CRMP and DoE).
Fig. 9. Schematic of the real reservoir.
Fig. 10. Total production rate of the entire reservoir.
Fig. 11. Underground connection map of the real reservoir (estimated by CRMP; a
thick line represents a relatively strong connection, a medium line represents a
medium connection, and a thin dashed line signifies a weak connection).
Fig. 12. Underground connection map of the reservoir (estimated by DoE).
Table 11
Results of CRM and DoE in Boston matrix.
Connection Weak (CRM) Medium (CRM) Strong (CRM)
Weak (DoE) 14 1 1
Medium (DoE) 4 1 0
Strong (DoE) 1 1 1
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 7
as a strong connection, but DoE considers weak. Also there is another
pair that DoE considers a strong connection while CRM considers
weak). The orange squares show the connection strengths that are in
slight disagreement (CRM considers a connection medium strength,
while DoE considers the connection to be a weak connection).
A criterion was developed to express the agreement between two
approaches in the Boston matrix. In the developed criterion, each
square with specific color awards a specific point. For example, 3
points are awarded to the pairs which are located in the green
squares of the Boston matrix, 1 point to the pairs of the orange
squares, and no points are awarded to the pairs located in the red
squares. Based on this criterion, it can be inferred that what percent
the estimated connection strength by DoE confirms the estimated
connection strength by CRM. The agreement parameter (AP) between
2 methods is calculated using the equation presented below:
AP ¼
3ng þ no
3ðng þ no þ nrÞ
 100 ð3Þ
where, ng is the number of pairs in the green squares, no is the num-
ber of pairs in the orange squares, and nr is the number of pairs in the
red squares. According to the Table 11, 16, pairs are in the green
squares, 6 pairs in the orange squares, and 2 pairs in the red region.
Consequently, there is 75% agreement between CRM and DoE in the
studied case.
Another advantage of using this method is the confidence about
the true underground interwell connections. For example, a strong
connection has been estimated between injector 3 and producer 3
(I3  P3) in both methods. So, this strong connection can be reliable.
CRM estimates there is a strong connection between injector 1 and
producer 5 (I1  P5), while DoE estimates that this connection is
weak. Hence, the connection between this pair is not trustable and
requires more surveillance techniques. According to the long distance
between injector 1 and producer 5 (see Fig. 12), the distrust about
this pair is not weird. The fact that events on an injection well are
correlated in time with events on a separate producer well does not
provide a predictive mechanism for how future production from that
well might usefully vary. However, a strong association is still indica-
tive of a communication path between the wells and this insight
could be great information to develop the reservoir in the future.
3.2.3. Comparison of DoE and ANN
Artificial neural networks (ANNs) also is used as a data-driven
method for estimating the underground interwell connections [5]. In
the present section, a comparison is performed between the ANN
and DoE approaches in terms of accuracy, data requirements, exper-
tise requirements, training algorithm, and processing times. To do so,
ANN was applied for the real case (field case) using the backpropaga-
tion algorithm and 1 hidden layer. Then, the normalized ANN connec-
tions compared with the CRM connections and presented in the form
of the Boston matrix (Table 12).
Accuracy: It is believed that CRM presents more reliable interwell
connections in comparison with ANN, as CRM is a physics-based
approach [13,14,26]. Therefore, the value of the agreement with CRM
(Eq. (3)) is selected as the accuracy criteria in this section. Regarding
Table 12 and using Eq. (3), there is a 68% agreement between the
obtained result from CRM and ANN. As the agreement parameter for
CRM and DoE was 75%, it can be inferred that the DoE is more consis-
tent with the CRM (more precise) in this studied case.
Data requirements: In terms of required data, both DoE and ANN
required the same set of data (the injection and production data).
Expertise requirements: When it comes to working with ANNs,
deep knowledge of the ANN theory and training parameters is cru-
cial. On the other hand, DoE does not requires any optimization algo-
rithm and only needs the proper identification of the parameters
used for the detection of the events.
Training algorithm: No training algorithm is required for the DoE,
while the performance of the ANN strongly depends on the training
algorithms.
Processing time: Both DoE and ANN methods are fast, where the
CPU time is less than a few minutes.
According to all these terms, it seems that DoE is more efficient
than ANN. DoE is in a better agreement with CRM and is easier to
apply. However, DoE only can be used for estimation of the interwell
connections and cannot predict the production rate of the reservoir.
4. Conclusions
In this paper, a novel data-driven approach (DoE) was proposed in
order to find the underground connections between injection and
production wells. The proposed method is based on finding the
impact of injection rate fluctuations on the production rates and scor-
ing the relevant impacts. Then, by establishing an effective calibration
between CRM and DoE, the confidence about the estimated under-
ground connections were analyzed. The proposed method was suc-
cessfully tested for both synthetic and real reservoirs, and the
estimations of these two methods (CRM and DoE) showed a good
agreement with each other (75%).
The proposed method can be used by reservoir engineers for
understanding the underground path of the injected fluid, and deci-
sion-making about the future. Although there is a high degree of sim-
ilarity between DoE and CRM, it is highly recommended to integrate
this methodology with other existing surveillance techniques, such
as tracer, streamline modeling, and 4D seismic to evaluate and moni-
tor the waterflooding process more accurately.
Declaration of Competing Interest
None.
References
[1] Adedigba SA, Oloruntobi O, Khan F, Butt S. ''Data-driven dynamic risk analysis of
offshore drilling operations. J Petrol Sci Eng 2018;165:444–52.
[2] Akin S. Analysis of tracer tests with simple spreadsheet models. Comput Geosci
2001;27(2):171–8.
[3] Amar MN, Ghahfarokhi AJ, Zeraibi N. ''Predicting thermal conductivity of carbon
dioxide using group of data-driven models. J Taiwan Inst Chem Eng 2020:3489.
[4] Amin MT, Khan F, Ahmed S, Imtiaz S. ''A novel datadriven methodology for fault
detection and dynamic risk assessment. Can J Chem Eng 2020.
[5] Artun E. Characterizing interwell connectivity in waterflooded reservoirs using
data-driven and reduced-physics models: a comparative study. Neural Comput
Appl 2017;28(7):1729–43.
[6] Chitrala Y, Moreno C, Sondergeld C, Rai C. ''An experimental investigation into
hydraulic fracture propagation under different applied stresses in tight sands
using acoustic emissions. J Petrol Sci Eng 2013;108:151–61.
[7] Eshraghi SE, Rasaei MR, Zendehboudi S. ''Optimization of miscible CO2 EOR and
storage using heuristic methods combined with capacitance/resistance and Gentil
fractional flow models. J Nat Gas Sci Eng 2016;32:304–18.
[8] Goudarzi A, Delshad M, Sepehrnoori K. ''A chemical EOR benchmark study of dif-
ferent reservoir simulators. Comput Geosci 2016;94:96–109.
[9] Guo Z, Reynolds AC, Zhao H. ''Waterflooding optimization with the INSIM-FT
data-driven model. Comput Geosci 2018:1–17.
[10] Heffer KJ, Fox RJ, McGill CA, Koutsabeloulis NC. ''Novel techniques show links
between reservoir flow directionality, earth stress, fault structure and geome-
chanical changes in mature waterfloods. SPE J 1997;2(02):91–8.
[11] Izgec O, Kabir CS. ''Quantifying nonuniform aquifer strength at individual wells.
SPE Reserv Evaluat Eng 2010;13(02):296–305.
[12] Jahangiri HR, Adler C, Shirzadi S, Bailey R, Ziegel E, Chesher J, White M. A data-
driven approach enhances conventional reservoir surveillance methods for
waterflood performance management in the North Sea. SPE intelligent energy
conference  exhibition, society of petroleum engineers; 2014.
Table 12
Results of CRM and ANN in the Boston matrix.
Connection Weak (CRM) Medium (CRM) Strong (CRM)
Weak (ANN) 12 2 1
Medium (ANN) 3 1 1
Strong (ANN) 2 1 1
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
8 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
[13] Jamali A, Ettehadtavakkol A. ''Application of capacitance resistance models to
determining interwell connectivity of large-scale mature oil fields. Petrol Explorat
Dev 2017;44(1):132–8.
[14] Jensen JL. Comment on “Characterizing interwell connectivity in waterflooded
reservoirs using data-driven and reduced-physics models: a comparative study”
by E. Artun.''. Neural Comput Appl 2017;28(7):1745–6.
[15] Kaviani D, Jensen JL, Lake LW. ''Estimation of interwell connectivity in the
case of unmeasured fluctuating bottomhole pressures. J Petrol Sci Eng
2012;90:79–95.
[16] Kaviani D, Soroush M, Jensen JL. ''How accurate are capacitance model connectiv-
ity estimates? J Petrol Sci Eng 2014;122:439–52.
[17] Kaviani D, Valk
o PP. ''Inferring interwell connectivity using multiwell productiv-
ity index (MPI). J Petrol Sci Eng 2010;73(12):48–58.
[18] Liang X. A simple model to infer interwell connectivity only from well-rate fluctu-
ations in waterfloods. J Petrol Sci Eng 2010;70(12):35–43.
[19] Mamghaderi A, Aminshahidy B, Bazargan H. ''Error behavior modeling in capaci-
tance-resistance model: a promotion to fast, reliable proxy for reservoir perfor-
mance prediction. J Nat Gas Sci Eng 2020:103228.
[20] Mamghaderi A, Bastami A, Pourafshary P. ''Optimization of waterflooding perfor-
mance in a layered reservoir using a combination of capacitance-resistive model
and genetic algorithm method. J Energy Resour Technol 2013;135(1):013102.
[21] Mirzayev M, Riazi N, Cronkwright D, Jensen JL, Pedersen PK. ''Determining well-
to-well connectivity using a modified capacitance model, seismic, and geology for
a Bakken Waterflood. J Petrol Sci Eng 2017;152:611–27.
[22] Moreno GA. Multilayer capacitanceresistance model with dynamic connectivi-
ties. J Petrol Sci Eng 2013;109:298–307.
[23] Moreno GA, Lake LW. ''On the uncertainty of interwell connectivity estimations
from the capacitance-resistance model. Petrol Sci 2014;11(2):265–71.
[24] Nhat DM, Venkatesan R, Khan F. ''Data-driven Bayesian network model for early
kick detection in industrial drilling process. Process Saf Environ Protect 2020.
[25] Onalo D, Adedigba S, Khan F, James LA, Butt S. ''Data driven model for sonic well
log prediction. J Petrol Sci Eng 2018;170:1022–37.
[26] Panda M, Chopra A. An integrated approach to estimate well interactions. SPE
India oil and gas conference and exhibition, society of petroleum engineers;
1998.
[27] Refunjol BT. Reservoir characterization of north buck draw field based on tracer
response and production/injection analysis. University of Texas at Austin; 1996.
[28] Sant'Anna, P.J. ''Estimating injectivity and lateral autocorrelation in heteroge-
neous media.''(1999).
[29] Sayarpour M. Development and application of capacitance-resistive models to
water/carbon dioxide floods. The University of Texas at Austin; 2008.
[30] Sayarpour M, Zuluaga E, Kabir CS, Lake LW. ''The use of capacitanceresistance
models for rapid estimation of waterflood performance and optimization. J Petrol
Sci Eng 2009;69(3):227–38.
[31] Shabani, A., M.S. Moosavi, D. Zivar and H.R. Jahangiri. ''Data-driven technique for
analyzing the injector efficiency in a waterflooding operation.'' SIMULATION:
0037549720923386.(2020).
[32] Soeriawinata T, Kelkar M. Reservoir management using production data. SPE mid-
continent operations symposium; 1999.
[33] Soroush M, Kaviani D, Jensen JL. ''Interwell connectivity evaluation in cases of
changing skin and frequent production interruptions. J Petrol Sci Eng
2014;122:616–30.
[34] Van SL, Chon BH. ''Effective prediction and management of a CO2 flooding process
for enhancing oil recovery using artificial neural networks. J Energy Resour Tech-
nol 2017;140(3) 032906-032906-032914.
[35] Wang D, Li Y, Chen B, Hu Y, Li B, Gao D, Fu. B. Ensemble-based optimization of
interwell connectivity in heterogeneous waterflooding reservoirs. J Nat Gas Sci
Eng 2017;38:245–56.
[36] Weber D, Edgar TF, Lake LW, Lasdon LS, Kawas S, Sayarpour M. Improvements in
capacitance-resistive modeling and optimization of large scale reservoirs. SPE
western regional meeting, society of petroleum engineers; 2009.
[37] Xu T, Valocchi AJ. ''Data-driven methods to improve baseflow prediction of a
regional groundwater model. Comput Geosci 2015;85:124–36.
[38] Yousef AA, Gentil PH, Jensen JL, Lake LW. A capacitance model to infer interwell
connectivity from production and injection rate fluctuations. SPE annual technical
conference and exhibition, society of petroleum engineers; 2005.
[39] Zhao H, Kang Z, Zhang X, Sun H, Cao L, Reynolds AC. ''A physics-based data-driven
numerical model for reservoir history matching and prediction with a field appli-
cation (associated discussion available as supporting information). SPE J 2016;21
(06):175. 2172194.
ARTICLE IN PRESS
JID: JTICE [m5G;November 19, 2020;9:37]
Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the
Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008
E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 9

More Related Content

What's hot

Session 83 Ola Wik
Session 83 Ola WikSession 83 Ola Wik
Session 83 Ola WikOla Wik
 
ConcreteBlockLCAReport
ConcreteBlockLCAReportConcreteBlockLCAReport
ConcreteBlockLCAReportJoseph Tursi
 
An efficient method for assessing water
An efficient method for assessing waterAn efficient method for assessing water
An efficient method for assessing waterijsc
 
Luca_Carniato_PhD_thesis
Luca_Carniato_PhD_thesisLuca_Carniato_PhD_thesis
Luca_Carniato_PhD_thesisLuca Carniato
 
Due diligence reviews of mineral resource estimates
Due diligence reviews of mineral resource estimatesDue diligence reviews of mineral resource estimates
Due diligence reviews of mineral resource estimatesPeter Ravenscroft
 
Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...
Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...
Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...ASC-HRSC
 
Progress report STRAINS
Progress report STRAINSProgress report STRAINS
Progress report STRAINSbiometrust
 
Team M Reservoir simulation-an extract from original ppt
Team M Reservoir simulation-an extract from original pptTeam M Reservoir simulation-an extract from original ppt
Team M Reservoir simulation-an extract from original pptMukesh Mathew
 
Isct group us llc tech article e&p aug 2017.v f
Isct group us llc tech article e&p aug 2017.v fIsct group us llc tech article e&p aug 2017.v f
Isct group us llc tech article e&p aug 2017.v fISCT GROUP US LLC
 

What's hot (15)

Session 83 Ola Wik
Session 83 Ola WikSession 83 Ola Wik
Session 83 Ola Wik
 
ConcreteBlockLCAReport
ConcreteBlockLCAReportConcreteBlockLCAReport
ConcreteBlockLCAReport
 
An efficient method for assessing water
An efficient method for assessing waterAn efficient method for assessing water
An efficient method for assessing water
 
Ijciet 10 01_169
Ijciet 10 01_169Ijciet 10 01_169
Ijciet 10 01_169
 
Luca_Carniato_PhD_thesis
Luca_Carniato_PhD_thesisLuca_Carniato_PhD_thesis
Luca_Carniato_PhD_thesis
 
IJESD
IJESDIJESD
IJESD
 
Mineral Exploration QAM
Mineral Exploration QAMMineral Exploration QAM
Mineral Exploration QAM
 
Due diligence reviews of mineral resource estimates
Due diligence reviews of mineral resource estimatesDue diligence reviews of mineral resource estimates
Due diligence reviews of mineral resource estimates
 
Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...
Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...
Improving Remedial Actions Through Integrated Use of Direct-Push HRSC Technol...
 
Progress report STRAINS
Progress report STRAINSProgress report STRAINS
Progress report STRAINS
 
Team M Reservoir simulation-an extract from original ppt
Team M Reservoir simulation-an extract from original pptTeam M Reservoir simulation-an extract from original ppt
Team M Reservoir simulation-an extract from original ppt
 
MOHAMED HELMY_CV_2016_UPDATED
MOHAMED HELMY_CV_2016_UPDATED MOHAMED HELMY_CV_2016_UPDATED
MOHAMED HELMY_CV_2016_UPDATED
 
und0802final
und0802finalund0802final
und0802final
 
Isct group us llc tech article e&p aug 2017.v f
Isct group us llc tech article e&p aug 2017.v fIsct group us llc tech article e&p aug 2017.v f
Isct group us llc tech article e&p aug 2017.v f
 
harsha_thesis
harsha_thesisharsha_thesis
harsha_thesis
 

Similar to dastgerdi2020 (2).pdf

Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...
Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...
Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...IRJET Journal
 
REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...
REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...
REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...ijiert bestjournal
 
Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...
Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...
Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...IRJET Journal
 
IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...
IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...
IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...IRJET Journal
 
Cfd analaysis of flow through a conical exhaust
Cfd analaysis of flow through a conical exhaustCfd analaysis of flow through a conical exhaust
Cfd analaysis of flow through a conical exhausteSAT Publishing House
 
Fluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFD
Fluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFDFluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFD
Fluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFDtheijes
 
Oilfield Modeling and Optimization of a Mature Field using Integrated Product...
Oilfield Modeling and Optimization of a Mature Field using Integrated Product...Oilfield Modeling and Optimization of a Mature Field using Integrated Product...
Oilfield Modeling and Optimization of a Mature Field using Integrated Product...IRJET Journal
 
IRJET- Experimental Study on Flow through Venturimeter
IRJET- Experimental Study on Flow through VenturimeterIRJET- Experimental Study on Flow through Venturimeter
IRJET- Experimental Study on Flow through VenturimeterIRJET Journal
 
OTC 14009 Deep Offshore Well Metering and Permutation Testing
OTC 14009 Deep Offshore Well Metering and Permutation TestingOTC 14009 Deep Offshore Well Metering and Permutation Testing
OTC 14009 Deep Offshore Well Metering and Permutation Testingguest467223b
 
FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMP
FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMPFLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMP
FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMPijiert bestjournal
 
Validation of Experimental and Numerical Techniques for Flow Analysis over an...
Validation of Experimental and Numerical Techniques for Flow Analysis over an...Validation of Experimental and Numerical Techniques for Flow Analysis over an...
Validation of Experimental and Numerical Techniques for Flow Analysis over an...IJERA Editor
 
Statistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters toStatistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters toeSAT Publishing House
 
Statistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters toStatistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters toeSAT Publishing House
 
Jurnalku pompa
Jurnalku pompaJurnalku pompa
Jurnalku pompaAmardhiana
 
Computational fluid dynamics
Computational fluid dynamicsComputational fluid dynamics
Computational fluid dynamicsZeeshan Inamdar
 
How can identify sensitivity of hydraulic characteristics of irrigation systems?
How can identify sensitivity of hydraulic characteristics of irrigation systems?How can identify sensitivity of hydraulic characteristics of irrigation systems?
How can identify sensitivity of hydraulic characteristics of irrigation systems?AI Publications
 
Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...
Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...
Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...Carlos R. Conde
 
Download-manuals-surface water-software-processingofstreamflowdata
 Download-manuals-surface water-software-processingofstreamflowdata Download-manuals-surface water-software-processingofstreamflowdata
Download-manuals-surface water-software-processingofstreamflowdatahydrologyproject0
 
Download-manuals-surface water-software-processingofstreamflowdata
 Download-manuals-surface water-software-processingofstreamflowdata Download-manuals-surface water-software-processingofstreamflowdata
Download-manuals-surface water-software-processingofstreamflowdatahydrologyproject0
 

Similar to dastgerdi2020 (2).pdf (20)

Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...
Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...
Experimental and Computational Fluid Dynamics (CFD) Analysis of Additively Ma...
 
REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...
REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...
REVIEW OF FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGA...
 
Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...
Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...
Artificial Neural Network Modelling for Pressure Drop Estimation of Oil-Water...
 
IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...
IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...
IRJET - Kinetic Model on the Performance of an Anaerobic Baffled Reactor for ...
 
Water pipes
Water pipesWater pipes
Water pipes
 
Cfd analaysis of flow through a conical exhaust
Cfd analaysis of flow through a conical exhaustCfd analaysis of flow through a conical exhaust
Cfd analaysis of flow through a conical exhaust
 
Fluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFD
Fluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFDFluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFD
Fluid Dynamics Simulation of Two-Phase Flow in a Separator Vessel through CFD
 
Oilfield Modeling and Optimization of a Mature Field using Integrated Product...
Oilfield Modeling and Optimization of a Mature Field using Integrated Product...Oilfield Modeling and Optimization of a Mature Field using Integrated Product...
Oilfield Modeling and Optimization of a Mature Field using Integrated Product...
 
IRJET- Experimental Study on Flow through Venturimeter
IRJET- Experimental Study on Flow through VenturimeterIRJET- Experimental Study on Flow through Venturimeter
IRJET- Experimental Study on Flow through Venturimeter
 
OTC 14009 Deep Offshore Well Metering and Permutation Testing
OTC 14009 Deep Offshore Well Metering and Permutation TestingOTC 14009 Deep Offshore Well Metering and Permutation Testing
OTC 14009 Deep Offshore Well Metering and Permutation Testing
 
FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMP
FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMPFLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMP
FLOW DISTRIBUTION NETWORK ANALYSIS FOR DISCHARGE SIDE OF CENTRIFUGAL PUMP
 
Validation of Experimental and Numerical Techniques for Flow Analysis over an...
Validation of Experimental and Numerical Techniques for Flow Analysis over an...Validation of Experimental and Numerical Techniques for Flow Analysis over an...
Validation of Experimental and Numerical Techniques for Flow Analysis over an...
 
Statistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters toStatistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters to
 
Statistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters toStatistical analysis to identify the main parameters to
Statistical analysis to identify the main parameters to
 
Jurnalku pompa
Jurnalku pompaJurnalku pompa
Jurnalku pompa
 
Computational fluid dynamics
Computational fluid dynamicsComputational fluid dynamics
Computational fluid dynamics
 
How can identify sensitivity of hydraulic characteristics of irrigation systems?
How can identify sensitivity of hydraulic characteristics of irrigation systems?How can identify sensitivity of hydraulic characteristics of irrigation systems?
How can identify sensitivity of hydraulic characteristics of irrigation systems?
 
Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...
Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...
Black Oil Reservoir Simulation Model Mis-Match Assessment through Computer-As...
 
Download-manuals-surface water-software-processingofstreamflowdata
 Download-manuals-surface water-software-processingofstreamflowdata Download-manuals-surface water-software-processingofstreamflowdata
Download-manuals-surface water-software-processingofstreamflowdata
 
Download-manuals-surface water-software-processingofstreamflowdata
 Download-manuals-surface water-software-processingofstreamflowdata Download-manuals-surface water-software-processingofstreamflowdata
Download-manuals-surface water-software-processingofstreamflowdata
 

Recently uploaded

MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINEMANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINESIVASHANKAR N
 
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICSHARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICSRajkumarAkumalla
 
What are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptxWhat are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptxwendy cai
 
Call Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile serviceCall Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile servicerehmti665
 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Serviceranjana rawat
 
Call Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur Escorts
Call Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur EscortsCall Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur Escorts
Call Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur EscortsCall Girls in Nagpur High Profile
 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCall Girls in Nagpur High Profile
 
Internship report on mechanical engineering
Internship report on mechanical engineeringInternship report on mechanical engineering
Internship report on mechanical engineeringmalavadedarshan25
 
Biology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptxBiology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptxDeepakSakkari2
 
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsCall Girls in Nagpur High Profile
 
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130Suhani Kapoor
 
Processing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptxProcessing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptxpranjaldaimarysona
 
Introduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxIntroduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxupamatechverse
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130Suhani Kapoor
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Dr.Costas Sachpazis
 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escortsranjana rawat
 
Call Girls in Nagpur Suman Call 7001035870 Meet With Nagpur Escorts
Call Girls in Nagpur Suman Call 7001035870 Meet With Nagpur EscortsCall Girls in Nagpur Suman Call 7001035870 Meet With Nagpur Escorts
Call Girls in Nagpur Suman Call 7001035870 Meet With Nagpur EscortsCall Girls in Nagpur High Profile
 

Recently uploaded (20)

MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINEMANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
MANUFACTURING PROCESS-II UNIT-2 LATHE MACHINE
 
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICSHARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
 
What are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptxWhat are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptx
 
Call Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile serviceCall Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile service
 
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptxExploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
 
Call Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur Escorts
Call Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur EscortsCall Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur Escorts
Call Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur Escorts
 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
 
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINEDJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
 
Internship report on mechanical engineering
Internship report on mechanical engineeringInternship report on mechanical engineering
Internship report on mechanical engineering
 
Biology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptxBiology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptx
 
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
 
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
 
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
 
Processing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptxProcessing & Properties of Floor and Wall Tiles.pptx
Processing & Properties of Floor and Wall Tiles.pptx
 
Introduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxIntroduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptx
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
 
Call Girls in Nagpur Suman Call 7001035870 Meet With Nagpur Escorts
Call Girls in Nagpur Suman Call 7001035870 Meet With Nagpur EscortsCall Girls in Nagpur Suman Call 7001035870 Meet With Nagpur Escorts
Call Girls in Nagpur Suman Call 7001035870 Meet With Nagpur Escorts
 

dastgerdi2020 (2).pdf

  • 1. Estimation of underground interwell connectivity: A data-driven technology Ehsan Jafari Dastgerdia , Ali Shabanib, *, Davood Zivarc, *, Hamid Reza Jahangirid a Department of Petroleum Engineering, Khazar University, Baku, Azerbaijan b Department of Chemical and Petroleum Engineering, Sharif University of Technology, Tehran, Iran c Special Core Analysis Lab, Petro Makhzan Kav (PMK), Tehran, Iran d Department of Chemical and Petroleum Engineering, Iran University of Science and Technology, Tehran, Iran A R T I C L E I N F O Article History: Received 21 August 2020 Revised 17 October 2020 Accepted 9 November 2020 Available online xxx A B S T R A C T Water injection into petroleum reservoirs is widely performed around the world for enhancing oil recovery. Understanding the underground fluid path is an important factor in improving reservoir performance under waterflooding operation. This may be used to optimize subsequent oil recovery by changing injection pat- terns, assignment of well priorities in operations, recompletion of wells, targeting infill drilling, and reduce the need for expensive surveillance activities. Most of the hydrocarbon reservoirs are equipped with sensors that measure the flow rate, pressure, and temperature in the wellbores continuously. Valuable and useful information about the interwell connections can be obtained from the measured data. In the present paper, a novel data-driven approach is developed that contributes to estimating the underground interwell connec- tions through analyzing and processing the injection and production data of a petroleum reservoir. This novel data-driven technology is named Detection of Events (DoE) and combines with the Capacitance Resistance Model (CRM) to quantify the connections between injectors and producers. The proposed approach has been applied to both synthetic and field cases. The results of the method show a good agreement between CRM and DoE for the field case (75%), and present a good insight into the better understanding of the underground connections. © 2020 Published by Elsevier B.V. on behalf of Taiwan Institute of Chemical Engineers. Keywords: Data-driven approach Capacitance resistance model (CRM) Detection of events (DoE) Interwell connections 1. Introduction Data acquisition systems are used in the petroleum industry to monitor the field operation continuously. The gathered data can be used for the well-log prediction, risk analysis, reservoir management, drilling engineering, and uncertainty analysis through applying dif- ferent data-driven methods [1,3,4,24,25]. Injection of water into the oil reservoirs for enhancing oil recovery is widely used around the world since the early days of the petroleum industry. Increasing the oil recovery by waterflooding can prolong the life of a petroleum res- ervoir. Waterflooding needs detailed management and continuous monitoring. During a waterflooding operation, understanding the underground fluid paths is necessary information for the efficient oil displacement, determination of infill drilling, and injection-produc- tion pattern. This information depends on the quality of the connec- tion between wells of the reservoir. There are different methods to quantify the underground connection among the wells including tra- ditional simulators, tracers, and data-driven approaches [2,8,12,15,17,30,31,33]. Traditional simulators have been used by the petroleum companies as a standard reservoir evaluation method. However, due to the limitations of the modeling techniques, reservoir simulators might not fully represent the true reservoir performance. Besides, maintaining a truly representative full reservoir simulation model for the routine and frequent analysis of waterflood perfor- mance prediction is a challenging process. Nowadays, using tracers is not common in the petroleum industry, because it takes a long time for the tracers to breakthrough, which makes the procedure time and money consuming. Over the past few years, several data-driven approaches based on available reservoir data have been introduced and developed for pre- diction of the production rates and the connections between wells [5,23,37]. Some researchers used a data-driven interwell numerical simulation model (INSIM) for modeling waterflooding [9,39]. Other researchers used spearman rank correlations to find the connectivi- ties between injectors and producers of a field [10,27,28,32]. Recently, capacitance resistance models were introduced for water- flooding studies and were used widely by companies and researchers for estimating the connections between the wells of a reservoir [6,7,16,1821,23,29,36,38]. This intuitive and numerically fast tech- nique utilizes only the injection and production data. It exhibits the * Corresponding authors. E-mail addresses: alishabani.sharif@gmail.com (A. Shabani), Davood_zivar@yahoo. com (D. Zivar). https://doi.org/10.1016/j.jtice.2020.11.008 1876-1070/© 2020 Published by Elsevier B.V. on behalf of Taiwan Institute of Chemical Engineers. ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 Journal of the Taiwan Institute of Chemical Engineers 000 (2020) 19 Contents lists available at ScienceDirect Journal of the Taiwan Institute of Chemical Engineers journal homepage: www.elsevier.com/locate/jtice
  • 2. strengths of the injector-producer connections without geological information [35]. Izgec and Kabir [11] introduced the CRM as a reli- able method that its results have a good agreement with traditional reservoir simulations [11]. Moreno [22] and Zhao et al. [39] tried to find the connectivity in multi-layer systems [22,39]. Shabani et al. [31] used CRM for understanding the injectors’ efficiency and analyz- ing the interwell connections. The neural network also can be used for prediction of the production rate and the interwell connections [5,26,34]. Artun [5] used weight factors achieved from the neural net- work to define the quality of connections between injectors and pro- ducers of a synthetic reservoir [5]. As the neural networks are not the physics based methods, they cannot estimate the interwell connec- tions as efficient as CRMs [13,14,26]. Jahangiri et al. [12] developed and applied Top-Down WaterFlood (TDWF) diagnostics and optimi- zation method for the North Sea and used it as a “virtual” interwell tracer. As demonstrated for the North Sea field, the efficiency of the most recent injector in the field will be understood without the need to await the results of a tracer program. In this paper, we present a novel approach named Detection of Events (DoE), based on relating the injection and production data of a real reservoir. The proposed approach complements CRM by data mining techniques to quantify and validate the CRM injector-pro- ducer connection estimates. This has implications when it comes to prioritizing water injectors when optimizing injection under various water constraints (e.g. individual well, well pad and/or total vol- umes). The proposed method is tested for the synthetic and real res- ervoirs, then the results are compared with the results of CRM. The structure of the paper is as follows: Section two presents the mathe- matical methodology. Results and discussions of the proposed method are presented in section three, and section four presents the conclusions. 2. Mathematical methodology Understanding the underground interwell connections presents valuable information about the performance of a waterflooding operation. For estimating the interwell connections in the present study, both analytical capacitance resistance model (CRM) and detec- tion of events (DoE) approach are used. These two complementary data-driven methods, which do not require any geological or dynamic reservoir models, quantify the underground connections between the injectors and the producers. 2.1. Capacitance resistance model (CRM) A detailed methodology of the capacitance resistance model is presented in this section. Capacitance resistance models are used in most engineering fields including electrical, chemical, petroleum, etc. In the petroleum industry, CRM enables a reservoir engineer to esti- mate production rates of producers, and understand the under- ground connection between each pair of injector-producer through history matching of the production data [22,29]. By assuming different control volumes, different forms of the CRM can be developed, including CRMT, CRMP, and CRMIP [29]. Among different forms of CRM, the CRMP (CRM Producer) can be used for a system of multi-well. It can estimate the production rate of producers and the quality of the underground connection between each pair of injector-producer. By assuming a constant productivity index (J), and a step variation of the injection rate and production pressure in the time interval of Dtk, the production rate of each producer can be esti- mated by the underlying equation [29]: qjðtkÞ ¼ qjðtk1Þe Dtk tj þ 1e Dtk tj X Ninj i¼1 fijIi k Jjtj DPk w;j Dtk # ð1Þ In this equation, qj, tj, and Jj are the production rate, the time con- stant, and the productivity index of the jth producer respectively. fij is the connection factor between the ith injector and the jth producer, Ii is the injection rate of the ith injector, and Njnj is the number of active injectors. The connection factor (fij), time constant (tj), productivity index (Jj), and the primary production rate of the producer (q0) are estimated using the optimization process. It is obvious that the sum of the connection factors between an injector and other producers must be less than one as presented below (Npro is the number of active producers in the reservoir): X Npro j¼1 fij1 ð2Þ The calculated connection factors (fij) create a matrix and are used for identification of the quality of the connection between each pair of injector-producer. The high value of the connection factor means that there is a stronger underground connection between the injector and producer. According to the magnitude of the connection factor between each pair of injector-producer, the quality of the connection is defined as presented in Table 1. 2.2. Detection of events (DoE) DoE is a quantitative method for identifying the flow path between injectors and producers of a reservoir. This method is based on finding relations between injection and production events. A flow rate or a pressure event is defined as the certain changes in that flow rate or pressure in a certain period of time. For determination of the Nomenclature AP Agreement parameter fij Fraction of the rate of injector i that is towards to the producer j Ii Injection rate of the ith injector (STB/Day) Jj Productivity index of the jth producer (STB/Day.psi) Ninj Number of injectors Npro Number of producers ng Number of the elements in the green squares of the Boston matrix no Number of the elements in the orange squares of the Boston matrix nr Number of the elements in the red squares of the Boston matrix Pw Wellbore pressure (psi) Q Flow rate (STB/Day) qj Production rate of the jth producer (STB/Day) Scoreij Score of the connection between the injector i and the producer j in DoE Tmin Minimum time lag for an event of an injector to be seen in a producer (Days) Tmax Maximum time lag for an event of an injector to be seen in a producer (Days) T Time (Days) a Constant variation derivate (STB/Day2 or psi/Day) b Minimum event variation Table 1 Quality of the connections according to the connection factor matrix in CRM. Amount of the connection factor Quality of the connection fij 0.2 Weak 0.2 fij 0.5 Medium 0.5 fij Strong ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 2 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
  • 3. quality of connection between each pair of injector-producer, DoE detects injector and producer events, then it identifies that every one of the producer events is due to changes in producer choke valve or caused by the certain injector event. According to the number of evi- dences that represent the connection between each pair of injector- producer, an interwell connection will be assumed for these wells. In order to discriminate between non-connection events (trig- gered by an operator intervention at the surface) and connection events (due to the reservoir response to other inputs), evidence for operator action is required. When the operator changes the choke valve size, the flow rate and wellbore pressure change simulta- neously. These variations lead to non-connection events that are not representative of a connection between wells. DoE requires injection rates of the injectors for identification of event start date and dura- tion. But in producers, DoE requires flow rate and wellbore pressure in order to check whether the events are made by a change in the choke valve or not. DoE is implemented in 3 steps including, 1) detection and evalua- tion of the events in the injection rates of the injection wells, 2) detection of the events in the production rates and bottom-hole pres- sure (BHP) of the production wells, and 3) comparison and evaluation of the event in order to find the connectivity and association between each pair of injector-producer. The steps of a DoE algorithm are pre- sented below: 2.2.1. Detection and evaluation of the events in the injectors In the first step, the events of injection rates are identified. The event of an injector rate means a sharp change in injection rate, shut- in of the injector, or its constant flow rate (during a certain period of time). For an injection flow rate, four types of events exist, including Increases, Decreases, No flow, and Smooth. In Table 2 and Fig. 1, four types of these events have been presented. No flow events are detected easily from the injection data. In order to detect other types of events, it is required to define a crite- rion (based on derivate of flow rate), to find out whether the flow rate is increasing, decreasing, or constant. For this purpose, it is required to define a parameter named constant variation derivate (a). A constant variation derivate is a constant number (derivate) for determining types of changes in the wellbore flow rate. The slope of the flow rate diagram versus time (@Q @T ) is compared with the value of a in order to recognize the type of events in each day, as presented in Table 3. In Table 3,Q is the flow rate and T is the time. After finding the type of events in each day, the same and sequen- tial daily events are merged. For example, 2 sequential increasing events in each day are merged to have an increasing event that lasts for 2 days. Then, in order to find the considerable Increasing and Decreasing events, the absolute change of flow rate to its primary value is compared with a parameter named minimum event variation (b) as presented in Table 4. Those events that their absolute variation is less than b are excluded from the Increasing and Decreasing events and allocated to the Smooth events. 2.2.2. Detection of the events in the production rate and pressure of the production wells For each producer, daily events of production rate and wellbore pressure are detected using the method presented in the previous section (same procedure as the injection flow rate). For daily events of wellbore pressures, the value of constant variation derivate (a) is different from the one for flow rates. No merging will be done for the events of production wells, and there is only daily events for the pro- duction rate and wellbore pressure. This is because these daily events are required to detect the events which are created by the operator (changing choke valve size). For instance, if the operator is opening the choke valve (Increasing event of the production rate in the pro- ducer), the pressure of the producer is decreasing so rapidly (Decreasing event of the pressure in the producer). This escalation in the production rate of the producer is not representative of an inter- well connection because it is not created by increasing the injection rate of other injectors. Fig. 2 shows an example of a change in choke valve size. The rate of the injector i is decreasing on day 62, and it seems that with 5 days lag time the flow rate of producer j is decreasing too (due to the injec- tor i). But, the pressure of the producer j is increasing. This means that the reduction in the producer rate is due to the choke valve closing by the operator, and these events are not representative of the connec- tion between injector i and producer j. 2.2.3. Comparison and evaluation of the events to find the connection between each pair of injector-producer As discussed in the previous section, although some events in the production wells are created by the operator, there are some other events in the production wells that have been created by the injec- tion wells as presented in Table 5. For instance, as can be observed in Fig. 3, the production rate event in the producer j is increasing (on day 110), meanwhile the pressure event of the producer is constant. If this increment in the producer rate was due to the opening of the choke valve, the pressure of the producer would be decreasing so rapidly which is not the case here. This means that an external factor is supporting the producer. In this situation, if the injection rate event in the injector (with a proper lag time) is increasing too, this pair of injector-producer will be assigned one score. As shown in Fig. 3, there is an increasing event in the injector i on the 100th day, so this pair of injector-producer will be assigned one score in DoE. The minimum lag time (Tmin) and the maximum lag time (Tmax) after the beginning of the injector event are selected as the boundaries for evaluation of producer events. Only the production rate and pressure events that their start time is in this timespan will be considered for scoring. The values of Tmin and Tmin depends on the size of the reservoir, and are as the input parameters determined by the user. In Fig. 4, two sequences of events that show the connection and non-connection between one pair of injector- producer (for a real case), have been presented. After surveying all of the events between each pair of injector- producer, a score matrix with Ninj columns and Npro rows is produced. Then, according to the 10 percentile and 50 percentile of the score matrix, the quality of the connection is determined as presented in Table 6. 3. Results and discussions For analysis the correctness of the proposed method, both syn- thetic and real waterflooding cases were studied in this section. For each case, CRM and DoE are run to find the connection between wells. Then, the estimated interwell connections of both methods are compared and discussed. 3.1. Synthetic case A simple synthetic waterflooding case was used in this section to perform a detailed analysis of the results of the DoE. The synthetic case is a homogeneous 1D reservoir that has been flooded for 100 Days. It consists of only one injection well and two production wells as presented in Fig. 5. For simplicity of the synthetic case, the Table 2 Events of the flow rate. Event name Definition Increasing An increase in flow rate Decreasing A decrease in flow rate No flow Shut-in well Smooth Flow rate is almost constant ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 3
  • 4. bottom-hole pressure of both producers is constant during water- flood operation, so the last term of Eq. (1) in CRM and the pressure events in DoE can be neglected. 3.1.1. Results of CRM for the synthetic case As was discussed in Section 2.1, the connection factors of CRMP (producer base representation of the CRM) can be used to find the quality of the connection between injectors and producers. Hence, this model is run for finding the interwell connections in the syn- thetic reservoir. History matching of injection and production data leads to a connection factor matrix (fij) with Ninj columns and Npro rows as presented in Table 7. Fig. 6 presents the simulated (from CRMP) and observed total (oil+water) production rate of the syn- thetic reservoir. According to Table 7, the connection factor between injector 1 and producer 1 (I1 P1) is 0.65, and for injector 1 and pro- ducer 2 (I1 P2) is 0.35. It means that there is a strong connection for the first pair and a medium connection for the second one (according to Table 1). This is because of the producer 1 closeness to the injector 1 in comparison with the producer 2. Fig. 1. Different types of events for the flow rate. Table 3 Daily event detection. Criterion Event type @Q @T a Increasing event @Q @T a Decreasing event a @Q @T a Smooth event Table 4 Detection of Smooth events. Criterion Event type j Q2Q1 Q1 jb Increasing or Decreasing event j Q2Q1 Q1 j b Smooth event Fig. 2. Events those are not representative of the connection between injector i and producer j. Table 5 Events that present the connection between a pair of injector-producer. Production pressure event Production rate event Injection rate event Increasing Increasing Increasing Smooth Increasing Increasing Increasing Smooth Increasing Decreasing Smooth Decreasing Decreasing Decreasing Decreasing Smooth Increasing Smooth Decreasing Decreasing No flow Decreasing Smooth No flow Fig. 3. Events that are representative of the connection between injector i and pro- ducer j. ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 4 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
  • 5. 3.1.2. Results of DoE for the synthetic case According to the discussions in the DoE methodology, some con- stants are needed for running a DoE. For the synthetic case, a is selected 50 (STB/Day2 ), b is 0.2. Tmin and Tmax are selected according to the size of the injectors events (Tmin = 0 Days and Tmax = 5 Days). Fig. 7 presents the events of the injector 1 versus the events of pro- ducers 1 and 2. Also, the details of these events have been presented in Table 8 and Table 9. Note that for the simplicity smooth events have not been reported here, because they don’t affect the connec- tions in this case. For the big data containing a huge number of events, a code can be developed for the detecting and scoring the events as described in Section 2.2. As presented in Fig. 7, there are 2 Increasing-Decreasing events in the injector. The magnitude of the first Increasing-Decreasing events are too large, so they create Increasing-Decreasing events in both producers. Hence, according to the size of these events (duration), the score of the first events are allocated to both pair of injector-pro- ducer (see Table 9). The second Increasing-Decreasing events are small, so they cannot be detected in producer 2 (a 50), and the sec- ond producer receives no score from these events. The score matrix (achieved from DoE) for the synthetic case is pre- sented in Table 10. According to the 50 percentile of the score matrix (see Table 6), producer 1 shows a strong connection with injector 1, while a medium connection is observed between producer 2 and the injector. Therefore, both of CRMP and DoE are in agreement and pres- ent the same results about the connections of the synthetic case. Fig. 8 presents the connection map of DoE and CRMP for the synthetic reservoir. In this figure, different connection strengths are shown by the line width. A thick line represents a relatively strong connection, and a medium line represents a medium connection. 3.2. Real case (field case) In this section, the injection and production data of a real res- ervoir have been studied. Workflows are implemented in this res- ervoir to continuously improve their performance. In the real case, it is important to highlight key assumptions to ensure higher confidence in results, especially in the application of CRM. One requirement is the judicious selection of the time interval for the study. The selected data should include a minimum of 120 days of injection and production (to ensure that the wells have affected each other), with a relatively constant number of the active injector and producer wells. The studied reservoir consists of several production and injec- tion wells, but only for 250 days, the number of active wells in this field is constant (no new well is drilled or perforated, and there are 6 active producers and 4 active injectors). So, we use the injection and production data of this specified timespan for DoE and CRM analyses. The schematic of the real reservoir has been presented in Fig. 9. 3.2.1. Results of CRM for the real case Similar to the synthetic case, CRMP is used for finding the inter- well connections in the real reservoir, and then the results are com- pared with the results of the DoE method. Fig. 10 shows the simulated (from CRMP) and observed total (oil+water) production rate of the entire reservoir. According to the high value of R-squared (0.916) presented in Fig. 10, the results of the simulated production rate (CRM) show a good agreement with observation data. Fig. 11 shows the connectivity map of the field achieved by CRM (according to Table 1). As presented in Fig. 11, CRMP estimates that there is a strong underground connection between injector 1 and producer 5 (I1 P5), and injector 3 and producer 3 (I3 P3). Fig. 4. Two sequences of connection and non-connection events in a real case. Table 6 Quality of the connections according to the score matrix created by DoE. The value of the score Quality of the connection Scoreij P10 Weak P10 Scoreij P50 Medium P50 Scoreij Strong Fig. 5. Schematic of the synthetic reservoir (red circles are the producers and the blue triangle is the injector).(For interpretation of the references to color in this figure leg- end, the reader is referred to the web version of this article.) Table 7 Connection fac- tor matrix for the synthetic case. fij I1 P1 0.65 P2 0.35 ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 5
  • 6. 3.2.2. Results of DoE for the real case Sensible values must be selected for each one of the DoE parame- ters. In the real case, a is 300 STB/Day2 for the flow rate events, and 50 psi/Day for the pressure events. b is 0.1, Tmin is 2 Days and Tmax is 8 Days. In Fig. 12, the results of the connectivity map of the DoE for the entire field have been shown. As presented in Fig. 12, DoE estimates that there is a strong underground connection between injector 2 and producer 3 (I2 P3), injector 3 and producer 3 (I3 P3), and injec- tor 3 and producer 4 (I3 P4). Injector 3 and producer 3 (I3 P3), was reported to have a strong connection by both DoE and CRM. DoE Fig. 6. Total production rate of the synthetic reservoir. Fig. 7. Events of the injector 1 versus the events of producers 1and 2 in the synthetic case. Table 8 Events of the injector 1 in the synthetic case. No. Event type Duration (Days) a b E1 Increasing 914 600 3 E2 Decreasing 1419 600 3 E3 Increasing 4951 200 1 E4 Decreasing 5156 200 1 Table 9 Events of the producers in the synthetic case. Well No. Event type Duration a b Connection Score Total score P1 E5 Increasing 915 195 1.74 Yes 6 28 E6 Decreasing 1526 103.7 0.61 Yes 11 E7 Increasing 4955 67.3 0.62 Yes 6 E8 Decreasing 5661 55.9 0.26 Yes 5 P2 E9 Increasing 1015 104.5 1.11 Yes 5 11 E10 Decreasing 1622 77 0.46 Yes 6 E11 Increasing 5055 37.7 0.51 No 0 E12 Decreasing 5661 26.4 0.23 No 0 Table 10 Score matrix for the synthetic case. Scoreij I1 P1 28 P2 11 ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 6 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
  • 7. considers the connection between injector 2 and producer 3 (I2 P3) as a strong connection, while CRM expresses that this connection is medium. CRM estimates that there is a strong connection between injector 1 and producer 5 (I1 P5), while DoE estimates that this is a weak connection. It is time-consuming for analyzing one by one of the estimated connections for a real reservoir. Hence, the Boston matrix has been used in order to have a quick and fair comparison between the con- nections calculated by CRM and DoE (see Table 11). In the Boston matrix, as the higher number of elements are on the main diagonal, the results are in a better agreement. The green squares show the number of pairs that have been estimated similarly by CRM and DoE (for example, 14 pairs with a weak connection are recognized by both CRM and DoE). The red squares show where the two models dis- agree with each other (for example, there is a pair that CRM considers Fig. 8. Connection map of the synthetic reservoir (estimated by CRMP and DoE). Fig. 9. Schematic of the real reservoir. Fig. 10. Total production rate of the entire reservoir. Fig. 11. Underground connection map of the real reservoir (estimated by CRMP; a thick line represents a relatively strong connection, a medium line represents a medium connection, and a thin dashed line signifies a weak connection). Fig. 12. Underground connection map of the reservoir (estimated by DoE). Table 11 Results of CRM and DoE in Boston matrix. Connection Weak (CRM) Medium (CRM) Strong (CRM) Weak (DoE) 14 1 1 Medium (DoE) 4 1 0 Strong (DoE) 1 1 1 ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 7
  • 8. as a strong connection, but DoE considers weak. Also there is another pair that DoE considers a strong connection while CRM considers weak). The orange squares show the connection strengths that are in slight disagreement (CRM considers a connection medium strength, while DoE considers the connection to be a weak connection). A criterion was developed to express the agreement between two approaches in the Boston matrix. In the developed criterion, each square with specific color awards a specific point. For example, 3 points are awarded to the pairs which are located in the green squares of the Boston matrix, 1 point to the pairs of the orange squares, and no points are awarded to the pairs located in the red squares. Based on this criterion, it can be inferred that what percent the estimated connection strength by DoE confirms the estimated connection strength by CRM. The agreement parameter (AP) between 2 methods is calculated using the equation presented below: AP ¼ 3ng þ no 3ðng þ no þ nrÞ 100 ð3Þ where, ng is the number of pairs in the green squares, no is the num- ber of pairs in the orange squares, and nr is the number of pairs in the red squares. According to the Table 11, 16, pairs are in the green squares, 6 pairs in the orange squares, and 2 pairs in the red region. Consequently, there is 75% agreement between CRM and DoE in the studied case. Another advantage of using this method is the confidence about the true underground interwell connections. For example, a strong connection has been estimated between injector 3 and producer 3 (I3 P3) in both methods. So, this strong connection can be reliable. CRM estimates there is a strong connection between injector 1 and producer 5 (I1 P5), while DoE estimates that this connection is weak. Hence, the connection between this pair is not trustable and requires more surveillance techniques. According to the long distance between injector 1 and producer 5 (see Fig. 12), the distrust about this pair is not weird. The fact that events on an injection well are correlated in time with events on a separate producer well does not provide a predictive mechanism for how future production from that well might usefully vary. However, a strong association is still indica- tive of a communication path between the wells and this insight could be great information to develop the reservoir in the future. 3.2.3. Comparison of DoE and ANN Artificial neural networks (ANNs) also is used as a data-driven method for estimating the underground interwell connections [5]. In the present section, a comparison is performed between the ANN and DoE approaches in terms of accuracy, data requirements, exper- tise requirements, training algorithm, and processing times. To do so, ANN was applied for the real case (field case) using the backpropaga- tion algorithm and 1 hidden layer. Then, the normalized ANN connec- tions compared with the CRM connections and presented in the form of the Boston matrix (Table 12). Accuracy: It is believed that CRM presents more reliable interwell connections in comparison with ANN, as CRM is a physics-based approach [13,14,26]. Therefore, the value of the agreement with CRM (Eq. (3)) is selected as the accuracy criteria in this section. Regarding Table 12 and using Eq. (3), there is a 68% agreement between the obtained result from CRM and ANN. As the agreement parameter for CRM and DoE was 75%, it can be inferred that the DoE is more consis- tent with the CRM (more precise) in this studied case. Data requirements: In terms of required data, both DoE and ANN required the same set of data (the injection and production data). Expertise requirements: When it comes to working with ANNs, deep knowledge of the ANN theory and training parameters is cru- cial. On the other hand, DoE does not requires any optimization algo- rithm and only needs the proper identification of the parameters used for the detection of the events. Training algorithm: No training algorithm is required for the DoE, while the performance of the ANN strongly depends on the training algorithms. Processing time: Both DoE and ANN methods are fast, where the CPU time is less than a few minutes. According to all these terms, it seems that DoE is more efficient than ANN. DoE is in a better agreement with CRM and is easier to apply. However, DoE only can be used for estimation of the interwell connections and cannot predict the production rate of the reservoir. 4. Conclusions In this paper, a novel data-driven approach (DoE) was proposed in order to find the underground connections between injection and production wells. The proposed method is based on finding the impact of injection rate fluctuations on the production rates and scor- ing the relevant impacts. Then, by establishing an effective calibration between CRM and DoE, the confidence about the estimated under- ground connections were analyzed. The proposed method was suc- cessfully tested for both synthetic and real reservoirs, and the estimations of these two methods (CRM and DoE) showed a good agreement with each other (75%). The proposed method can be used by reservoir engineers for understanding the underground path of the injected fluid, and deci- sion-making about the future. Although there is a high degree of sim- ilarity between DoE and CRM, it is highly recommended to integrate this methodology with other existing surveillance techniques, such as tracer, streamline modeling, and 4D seismic to evaluate and moni- tor the waterflooding process more accurately. Declaration of Competing Interest None. References [1] Adedigba SA, Oloruntobi O, Khan F, Butt S. ''Data-driven dynamic risk analysis of offshore drilling operations. J Petrol Sci Eng 2018;165:444–52. [2] Akin S. Analysis of tracer tests with simple spreadsheet models. Comput Geosci 2001;27(2):171–8. [3] Amar MN, Ghahfarokhi AJ, Zeraibi N. ''Predicting thermal conductivity of carbon dioxide using group of data-driven models. J Taiwan Inst Chem Eng 2020:3489. [4] Amin MT, Khan F, Ahmed S, Imtiaz S. ''A novel datadriven methodology for fault detection and dynamic risk assessment. Can J Chem Eng 2020. [5] Artun E. Characterizing interwell connectivity in waterflooded reservoirs using data-driven and reduced-physics models: a comparative study. Neural Comput Appl 2017;28(7):1729–43. [6] Chitrala Y, Moreno C, Sondergeld C, Rai C. ''An experimental investigation into hydraulic fracture propagation under different applied stresses in tight sands using acoustic emissions. J Petrol Sci Eng 2013;108:151–61. [7] Eshraghi SE, Rasaei MR, Zendehboudi S. ''Optimization of miscible CO2 EOR and storage using heuristic methods combined with capacitance/resistance and Gentil fractional flow models. J Nat Gas Sci Eng 2016;32:304–18. [8] Goudarzi A, Delshad M, Sepehrnoori K. ''A chemical EOR benchmark study of dif- ferent reservoir simulators. Comput Geosci 2016;94:96–109. [9] Guo Z, Reynolds AC, Zhao H. ''Waterflooding optimization with the INSIM-FT data-driven model. Comput Geosci 2018:1–17. [10] Heffer KJ, Fox RJ, McGill CA, Koutsabeloulis NC. ''Novel techniques show links between reservoir flow directionality, earth stress, fault structure and geome- chanical changes in mature waterfloods. SPE J 1997;2(02):91–8. [11] Izgec O, Kabir CS. ''Quantifying nonuniform aquifer strength at individual wells. SPE Reserv Evaluat Eng 2010;13(02):296–305. [12] Jahangiri HR, Adler C, Shirzadi S, Bailey R, Ziegel E, Chesher J, White M. A data- driven approach enhances conventional reservoir surveillance methods for waterflood performance management in the North Sea. SPE intelligent energy conference exhibition, society of petroleum engineers; 2014. Table 12 Results of CRM and ANN in the Boston matrix. Connection Weak (CRM) Medium (CRM) Strong (CRM) Weak (ANN) 12 2 1 Medium (ANN) 3 1 1 Strong (ANN) 2 1 1 ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 8 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19
  • 9. [13] Jamali A, Ettehadtavakkol A. ''Application of capacitance resistance models to determining interwell connectivity of large-scale mature oil fields. Petrol Explorat Dev 2017;44(1):132–8. [14] Jensen JL. Comment on “Characterizing interwell connectivity in waterflooded reservoirs using data-driven and reduced-physics models: a comparative study” by E. Artun.''. Neural Comput Appl 2017;28(7):1745–6. [15] Kaviani D, Jensen JL, Lake LW. ''Estimation of interwell connectivity in the case of unmeasured fluctuating bottomhole pressures. J Petrol Sci Eng 2012;90:79–95. [16] Kaviani D, Soroush M, Jensen JL. ''How accurate are capacitance model connectiv- ity estimates? J Petrol Sci Eng 2014;122:439–52. [17] Kaviani D, Valk o PP. ''Inferring interwell connectivity using multiwell productiv- ity index (MPI). J Petrol Sci Eng 2010;73(12):48–58. [18] Liang X. A simple model to infer interwell connectivity only from well-rate fluctu- ations in waterfloods. J Petrol Sci Eng 2010;70(12):35–43. [19] Mamghaderi A, Aminshahidy B, Bazargan H. ''Error behavior modeling in capaci- tance-resistance model: a promotion to fast, reliable proxy for reservoir perfor- mance prediction. J Nat Gas Sci Eng 2020:103228. [20] Mamghaderi A, Bastami A, Pourafshary P. ''Optimization of waterflooding perfor- mance in a layered reservoir using a combination of capacitance-resistive model and genetic algorithm method. J Energy Resour Technol 2013;135(1):013102. [21] Mirzayev M, Riazi N, Cronkwright D, Jensen JL, Pedersen PK. ''Determining well- to-well connectivity using a modified capacitance model, seismic, and geology for a Bakken Waterflood. J Petrol Sci Eng 2017;152:611–27. [22] Moreno GA. Multilayer capacitanceresistance model with dynamic connectivi- ties. J Petrol Sci Eng 2013;109:298–307. [23] Moreno GA, Lake LW. ''On the uncertainty of interwell connectivity estimations from the capacitance-resistance model. Petrol Sci 2014;11(2):265–71. [24] Nhat DM, Venkatesan R, Khan F. ''Data-driven Bayesian network model for early kick detection in industrial drilling process. Process Saf Environ Protect 2020. [25] Onalo D, Adedigba S, Khan F, James LA, Butt S. ''Data driven model for sonic well log prediction. J Petrol Sci Eng 2018;170:1022–37. [26] Panda M, Chopra A. An integrated approach to estimate well interactions. SPE India oil and gas conference and exhibition, society of petroleum engineers; 1998. [27] Refunjol BT. Reservoir characterization of north buck draw field based on tracer response and production/injection analysis. University of Texas at Austin; 1996. [28] Sant'Anna, P.J. ''Estimating injectivity and lateral autocorrelation in heteroge- neous media.''(1999). [29] Sayarpour M. Development and application of capacitance-resistive models to water/carbon dioxide floods. The University of Texas at Austin; 2008. [30] Sayarpour M, Zuluaga E, Kabir CS, Lake LW. ''The use of capacitanceresistance models for rapid estimation of waterflood performance and optimization. J Petrol Sci Eng 2009;69(3):227–38. [31] Shabani, A., M.S. Moosavi, D. Zivar and H.R. Jahangiri. ''Data-driven technique for analyzing the injector efficiency in a waterflooding operation.'' SIMULATION: 0037549720923386.(2020). [32] Soeriawinata T, Kelkar M. Reservoir management using production data. SPE mid- continent operations symposium; 1999. [33] Soroush M, Kaviani D, Jensen JL. ''Interwell connectivity evaluation in cases of changing skin and frequent production interruptions. J Petrol Sci Eng 2014;122:616–30. [34] Van SL, Chon BH. ''Effective prediction and management of a CO2 flooding process for enhancing oil recovery using artificial neural networks. J Energy Resour Tech- nol 2017;140(3) 032906-032906-032914. [35] Wang D, Li Y, Chen B, Hu Y, Li B, Gao D, Fu. B. Ensemble-based optimization of interwell connectivity in heterogeneous waterflooding reservoirs. J Nat Gas Sci Eng 2017;38:245–56. [36] Weber D, Edgar TF, Lake LW, Lasdon LS, Kawas S, Sayarpour M. Improvements in capacitance-resistive modeling and optimization of large scale reservoirs. SPE western regional meeting, society of petroleum engineers; 2009. [37] Xu T, Valocchi AJ. ''Data-driven methods to improve baseflow prediction of a regional groundwater model. Comput Geosci 2015;85:124–36. [38] Yousef AA, Gentil PH, Jensen JL, Lake LW. A capacitance model to infer interwell connectivity from production and injection rate fluctuations. SPE annual technical conference and exhibition, society of petroleum engineers; 2005. [39] Zhao H, Kang Z, Zhang X, Sun H, Cao L, Reynolds AC. ''A physics-based data-driven numerical model for reservoir history matching and prediction with a field appli- cation (associated discussion available as supporting information). SPE J 2016;21 (06):175. 2172194. ARTICLE IN PRESS JID: JTICE [m5G;November 19, 2020;9:37] Please cite this article as: E.J. Dastgerdi et al., Estimation of underground interwell connectivity: A data-driven technology, Journal of the Taiwan Institute of Chemical Engineers (2020), https://doi.org/10.1016/j.jtice.2020.11.008 E.J. Dastgerdi et al. / Journal of the Taiwan Institute of Chemical Engineers 00 (2020) 19 9