See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/365442089

A Study of Modern Decline Curve Analysis Models Based on Flow Regime

Identification

Article in Journal of University of Shanghai for Science and Technology · June 2022
DOI: 10.51201/JUSST/22/0640

CITATIONS READS
4 288

4 authors, including:

Ali Wahba Hamid Khattab

Suez University Suez University
8 PUBLICATIONS 40 CITATIONS 19 PUBLICATIONS 142 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ali Wahba on 16 November 2022.

The user has requested enhancement of the downloaded file.

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

A Study of Modern Decline Curve Analysis Models Based on Flow

Regime Identification
Wahba, A. M. *, Khattab, H. M., Tantawy, M. A., Gawish, A. A.
Petroleum Engineering Department, Faculty of Petroleum and Mining Engineering, Suez University, Suez,
*Corresponding author e-mail:
[email protected]

Abstract
Unconventional reservoirs, including tight and shale reservoirs, have become a
significant source of petroleum supply in the world. Exploitation of these reservoirs is a
challenge due to their very low permeability. Advanced technologies such as multi-stage hydraulic
fracturing or pad drilling are necessary to enhance their productivity. Production behavior of
unconventional reservoirs is characterized by long-term transient flow followed by boundary-dominated
flow. Various production data analysis methods have been developed to analyze this production
behavior; straight-line analysis methods are used to identify the flow-regime type while decline curve
analysis methods are used to predict the future production behavior. This paper summarized flow-
regime types which appear during well production in unconventional reservoirs and reviewed different
decline curve analysis methods used to simulate this production behavior. A comparative study was
conducting between the most used decline curve analysis models; Arps model, power-law exponential
model, T-model and Duong model to determine when to use each model. A software application was
written in python language to facilitate conducting this study. This software application is a powerful
tool and provides reliable results.

Keywords: Unconventional reservoirs, Flow regimes, Decline curve analysis, Field data.

1. Introduction

Unconventional resources are subsurface hydrocarbon-bearing rocks which are

characterized by very low permeability. Tight and shale formations are the two main types.
Drilling multi-fractured horizontal wells (MFHWs) and multiple horizontal wells from a
single pad are necessary as they make the reservoir [1-4]. The long horizontal section of the
well maximizes the contact area with the reservoir and fractures work as paths for
hydrocarbon to flow and as a result well productivity increases. Various decline curve analysis
(DCA) methods have been developed to simulate production behavior of wells in
unconventional reservoirs. Then, they can be used to predict the future production behavior.
Kanfar and Wattenbarger [5], Tan et al. [6], Mahmoud et al. [7], Ibrahim et al. [8] and Liang et
al. [9] reviewed and evaluated different DCA models. Production behavior of wells in
unconventional reservoirs shows long period of transient flow regime which is followed by
a period of boundary-dominated flow (BDF) regime. The transient flow may be linear,
bilinear or a combination of both flow-regime types depending on the reservoir, well and/or
fractures characteristics. A backflow period may occur before the transient flow due to
cleaning the well from the fracturing fluid. Many

Volume 24, Issue 6, June - 2022 134

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

straight-line analysis methods have been developed to identify these flow-regime types. This
paper summarized the different types of flow regime occurred in unconventional reservoirs.
The paper also reviewed different DCA models and introduced a comparative study between
the most used ones to show when to use each model according to the occurring flow-regime
types. Seven actual data sets were used in the comparative study and a software application
was written in python language to minimize computation time.

2. Flow-Regime Types

Flow regimes are characteristic flow patterns or geometries which result from fluid flow
dynamics in the reservoir. They are divided into two main types: transient and BDF. Transient
flow occurs when the reservoir act as an infinite system. It converts into BDF when the flow
disturbance reaches the boundaries. These flow regimes have different geometries such as
linear, bilinear, elliptical or pseudo-radial flow geometries; Table 1 describes these flow
geometries. Production of wells in unconventional reservoirs shows a sequence of different
flow-regime types [10-15]. The occurred flow-regimes sequence depends on the properties of
the reservoir, the well and/or the hydraulic fractures as shown in Figures 1-5.
Table 1 Description of flow-regime geometries

Volume 24, Issue 6, June - 2022 135

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 1 Flow-regimes sequence for a hydraulically fractured vertical well in a homogeneous

tight reservoir (Modified after Cinco-Ley and Samaniego-V, 1981) [10]

Figure 2 Flow-regimes sequence for a multi-fractured horizontal well in a homogeneous tight

reservoir (Modified after Chen and Raghavan, 1997; Raghavan et al., 1997) [11,12]

Volume 24, Issue 6, June - 2022 136

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 3 Flow-regimes sequence for a hydraulically fractured vertical well in a very tight
reservoir (modified after Wattenbarger et al., 1998) [13]

Figure 4 Flow-regimes sequence for a multiple transverse fracture horizontal well in shale gas
reservoir (Modified after Song and Ehlig-Economides, 2011) [14]

Volume 24, Issue 6, June - 2022 137

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 5 Illustration of the five flow regions for a slab matrix dual porosity linear shale gas
reservoir (Modified after Bello and Wattenbarger, 2010) [15]

These figures showed that the production behavior of wells in unconventional reservoirs
exhibits long-term transient flow followed by BDF. The transient flow may be linear, bilinear or
a combination of both types [10-15]. Transitional flow (elliptical and pseudo-radial) disappears
when using shortly spaced hydraulic fractures with long fracture half-length [11,12]. And in
practical, there is no flow from the outside of fractures tips [14]. Straight-line analysis methods
are used to identify the occurred flow-regime types. They are analogous to those used in
pressure transient analysis [16]. On log-log diagnostic plot of flow rate versus time, linear flow

Volume 24, Issue 6, June - 2022 138

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

appears as a straight-line with 1/2 slope while bilinear flow appears as a straight-line with 1/4
slope. Then, the straight-line bends downward as an indication of BDF occurrence.

3. Decline Curve Analysis Methods

Different production data analysis (PDA) methods have been developed for
conventional and unconventional reservoirs. Decline curve analysis methods are the simplest
PDA methods which require the least amount of data. They are also the most used methods to
predict the future production behavior of the well in conventional and unconventional
reservoirs. Ahmed [3] has called DCA methods used for unconventional reservoirs by modern
DCA methods to differentiate them from those used for conventional reservoirs. DCA methods
may be single models or combined models as summarized in the following sections.
3.1 Single Models
A single model involves one model which is used to simulate the whole production
behavior. In this paper, we reviewed the most popular eight models.
3.1.1 Arps Model
Based on definition of loss-ratio proposed by Johnson and Bollens [17] and production
decline behaviors concluded by Pirson [18], Arps [19] derived three different types for decline-
curve behavior as given in Table 2.
Table 2 Mathematical expressions of the three types of Arps decline curves

Exponential Decline Hyperbolic Decline Harmonic Decline

𝑏=0 0<𝑏<1 𝑏=1
𝐷𝑖 𝐷
𝐷(𝑡) = 𝐷𝑖 𝐷(𝑡) = 𝐷(𝑡) = 𝑖
1+𝑏 𝐷𝑖 𝑡 1+𝐷𝑖 𝑡
(Eq. (1)) (Eq. (2)) (Eq. (3))
𝑞𝑖 𝑞
𝑞(𝑡) = 𝑞𝑖 exp(−𝐷𝑖 𝑡) 𝑞(𝑡) = 1 𝑞(𝑡) = (1+𝐷𝑖
(1+𝑏𝐷𝑖 𝑡)𝑏 𝑖 𝑡)

(Eq. (4)) (Eq. (5)) (Eq. (6))

𝑄(𝑡) =
𝑞𝑖 −𝑞(𝑡) 𝑞𝑖 𝑞(𝑡) 1−𝑏 𝑞
𝑄(𝑡) = � 𝑖 � ln �
𝑞𝑖
�
𝐷𝑖 𝑄(𝑡) = � � �1 −� � � 𝐷𝑖 𝑞(𝑡)
𝐷𝑖 (1−𝑏) 𝑞𝑖
(Eq. (7)) (Eq. (8)) (Eq. (9))

Where:
𝑏 = Arps decline-curve exponent.
𝐷𝑖 = Initial decline rate, day -1.
𝐷(𝑡) = Decline rate at time t, day -1.
𝑡 = Time, day.
𝑞𝑖 = Initial flow rate at time 𝑡 = 0, MSTB/day or MMSCF/day.
𝑞(𝑡) = Flow rate at time t, MSTB/day or MMSCF/day.
𝑄(𝑡) = Cumulative production, MSTB or MMSCF.

Volume 24, Issue 6, June - 2022 139

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Arps model was firstly developed to simulate production behavior of conventional

reservoirs as it exhibits a long period of BDF regime. In contrast, production behavior of
unconventional reservoirs exhibits a long period of transient flow and Arps model fails to
simulate it. So, modern DCA models have been developed to simulate this transient flow. The
most popular modern DCA models are summarized in the following sections.
3.1.2 Power-Law Exponential (PLE) Model
Ilk et al. [20] developed the following expressions to simulate decline rate and
production rate for tight and shale reservoirs.

𝐷(𝑡) = 𝐷∞ + 𝐷1 𝑡 −(1−𝑛) (10)

𝑞(𝑡) = 𝑞�𝑖 exp�−𝐷∞ 𝑡 − 𝐷�𝑖 𝑡 𝑛 � (11)

Where:
𝑞(𝑡) = Flow rate at time 𝑡, MSTB/day or MMSCF/day.
𝑞�𝑖 = Rate intercept at 𝑡 = 0 [This parameter has a different interpretation than 𝑞𝑖 ].
𝐷(𝑡) = Decline rate at time 𝑡, day -1.
𝐷1 = Decline constant intercept at 𝑡 = 1 day.
𝐷∞ = Decline constant at infinite time.
𝐷�𝑖 = Decline constant [𝐷�𝑖 = 𝐷1 ⁄𝑛] [This parameter has a different interpretation than 𝐷𝑖 .
𝑡 = Time, day.
𝑛 = Time exponent.
3.1.3 Stretched Exponential Production Decline (SEPD) Model
Valkό [21] developed an empirical decline curve analysis model. This model is called
stretched exponential production decline as production rate obeys a decreasing exponential
relation.
𝑡 𝑛 (12)
𝑞 = 𝑞𝑖 exp �− � � �
𝜏

Where:
𝑞 = Produced rate in period, MSCF/month.
𝑞𝑖 = Maximum observed production rate, MSCF/month.
𝑡 = Number of periods, months.
𝑛 = Model exponent.
𝜏 = Characteristic number of periods.

The author also developed a mathematical expression for cumulative production.

𝑞𝑖 𝜏 1 1 𝑡 𝑛 (13)
𝑄= �Γ � � − Γ � , � � ��
𝑛 𝑛 𝑛 𝜏

Volume 24, Issue 6, June - 2022 140

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Where:
𝑄 = Cumulative production, MSCF.
𝛤� �
1 = Complete gamma function.
𝑛

𝛤 � , � � = Incomplete gamma function.

1 𝑡
𝑛 𝜏

3.1.4 T-Model
After Huang et al. [22] and Hu and Chen [23], Dou et al. [24] integrated T-model
equation and introduced the following cumulative-time and rate-time relations which describe
production behavior in unconventional reservoirs:

𝑁𝑝 = 𝑁𝑅 exp �
𝑎
𝑡 𝑏+1 � (14)
𝑏+1

𝑞(𝑡) = 𝑎 𝑁𝑅 𝑡 𝑏 exp �
𝑎
𝑡 𝑏+1 � (15)
𝑏+1
Where:
𝑞(𝑡) = Oil production rate, STB/day.
𝑁𝑝 = Cumulative oil production, STB.
𝑁𝑅 = Ultimate oil recovery at 𝑡 → ∞, STB.
𝑡 = Producing time, day.
𝑎 = Model’s constant.
𝑏 = Model’s constant.

3.1.5 Logistic Growth Model (LGM)

Clark et al. [25] derived two mathematical expressions by alternating the logistic growth
model used by Spencer and Coulombe [26]. These mathematical expressions can be used to
predict production of tight and unconventional reservoirs.
LGM cumulative-time relation:
𝐾𝑡𝑛 (16)
𝑄(𝑡) =
𝑎+𝑡 𝑛

LGM rate-time relation:

𝑞(𝑡) =
𝐾 𝑛 𝑎 𝑡 𝑛−1 (17)
(𝑎+𝑡 𝑛 ) 2

Where:
𝑄(𝑡) = Cumulative production.
𝑞(𝑡) = Production rate.
𝐾 = Carrying capacity or estimated ultimate recovery
𝑎 = Model constant.
𝑛 = Hyperbolic exponent.
𝑡 = Time.

Volume 24, Issue 6, June - 2022 141

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

3.1.6 Duong Model

To simulate production behavior of fractured wells in very tight and shale reservoirs,
Duong [27] developed the following approach:
𝑞
Duong time/material balance-time relation , = 𝑎 𝑡 −1 (18)
𝐺𝑃
𝑞
Duong modified time/material-balance-time , = 𝑎 𝑡 −𝑚 (19)
𝐺𝑃
relation
Duong rate-time relation , 𝑞
= 𝑡 − 𝑚 exp �
𝑎
(𝑡 1−𝑚 − 1)� (20)
𝑞1 1−𝑚

Duong cumulative-time relation , 𝐺 = 𝑞1 exp � 𝑎

(𝑡 1−𝑚 − 1)� (21)
𝑃 𝑎 1−𝑚
Where:
𝑞 = Production rate, MMSCF/day.
𝑞1 = Production rate at 𝑡 = 1 day, MMSCF/day.
𝐺𝑃 = Cumulative gas production, MMSCF.
𝑡 = Time, day.
𝑎 = Intercept constant defined by Eq. (19).
𝑚 = Slope defined by Eq. (19).
3.1.7 YM-SEPD Model
Yu and Miocevic [28] modified the methodology of using SEPD model in simulating
production data. This modification is to overcome non-uniqueness problem of determining
SEPD model’s parameters. They proposed plotting 𝑙𝑛(𝑞𝑖 ⁄𝑞(𝑡)) versus 𝑡 on a log-log scale and
determining 𝑛 and 𝜏 parameters according to the following relation:

𝑙𝑛 �
𝑞𝑖
� = 𝜏− 𝑛 𝑡𝑛 (22)
𝑞(𝑡)

3.1.8 Extended Exponential Decline Curve (EED)

After Fetkovich [29], Zhang et al. [30] developed an empirical equation to describe the
behavior of 𝑎-parameter in Eq. (23). This equation enabled Fetkovich equation to simulate flow
regimes occurring in unconventional reservoirs such as transient, transitional, and boundary-
dominated flow regimes.
Fetkovich equation:
𝑞 = 𝑞𝑖 exp(− 𝑎 𝑡) (23)
Where:
𝑞 = Flow rate at time 𝑡.
𝑞𝑖 = Initial flow rate.
𝑎 = Nominal decline rate.

Volume 24, Issue 6, June - 2022 142

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Zhang et al. empirical equation:

𝑎 = 𝛽𝑙 + 𝛽𝑒 exp(− 𝑡 𝑛 ) (24)
Where:
𝛽𝑒 = Constant to account for the early (fully transient) period, which should be larger than 𝛽𝑙
as recommended.
𝛽𝑙 = Constant to account for the late-life (BDF) period.
𝑛 = Empirical exponent, with a recommended range of 0 to 0.7.
𝑡 = Time in months.
3.2 Combined Models
As production behavior of unconventional reservoirs exhibits a sequence of flow-
regimes, some authors recommended using a DCA model for each flow-regime type. The used
model may involve two or three models and hence it called combined model. Combined models
containing two models, also called dual models, are such as PLE + Arps (modified PLE) [31] and
Duong + Arps (modified Duong) [32,33]. The switch point of these models has no physical basis
and just assumed based on experience. Chen et al. [34] proposed a combined model by using
SEPD, Duong, and Arps. This model is more complex and has more undetermined coefficients
which make it harder to use. In this study, we ignored combined models as they aren’t basic
models in the strict sense, as no new models are created

4. Methodology and PDAS Application

Arps model, PLE model, T-model and Duong model are the most used DCA models in
petroleum industry. They are considered as modern DCA models except Arps model. In this
study, we compared between the accuracy of these four models in simulating the production
behavior of unconventional reservoirs and their ability to predict the future production
behavior. This study was based on flow-regime identification which helps to determine when to
use each model. Seven actual data sets were used in this study. Figure 6 shows five steps
followed in this study to analyze the data sets.

Quality Check Curve- Prediction of

Flow-Regime Prediction
of Production Fitting of Future Production
Identification Results
Data DCA Model Behavior

Figure 6 Flow chart of production data analysis.

The first step, quality check of production data, was used to remove the very scattered
data point to enhance the quality of the production data. Other steps were programmed in a
software application written in python language which we called production data analysis

Volume 24, Issue 6, June - 2022 143

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

software (PDAS) application. The PDAS application consists of seven tabs. In this study, we only
used four tabs, which are:
i. Identification tab: to identify flow-regime types occurred in the production behavior of
the well.
ii. Curve fitting tab: to calibrate parameters of DCA models to production data. Statistical
parameters such as coefficient of determination (R2) and root mean square error (RMSE)
are determined to show the accuracy of curve fitting.
iii. Future prediction tab: to predict the future production behavior.
iv. Results tab: to calculate values of time and cumulative production corresponding to
production rate at economic limit.
There four tabs were used, respectively, to perform steps 2-5 in the flow chart.

5. Results and Discussion

Seven data sets (5 gas cases and 2 oil cases) were used in the comparative study. They
were from unconventional reservoirs (tight and shale). Symbols used in this study are given in
Table 3. Figures 7-13 show the flow-regime identification of each case and Table 4 summarizes
results of reservoir type, flow-regimes sequence and time at start of BDF (t BDF ) for each case.
Backflow appears in cases 1, 4, 5, 6 and 7 where multi-stage hydraulic fracturing was applied.
According to the dimensionless fracture conductivity, bilinear flow appears in some cases and
disappears in the others. Linear flow occurs in all cases as it is the characteristic flow of
unconventional reservoirs. In all cases, flow disturbance reached the boundaries and
production was under the dominant of BDF regime in late-time stage. The time at start of BDF
(t BDF ) differs from case to another depending on the fracture spacing. As fracture spacing is
reduced, t BDF decreases.
Table 3 Symbols used in the comparative study of DCA Models

EUR Estimated ultimate recovery or cumulative production at economic limit

GP Cumulative gas production
NP Cumulative oil production
Q𝑔 Gas production rate
Q limit Production rate at economic limit
Q𝑜 Oil production rate
RMSE (GP ) RMSE for cumulative gas production
RMSE (NP ) RMSE for cumulative oil production
RMSE (Q 𝑔 ) RMSE for gas production rate
RMSE (Q o ) RMSE for oil production rate
R2 (GP ) R2 for cumulative gas production
R2 (NP ) R2 for cumulative oil production
R2 (Q 𝑔 ) R2 for gas production rate
R2 (Q o ) R2 for oil production rate
t Producing time
t limit Time at economic limit

Volume 24, Issue 6, June - 2022 144

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 7 The identification of flow-regimes sequence for case 1

Figure 8 The identification of flow-regimes sequence for case 2

Volume 24, Issue 6, June - 2022 145

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 9 The identification of flow-regimes sequence for case 3

Figure 10 The identification of flow-regimes sequence for case 4

Volume 24, Issue 6, June - 2022 146

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 11 The identification of flow-regimes sequence for case 5

Figure 12 The identification of flow-regimes sequence for case 6

Volume 24, Issue 6, June - 2022 147

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 13 The identification of flow-regimes sequence for case 7

Table 4 Summary of flow-regime identification for the seven cases

Case Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7

Unconventional Tight Tight Shale Shale Eagle Eagle
Reservoir type
dry gas gas gas gas gas Ford oil Ford oil
Flow- Backflow  - -    
regimes Bilinear flow - - -   - 
sequence Linear flow       
BDF       
𝐭 𝐁𝐃𝐅 , day 86 5730 5294 143 223 85 190

Parameters of each DCA model are calibrated to the production data of each case.
Calibrated parameters are used to check the curve-fit accuracy through determining values of
R2 and RMSE. Better accuracy is achieved when R2 value is close to 1 and RMSE value is low.
Tables 5-11 summarizes curve fitting parameters and statistical values of DCA models for the
seven cases. As expected, Arps model could simulate the production behavior of cases 1, 4, 5, 6
and 7 where BDF started after a short period of transient flow and most of the production data
falls in BDF period. Cases 2 and 3 show long-term transient linear flow and as a result the curve-
fit accuracy of Arps model decreases. Also, PLE model couldn’t simulate the production
behavior of these two cases as it basically consists of two segments; one to simulate early-time
stage flow period and the other to simulate late-time stage flow period and there is a smooth
transition between the two flow periods. So, PLE model could simulate cases 1, 4, 5, 6 and 7
with high accuracy where flow-regime type changes from backflow to BDF with linear and/or

Volume 24, Issue 6, June - 2022 148

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

bilinear flow in between. The PLE model records the highest accuracy for case 5 as the flow-
regime smoothly changes from type to another and backflow period is negligible. T and Duong
models can be used interchangeably as they have similar mathematical expressions and behave
with the same manner. They record nearly the same curve-fit accuracy for most of the cases.
Their highest curve-fit accuracy is recorded for cases 1, 2, 5 and 6. In this study, we modified
the calibration method of T and Duong models. We only calibrated the parameters of these
models to the whole production data except the part of backflow period. This modification
enhanced the curve-fit accuracy of these models.
Table 5 Curve fitting parameters and statistical values of DCA models for case 1
Arps Model PLE Model T-Model Duong Model
𝑞𝑖 = 5.2806 𝑞𝑖 = 5.2316 𝑎 = 4.6717 𝑞1 = 0.2277
𝐷𝑖 = 0.0106 𝐷1 = 0.0140 𝑏 = -1.4648 𝑎 = 4.6717
𝑏 = 0.3612 𝐷∞ = 0.0000 𝐺𝑅 = 1150.2 𝑚 = 1.4648
𝑛 = 0.8612
𝑅2 (𝑄𝑔 ) = 0.9197 𝑅2 (𝑄𝑔 ) = 0.9119 𝑅2 (𝑄𝑔 ) = 0.9514 𝑅2 (𝑄𝑔 ) = 0.9529
𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.3382 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.3544 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.2632 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.2590
𝑅2 (𝐺𝑃 ) = 0.9970 𝑅2 (𝐺𝑃 ) = 0.9945 𝑅2 (𝐺𝑃 ) = 0.9949 𝑅2 (𝐺𝑃 ) = 0.9921
𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 8.7177 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 11.700 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 11.267 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 14.054

Table 6 Curve fitting parameters and statistical values of DCA models for case 2
Arps Model PLE Model T-Model Duong Model
𝑞𝑖 = 7.4415 𝑞𝑖 = - 𝑎 = 1.2488 𝑞1 = 15.440
𝐷𝑖 = 0.0024 𝐷1 = - 𝑏 = -1.1287 𝑎 = 1.2494
𝑏 = 1.0000 𝐷∞ = - 𝐺𝑅 = 2.2*105 𝑚 = 1.1288
𝑛 = -
𝑅2 (𝑄𝑔 ) = 0.8573 𝑅2 (𝑄𝑔 ) = - 𝑅2 (𝑄𝑔 ) = 0.9612 𝑅2 (𝑄𝑔 ) = 0.9688
𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.3121 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = - 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.1628 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.1459
2 2 2 2
𝑅 (𝐺𝑃 ) = 0.8661 𝑅 (𝐺𝑃 ) = - 𝑅 (𝐺𝑃 ) = 0.9869 𝑅 (𝐺𝑃 ) = 0.9387
𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 1191.9 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = - 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 373.09 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 806.61

Table 7 Curve fitting parameters and statistical values of DCA models for case 3
Arps Model PLE Model T-Model Duong Model
𝑞𝑖 = 1.5061 𝑞𝑖 = - 𝑎 = 1.5892 𝑞1 = 2.0171
𝐷𝑖 = 0.0027 𝐷1 = - 𝑏 = -1.1772 𝑎 = 1.5899
𝑏 = 1.0000 𝐷∞ = - 𝐺𝑅 = 1.2*104 𝑚 = 1.1772
𝑛 = - -
𝑅2 (𝑄𝑔 ) = 0.8256 𝑅2 (𝑄𝑔 ) = - 𝑅2 (𝑄𝑔 ) = 0.825 𝑅2 (𝑄𝑔 ) = 0.8843
𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.0916 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = - 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.0917 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.0746
𝑅2 (𝐺𝑃 ) = 0.9161 𝑅2 (𝐺𝑃 ) = - 𝑅2 (𝐺𝑃 ) = 0.9517 𝑅2 (𝐺𝑃 ) = 0.7966
𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 151.56 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = - 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 115.06 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 235.95

Volume 24, Issue 6, June - 2022 149

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Table 8 Curve fitting parameters and statistical values of DCA models for case 4
Arps Model PLE Model T-Model Duong Model
𝑞𝑖 = 16.3407 𝑞𝑖 = 18.2077 𝑎 = 2.1646 𝑞1 = 12.4432
𝐷𝑖 = 0.0152 𝐷1 = 0.0413 𝑏 = -1.3348 𝑎 = 2.1646
𝑏 = 0.8151 𝐷∞ = 0.0000 𝐺𝑅 = 4513.42 𝑚 = 1.3348
𝑛 = 0.5889
𝑅2 (𝑄𝑔 ) = 0.9741 𝑅2 (𝑄𝑔 ) = 0.9715 𝑅2 (𝑄𝑔 ) = 0.7345 𝑅2 (𝑄𝑔 ) = 0.856
𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.5582 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.5852 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 1.7869 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 1.3158
2 = 0.9985 2 = 0.9999 2 = 0.9688 2 = 0.7155
𝑅 (𝐺𝑃 ) 𝑅 (𝐺𝑃 ) 𝑅 (𝐺𝑃 ) 𝑅 (𝐺𝑃 )
𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 22.1799 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 6.7575 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 100.312 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 302.67

Table 9 Curve fitting parameters and statistical values of DCA models for case 5
Arps Model PLE Model T-Model Duong Model
𝑞𝑖 = 6.7696 𝑞𝑖 = 24.186 𝑎 = 1.2821 𝑞1 = 7.5428
𝐷𝑖 = 0.0106 𝐷1 = 0.1654 𝑏 = -1.1413 𝑎 = 1.2821
𝑏 = 1.0000 𝐷∞ = 0.0000 𝐺𝑅 = 50873 𝑚 = 1.1413
𝑛 = 0.2303
𝑅2 (𝑄𝑔 ) = 0.9706 𝑅2 (𝑄𝑔 ) = 0.9945 𝑅2 (𝑄𝑔 ) = 0.9923 𝑅2 (𝑄𝑔 ) = 0.9926
𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.2008 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.0865 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.1028 𝑅𝑀𝑆𝐸 (𝑄𝑔 ) = 0.1009
2 2 2 2
𝑅 (𝐺𝑃 ) = 0.9949 𝑅 (𝐺𝑃 ) = 0.9984 𝑅 (𝐺𝑃 ) = 0.9993 𝑅 (𝐺𝑃 ) = 0.9986
𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 29.242 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 16.522 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 10.885 𝑅𝑀𝑆𝐸 (𝐺𝑃 ) = 15.322

Table 10 Curve fitting parameters and statistical values of DCA models for case 6
Arps Model PLE Model T-Model Duong Model
𝑞𝑖 = 935.63 𝑞𝑖 = 9776.3 𝑎 = 1.6799 𝑞1 = 722.56
𝐷𝑖 = 0.0319 𝐷1 = 0.2978 𝑏 = -1.2545 𝑎 = 1.6799
𝑏 = 1.0000 𝐷∞ = 0.0000 𝑁𝑅 = 3.2*105 𝑚 = 1.2545
𝑛 = 0.1713
𝑅2 (𝑄𝑜 ) = 0.9716 𝑅2 (𝑄𝑜 ) = 0.9805 𝑅2 (𝑄𝑜 ) = 0.9769 𝑅2 (𝑄𝑜 ) = 0.9774
𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 17.343 𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 14.394 𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 15.651 𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 15.466
2 = 0.9926 2 = 0.9696 2 = 0.9956 2 = 0.9917
𝑅 (𝑁𝑃 ) 𝑅 (𝑁𝑃 ) 𝑅 (𝑁𝑃 ) 𝑅 (𝑁𝑃 )
𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 1956.2 𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 3973.2 𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 1509.0 𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 2073.9

Table 11 Curve fitting parameters and statistical values of DCA models for case 7
Arps Model PLE Model T-Model Duong Model
𝑞𝑖 = 570.07 𝑞𝑖 = 572.35 𝑎 = 6.5785 𝑞1 = 3.0505
𝐷𝑖 = 0.0042 𝐷1 = 0.0077 𝑏 = -1.5086 𝑎 = 6.5785
𝑏 = 0.4349 𝐷∞ = 0.0000 𝑁𝑅 = 2.4*105 𝑚 = 1.5086
𝑛 = 0.8095
𝑅2 (𝑄𝑜 ) = 0.9008 𝑅2 (𝑄𝑜 ) = 0.8901 𝑅2 (𝑄𝑜 ) = 0.5887 𝑅2 (𝑄𝑜 ) = 0.7427
𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 42.130 𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 44.345 𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 85.798 𝑅𝑀𝑆𝐸 (𝑄𝑜 ) = 67.867
2 = 0.9983 2 = 0.9965 2 = 0.9355 2 = 0.5429
𝑅 (𝑁𝑃 ) 𝑅 (𝑁𝑃 ) 𝑅 (𝑁𝑃 ) 𝑅 (𝑁𝑃 )
𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 2006.3 𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 2838.4 𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 12252 𝑅𝑀𝑆𝐸 (𝑁𝑃 ) = 32625

Volume 24, Issue 6, June - 2022 150

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figures 14-20 show the future production behavior of the seven cases as predicted by
DCA models while EUR values are summarized in Tables 12-18. The results showed that T and
Duong models could predict the future production behavior of cases 1, 2, 5 and 6 with high
accuracy and with the same manner. They recorded nearly the same EUR values. While Arps
model shows good prediction for cases 1, 5, 6 and 7. PLE model predicted the future production
behavior of cases 4, 5 and 6 with high accuracy. And for case 5, EUR value predicted using PLE
model is equal to 2.67 MMMSCF, as given in Table 16, which is consistent with that calculated
by Al-Ahmadi et al. [35] (2.74 MMMSCF) when using the linear flow analysis [36]. Also, EUR
values predicted by other models are close to that value. This proved that the PDAS application
is a powerful tool and gives reliable results.

Figure 14 Future prediction of production behavior for case 1 using DCA models

Figure 15 Future prediction of production behavior for case 2 using DCA models

Volume 24, Issue 6, June - 2022 151

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 16 Future prediction of production behavior for case 3 using DCA models

Figure 17 Future prediction of production behavior for case 4 using DCA models

Volume 24, Issue 6, June - 2022 152

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 18 Future prediction of production behavior for case 5 using DCA models

Figure 19 Future prediction of production behavior for case 6 using DCA models

Volume 24, Issue 6, June - 2022 153

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 20 Future prediction of production behavior for case 7 using DCA models

Table 12 Estimated ultimate recovery of DCA models for case 1

Arps Model PLE Model T-Model Duong Model
MMSCF
Q limit , day = 0.10 Q limit ,
MMSCF = 0.10 Q limit ,
MMSCF = 0.10 Q limit ,
MMSCF = 0.10
day day day
t limit , day = 834 t limit , day = 590 t limit , day = 1331 t limit , day = 1312

EUR, MMMSCF = 0.72 EUR, MMMSCF = 0.66 EUR, MMMSCF = 0.81 EUR, MMMSCF = 0.79

Table 13 Estimated ultimate recovery of DCA models for case 2

Arps Model PLE Model T-Model Duong Model
MMSCF
Q limit , day = 0.10 Q limit ,
MMSCF = - Q limit ,
MMSCF = 0.10 Q limit ,
MMSCF = 0.10
day day day
t limit , day = 31236 t limit , day = - t limit , day = 63380 t limit , day = 57929

EUR, MMMSCF = 13.65 EUR, MMMSCF = - EUR, MMMSCF = 21.05 EUR, MMMSCF = 19.03

Table 14 Estimated ultimate recovery of DCA models for case 3

Arps Model PLE Model T-Model Duong Model
Q limit ,
MMSCF = 0.05 Q limit ,
MMSCF = - Q limit ,
MMSCF = 0.05 Q limit ,
MMSCF = 0.05
day day day day
t limit , day = 10767 t limit , day = - t limit , day = 12942 t limit , day = 10858

EUR, MMMSCF = 1.896 EUR, MMMSCF = - EUR, MMMSCF = 2.179 EUR, MMMSCF = 1.773

Volume 24, Issue 6, June - 2022 154

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Table 15 Estimated ultimate recovery of DCA models for case 4

Arps Model PLE Model T-Model Duong Model
Q limit ,
MMSCF = 0.10 Q limit ,
MMSCF = 0.10 Q limit ,
MMSCF = 0.10 Q limit ,
MMSCF = 0.10
day day day day
t limit , day = 5048 t limit , day = 1498 t limit , day = 4055 t limit , day = 3430

EUR, MMMSCF = 3.54 EUR, MMMSCF = 2.50 EUR, MMMSCF = 3.02 EUR, MMMSCF = 2.42

Table 16 Estimated ultimate recovery of DCA models for case 5

Arps Model PLE Model T-Model Duong Model
Q limit ,
MMSCF = 0.200 Q limit ,
MMSCF = 0.200 Q limit ,
MMSCF = 0.200 Q limit ,
MMSCF = 0.200
day day day day
t limit , day = 3088 t limit , day = 3807 t limit , day = 6936 t limit , day = 7023

EUR, MMMSCF = 2.241 EUR, MMMSCF = 2.670 EUR, MMMSCF = 3.775 EUR, MMMSCF = 3.829

Table 17 Estimated ultimate recovery of DCA models for case 6

Arps Model PLE Model T-Model Duong Model
MSTB
Q limit , day = 0.010 MSTB
Q limit , day = 0.010 Q limit ,
MSTB = 0.010 Q limit ,
MSTB = 0.010
day day
t limit , day = 2902 t limit , day = 3086 t limit , day = 2985 t limit , day = 2936

EUR, MMSTB = 0.133 EUR, MMSTB = 0.146 EUR, MMSTB = 0.136 EUR, MMSTB = 0.133

Table 18 Estimated ultimate recovery of DCA models for case 7

Arps Model PLE Model T-Model Duong Model
MSTB
Q limit , day = 0.010 MSTB
Q limit , day = 0.010 Q limit ,
MSTB = 0.010 Q limit ,
MSTB = 0.010
day day
t limit , day = 2633 t limit , day = 1754 t limit , day = 2400 t limit , day = 2015

EUR, MMSTB = 0.216 EUR, MMSTB = 0.195 EUR, MMSTB = 0.191 EUR, MMSTB = 0.147

6. Conclusions

Unconventional reservoirs produced under long-term transient flow which is followed

by a BDF. This production behavior requires DCA models other than Arps model to simulate it. If
shortly spaced hydraulic fractures are used, BDF starts after a short period and Arps model
simulate the production behavior with high accuracy. Single and combined DCA models have
been developed to simulate the production behavior of unconventional reservoirs. In this study,
we conducted a comparison between the most used models; Arps model, PLE model, T-model
and Duong model. This comparison was based on flow-regime identification. As different flow-
regime sequences appear in unconventional reservoirs depending on the characteristics of the
reservoir, the well and/or the fractures. The most appeared flow-regime sequences are:
i. Backflow + Bilinear Flow + Linear Flow + BDF.
ii. Backflow + Linear Flow + BDF.
iii. Linear flow + BDF.

Volume 24, Issue 6, June - 2022 155

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

A software application called PDAS application has been built to facilitate conducting the
comparative study. Conclusions emanating from this study are as follows:
1) Arps model: works well when BDF regime occurs and most of the production data is
during this flow regime. This is the case when MFHWs are used.
2) PLE model: shows high curve-fit accuracy when flow regime smoothly changes from
type to another along production life and backflow period is negligible. In this case, PLE
model also makes good prediction for the future production behavior and gives reliable
results for EUR value.
3) T and Duong models: have similar mathematical expressions and can simulate transient
flow with high accuracy and with the same manner. The presence of BDF and noisy data
decreases their curve-fit accuracy. They can’t simulate the backflow period and
production data of this period must be removed before curve-fit process.
4) The PDAS application is a powerful tool and provides reliable results.
References
[1] R. Zhang, L. Zhang, R. Wang, Y. Zhao, R. Huang, Simulation of a Multistage Fractured
Horizontal Well with Finite Conductivity in Composite Shale Gas Reservoir through
Finite-Element Method, Energy & Fuels, 30 (11) (2016) 9036–9049.
https://doi.org/10.1021/acs.energyfuels.6b01565.
[2] L. Ren, R. Lin, J. Zhao, V. Rasouli, J. Zhao, H. Yang, Stimulated Reservoir Volume
Estimation for Shale Gas Fracturing: Mechanism and Modelling Approach, J. Pet. Sci.
Eng., 166 (2018) 290–304. https://doi.org/10.1016/j.petrol.2018.03.041.
[3] T. Ahmed, Modern Decline Curve Analysis, in Reservoir Engineering Handbook, fifth ed.,
Gulf Professional Publishing, 2019, pp. 1389–1461. https://doi.org/10.1016/B978-0-12-
813649-2.00018-9.
[4] W. Wang, D. Fan, G. Sheng, Z. Chen, Y. Su, A review of analytical and semi-analytical fluid
flow models for ultra-tight hydrocarbon reservoirs, Fuel, 256 (2019) 115737.
https://doi.org/10.1016/j.fuel.2019.115737.
[5] M.S. Kanfar and R.A. Wattenbarger, Comparison of Empirical Decline Curve Methods for
Shale Wells, in Paper presented at the SPE Canadian Unconventional Resources
Conference, Calgary, Alberta, Canada, October 2012. https://doi.org/10.2118/162648-
MS.
[6] L. Tan, L. Zuo, B. Wang, Methods of Decline Curve Analysis for Shale Gas Reservoirs,
Energies, 11 (3) (2018) 552. https://doi.org/10.3390/en11030552.
[7] O. Mahmoud, M. Ibrahim, C. Pieprzica, L. Shane, EUR Prediction for Unconventional
Reservoirs: State of the Art and Field Case, in Paper presented at the SPE Trinidad and
Tobago Section Energy Resources Conference, Port of Spain, Trinidad and Tobago, June
2018. https://doi.org/10.2118/191160-MS.
[8] M. Ibrahim, O. Mahmoud, C. Pieprzica, A New Look at Reserves Estimation of
Unconventional Gas Reservoirs, in Paper presented at the SPE/AAPG/SEG

Volume 24, Issue 6, June - 2022 156

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Unconventional Resources Technology Conference, Houston, Texas, USA, July 2018.

https://doi.org/10.15530/URTEC-2018-2903130.
[9] H. Liang, L. Zhang, Y. Zhao, B. Zhang, C. Chang, M. Chen, M. Bai, Empirical methods of
decline-curve analysis for shale gas reservoirs: Review, evaluation, and application, J.
Nat. Gas Sci. Eng., 83 (2020) 103531. https://doi.org/10.1016/j.jngse.2020.103531.
[10] H. Cinco-Ley, F. Samaniego-V, Transient Pressure Analysis for Fractured Wells, Journal
of Petroleum Technology, 9 (33) (1981) 1749-1766. https://doi.org/10.2118/7490-PA.
[11] C. C. Chen, R. Raghavan, A Multiply-Fractured Horizontal Well in a Rectangular
Drainage Region, Society of Petroleum Engineers Journal, 2 (4) (1997) 455-465.
https://doi.org/10.2118/37072-PA.
[12] R. Raghavan, C. C. Chen, B. Agarwal, An Analysis of Horizontal Wells Intercepted by
Multiple Fractures, Society of Petroleum Engineers Journal, 2 (3) (1997) 235-245.
https://doi.org/10.2118/27652-PA.
[13] R. A. Wattenbarger, A. H. El-Banbi, M. E. Villegas, J. B. Maggard, Production Analysis of
Linear Flow Into Fractured Tight Gas Wells, in Paper presented at the SPE Rocky
Mountain Regional/Low-Permeability Reservoirs Symposium, Denver, Colorado, 1998.
https://doi.org/10.2118/39931-MS.
[14] B. Song, C. Ehlig-Economides, Rate-Normalized Pressure Analysis for Determination of
Shale Gas Performance, in Paper presented at the North American Unconventional Gas
Conference, The Woodlands, Texas, 2011. https://doi.org/10.2118/144031-MS.
[15] R. O. Bello, R. A. Wattenbarger, Multi-Stage Hydraulically Fractured Shale Gas Rate
Transient Analysis, in Paper presented at the North Africa Technical Conference and
Exhibition, Cairo, Egypt, February 2010. https://doi.org/10.2118/126754-MS.
[16] J. Lee, J. B. Rollins, J.P. Spivey, Pressure Transient Testing, Textbook Series, vol. 9,
Society of Petroleum Engineers, Richardson, Texas, 2003.
[17] R.H. Johnson, A.L. Bollens, The Loss Ratio Method of Extrapolating Oil Well Decline
Curves, Trans. AIME, 77 (1) (1927) 771–778. https://doi.org/10.2118/927771-G.
[18] S.J. Pirson, Production Decline Curve of Oil Well May Be Extrapolated by Loss-Ratio, Oil
and Gas J., (1935).
[19] J.J. Arps, Analysis of Decline Curves, Trans. AIME, 160 (1) (1945) 228–247.
https://doi.org/10.2118/945228-G.
[20] D. Ilk, J.A. Rushing, A.D. Perego, T.A. Blasingame, Exponential vs. Hyperbolic Decline in
Tight Gas Sands: Understanding the Origin and Implications for Reserve Estimates Using
Arps' Decline Curves, in Paper presented at the SPE Annual Technical Conference and
Exhibition, Denver, Colorado, USA, September 2008. https://doi.org/10.2118/116731-
MS.
[21] P.P. Valkό, Assigning Value to Stimulation in the Barnett Shale: A Simultaneous Analysis
of 7000 Plus Production Histories and Well Completion Records,” in Paper presented at
the SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, January
2009. https://doi.org/10.2118/119369-MS.

Volume 24, Issue 6, June - 2022 157

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

[22] F.S. Huang, Y.S. Zhao, Q. N. Liu, A New Model for Oilfield Performance Prediction,
Petroleum Geology and Oilfield Development in Daqing, 6 (4) (1987) 55–62.
[23] J. Hu, Y. Chen, Application and Discussion on T-Model, Natural Gas Industry, 15 (4)
(1995) 26–29.
[24] H. Dou, C. Chen, Y. Chang, Y. Fang, X. Chen, W. Liu, W. Cai, Analysis and Comparison of
Decline Models: A Field Case Study for the Intercampo Oil Field, Venezuela, SPE Reserv.
Eval. Eng., 12 (1) (2009) 68–78. https://doi.org/10.2118/106440-PA.
[25] A.J. Clark, L.W. Lake, T.W. Patzek, Production Forecasting with Logistic Growth Models,
in Paper presented at the SPE Annual Technical Conference and Exhibition, Denver,
Colorado, USA, October 2011. https://doi.org/10.2118/144790-MS.
[26] R.P. Spencer, M.J. Coulombe, Quantitation of Hepatic Growth and Regeneration,
Growth, Development and Aging, 30 (3) (1966) 277–284.
[27] A.N. Duong, Rate-Decline Analysis for Fracture-Dominated Shale Reservoirs, SPE
Reserv. Eval. Eng., 14 (3) (2011) 377–387. https://doi.org/10.2118/137748-PA.
[28] S. Yu, D.J. Miocevic, An Improved Method to Obtain Reliable Production and EUR
Prediction for Wells with Short Production History in Tight/Shale Reservoirs, in Paper
presented at the SPE/AAPG/SEG Unconventional Resources Technology Conference,
Denver, Colorado, USA, August 2013, pp. 29–39. https://doi.org/10.1190/urtec2013-
003.
[29] M.J. Fetkovich, Decline Curve Analysis Using Type Curves, J. Pet. Technol., 32 (6) (1980)
1065–1077. https://doi.org/10.2118/4629-PA.
[30] H. Zhang, M. Cocco, D. Rietz, A. Cagle, An Empirical Extended Exponential Decline Curve
for Shale Reservoirs, in Paper presented at the SPE Annual Technical Conference and
Exhibition, Houston, Texas, USA, September 2015. https://doi.org/10.2118/175016-MS.
[31] L. Mattar, S. Moghadam, Modified Power Law Exponential Decline for Tight Gas, in
Paper presented at the Canadian International Petroleum Conference, Calgary, Alberta,
June 2009. https://doi.org/10.2118/2009-198.
[32] K. Joshi, J. Lee, Comparison of Various Deterministic Forecasting Techniques in Shale
Gas Reservoirs, in Paper presented at the SPE Hydraulic Fracturing Technology
Conference, The Woodlands, Texas, USA, February 2013.
https://doi.org/10.2118/163870-MS.
[33] N.T. Wang, Z.L. Chen, M.Q. Zhu, H. Zhu, Analysis of the combined model for the
production decline of the shale-gas fractured horizontal well, Pet. Geol. Oilfield Dev.
Daqing, 37 (5) (2018) 138–143. https://doi.org/10.19597/J.ISSN.1000-3754.201707046.
[34] Q. Chen, N. Wang, K. Ruan, M. Zhang, Selection of production decline analysis method
of shale gas well, Reserv. Eval. Develop, 8 (2) (2018) 76–79.
https://doi.org/10.13809/j.cnki.cn32-1825/te.2018.02.014.
[35] H.A. Al-Ahmadi, A.M. Almarzooq, R.A. Wattenbarger, Application of Linear Flow
Analysis to Shale Gas Wells—Field Cases, in Paper presented at the SPE Unconventional
Gas Conference, Pittsburgh, Pennsylvania, USA, February 2010.
https://doi.org/10.2118/130370-MS.

Volume 24, Issue 6, June - 2022 158

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

[36] M. Ibrahim, R.A. Wattenbarger, Analysis of Rate Dependence in Transient Linear Flow
in Tight Gas Wells, in Paper presented at the Canadian International Petroleum
Conference, Calgary, Alberta, June 2005. https://doi.org/10.2118/2005-057.

Volume 24, Issue 6, June - 2022 159

View publication stats