SPE-217288-MS

A Fit-For-Purpose Automated Decline Curve Analysis Using Python and BI

Dashboard

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

A. Kumar, S. T. Nukala, V. Deshpande, N. K. Mishra, D. Solohaga, A. Singh, T. B. Wydiabhakti, S. Rajput, S.
Ashish, V. S. Lopes, D. Ghosal, V. Kumar, and S. Kumar, SLB

Copyright 2023, Society of Petroleum Engineers DOI 10.2118/217288-MS

This paper was prepared for presentation at the Asia Pacific Unconventional Resources Symposium held in Brisbane, Australia on 14 – 15 November 2023.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Decline curve analysis (DCA) is a curve fitting procedure used for analyzing declining production rates and
forecasting future performance of oil and gas wells. Certain types of standard curves are fit, based on the
past production performance which are then extrapolated to predict the future well performance. It is the
basic tool for estimating recoverable reserves and can be used when the production history is long enough
that a decline trend can be identified.
One of the key technologies used in the process is an automated DCA methodology which can be
automatically run on a machine without any human interaction and with minimum errors. The traditional
DCA is derived by Arp's equation, where three types of decline curves have been identified: exponential,
hyperbolic, and harmonic.
In this research work, authors have developed a python code that apply DCA methods to wells in an
unbiased, systematic, and automated manner. This method contrasts with manual DCA, which was the
widespread practice of the industry. As industry has entered digital transformation era, it is imperative that
digital capabilities be coupled with conventional knowledge of DCA. This can be used as a handy tool to
generate type curves for fitting decline curves and to further analyze using business intelligence (BI) tools,
for a batch of oil wells under study.

Introduction
Lot of work in the past has been conducted for theoretical and practical applications of DCA manually.
It is subjective and based on experience of reservoir/ production engineer's judgment. In recent years,
advancements in data science and automation have led to the development of automated DCA methods,
allowing engineers and analysts to streamline the process of analyzing production data and extracting
meaningful insights. Python, a popular programming language in the data science community, has played
a significant role in enabling the automation of DCA.
Paper discusses DCA techniques and its evolution in chronological order. Python libraries and its
functionalities developed by authors would be discussed in detail in this paper. Test cases were developed
for verifying the workings of the code, which culminate onto developing formation characteristics-based
2 SPE-217288-MS

decline curve groupings. This would aid readers to understand application of programming techniques for
solving everyday petroleum engineering problems.

Literature Review of DCA Techniques

Arps decline curve model is one of the most widely used decline curve models (Arps 1945) in the oil and
gas industry. It accommodates a more gradual decline in production compared to exponential decline. The
hyperbolic decline equation is given by equation 1 below:

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

(1)

Where:
Q(t) = Production rate at time t
Q(O) = Initial production rate
b = Hyperbolic constant
D = Decline rate
t = Time
In above equation, the matching parameters are b and D. The values b = 0 and b = 1 represent exponential
and harmonic decline, respectively. During pseudo steady state for liquid systems, an exponential decline
is a characteristic of constant pressure production whereas a harmonic decline occurs with constant rate
production. Any other value of b represents a hyperbolic decline.
A set of graphical type curves were proposed that represent different decline behaviors based on the
hyperbolic decline equation. These type curves allowed engineers to match production data to a specific
curve and estimate decline parameters.
This work was expanded upon by introducing additional type curves that accounted for more complex
production scenarios, including wells subjected to boundary-dominated flow or various combinations of
boundary and transient flow (Fetkovich et al. 1980). Fetkovich's type curves offered improved accuracy in
analyzing unconventional reservoirs and transient flow conditions.
In 1990's, the concept of type curves was further extended (Palacio et al. 1993) by developing diagnostic
type curves that helped identify and quantify specific wellbore damage, fracture closure, and reservoir
heterogeneity effects on production decline. These diagnostic curves added a layer of interpretation to DCA
beyond simple parameter estimation.
In 2000's, type curves were introduced specifically for unconventional reservoirs which often exhibit
complex transient flow behavior (Holditch et al. 2006). These type curves considered the unique
characteristics of shale gas and tight oil reservoirs, helping engineers better interpret production data from
these challenging environments.
Since 2010, As reservoir simulation techniques became more sophisticated, researchers started
integrating numerical simulations with type curves. This integration allowed for the development of
integrated type curves (Purvis et al., 2016) that incorporated both boundary-dominated and transient flow
behaviors. These type curves provided more accurate predictions by considering a broader range of reservoir
condition.
The earlier automations in DCA were based on usage of non-linear parametric estimations (Chen et al.
1993). In recent times, artificial intelligence has been recently used to perform regressions like the use
of probabilistic DCA using methods of quantile regression (Kianinejad et al. 2019). Artificial intelligence
based DCA has been used in integrated management of reservoir performance and field production
(Carpenter. 2020). Monte-Carlo methods were used for deriving the probabilistic estimates at the field level
(Castiñeira et al. 2014). The MC simulations were used to compute reserves for over 9,000 wells at the field
level (Chavali et al. 2021). The various AI probabilistic workflows integrating petrophysical evaluation has
SPE-217288-MS 3

been introduced to increase the accuracy (Snow et al. 2020). The results of automated DCA are further
integrated into the reservoir management digital portals for real time decision-making (Nuaim et al. 2011). A
SCADA system can be used for full-scale end-to-end automation including data import/export, processing,
visualization, and analysis (Jordan et al. 2022). The usage is further extended to unconventional coalbed
methane wells (Kuznetsov et al. 2018), coal seam gas wells (Falcon et al. 2021), using three-segment DCA
using open-source programming language Python (Mogollón et al. 2021). These automated DCA methods
help in significantly reducing the cycle time involved in forecasting (Kumar et al. 2019).
Method adopted for this work using python libraries and its functionalities to automate DCA workflow

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

is briefly discussed in the next section.

Current methodology
This paper showcases the extensive research work done to automate the Decline Curve Analysis (DCA)
workflow using Python libraries and their functionalities is outlined briefly below.

Data Collection and Preprocessing.

The first step involves collecting historical production data, which includes information about production
rates and cumulative production over time. Python libraries such as Pandas, NumPy, SciPy were used to
efficiently handle, clean, and preprocess the data.

Curve Fitting.
Automated DCA in Python involves fitting decline curves to the production data. There are several decline
curve models available, such as the exponential decline model, hyperbolic decline model, and harmonic
decline model. Libraries like SciPy provide tools for curve fitting, enabling the selection of the most
appropriate model for a given dataset.

Parameter Estimation.
Once a decline curve model is chosen, Python can be used to estimate the parameters of the model that best
fit the production data. Optimization techniques available in libraries like SciPy can automatically find the
optimal parameter values that minimize the difference between the model and the actual data.

Production Forecasting.
Automated DCA allows engineers to forecast future production rates and cumulative production based
on the fitted decline curve model. Python scripts can be developed to project production behavior under
different scenarios and operational conditions.

Visualization.
Python's visualization libraries, such as Matplotlib, Seaborn, Plotly etc., or BI dashboard make it easy
to create informative graphs and plots that visually represent the DCA results. These visualizations help
stakeholders understand the trends and patterns in the production data.

Decision Support.
By automating the DCA process, decision-makers can quickly assess the performance of wells, identify
anomalies, and make informed decisions about well interventions, optimization strategies, and future
investments.
As programming languages continue to evolve toward greater user-friendliness, harnessing Python
for Decline Curve Analysis (DCA) emerges as a potent avenue for scrutinizing production data and
deriving decline parameters. Thus, while crafting an automated DCA solution, a foundational grasp of
Python libraries and their functionalities becomes imperative. Proficiency in programming languages has
4 SPE-217288-MS

emerged as a pivotal skillset within modern solutions for petroleum engineering, especially in the context
of optimizing production strategies for mature fields. This emphasis on technology-driven production
optimization underscores the evolving landscape of the industry, highlighting the significance of adapting
to new-age methodologies and tools.

Python Libraries & Implementation

Pandas stands as a pervasive Python library extensively employed in the realm of data manipulation and
analysis. It furnishes a comprehensive suite of data structures and operations tailored to streamline tasks

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

involving structured data, encompassing tables, spreadsheets, and databases. In the context of the current
endeavor, several pivotal functionalities of pandas were harnessed, including:
Reading and Writing Data: Pandas can read data from various file formats (CSV, Excel, SQL databases,
etc.) and write data back to these formats. Using this DataFrame functionality, production data loaded in the
BI dashboard in the excel format was read and data transformations were automated in the BI workflow.
Converting data into readable format and storing is achieved through data different types of data (numeric,
string, etc.). Using this DataFrame functionality, production data was stored in the structure after reading
the data structures. Two types of data structures were employed. DataFrame is a two-dimensional labeled
data structure, resembling a table with rows and columns. Each column can hold different types of data
(numeric, string, etc.). Using this DataFrame functionality, production data was stored in the structure after
reading the data. Series is a one-dimensional labeled array that can hold any data type. Series is used to
store columns in a DataFrame. Using this DataFrame functionality, production data columns were extracted
for automation algorithm.
The key concepts for accessing the data used in the workflow were Index and Column. The presence of an
index for every DataFrame and Series plays a pivotal role in labeling rows, facilitating efficient data retrieval
and manipulation. Capitalizing on this inherent capability, a DataFrame was populated with forecasted
data, leveraging the index structure. Moreover, the potential of the index was harnessed to seamlessly
update decline curve parameters, thereby streamlining the process of incorporating new information into
the analysis. The columns within a DataFrame are endowed with labels, enabling straightforward data
identification and selection. Capitalizing on this feature, the process involved accessing specific columns
for the purpose of updating tables, and further, the creation of new columns was facilitated seamlessly.
The empty DataFrames were created, which were later populated with forecast data and decline curve
parameters. The DataFrame comprising forecasted data served as a valuable resource for conducting a
comprehensive evaluation, aimed at gauging the congruence between historical production records and the
forecasted production data and the DataFrame comprising decline curve parameters served as input data for
generating zone wise type curves and for understanding the decline trends in the field.
The basic data preparation steps include sorting, data cleaning, filtering and aggregation. As part of the
data preprocessing phase, the DataFrame underwent sorting operations based on the well name and date,
thereby arranging the data in a structured manner that aligns with the order of well names and chronological
dates. As the second pivotal stride within the data preprocessing continuum, Python functions were wielded
adeptly to excise well shut-in data, effecting a discernible refinement in the dataset's quality and integrity.
Next, data manipulation was orchestrated by selectively filtering rows and columns using specific criteria,
such as well name, to facilitate the derivation of decline curve parameters for individual wells. Zone
names were similarly employed to construct type curves tailored to each distinct zone, while decline curve
parameters were harnessed to visualize wells that align with decline trends. This comprehensive approach
enabled targeted analysis and interpretation of data within the context of decline analysis. Following the
data filtration process tailored to specific needs, a range of aggregation operations such as minimum,
maximum, mean etc. were executed. Subsequently, leveraging the aggregated values, distinct type curves
were generated for each zone, encapsulating the characteristics of the decline trend within those zones.
SPE-217288-MS 5

NumPy is a fundamental Python library for numerical computations, particularly when working with
arrays and matrices. NumPy arrays exhibit a distinctive feature of contiguous memory storage, setting
them apart from lists. As a result, they facilitate streamlined accessibility and manipulation by various
processes, ultimately leading to enhanced operational efficiency. It forms the foundation for many other
scientific computing libraries. It was used for basic arithmetic calculations such as addition, subtraction,
multiplication, division on columns, universal functions for elementwise calculations such as applying row
wise function to forecast production data based on decline curve parameters and model function used in
the curve fitting algorithm and for list generation to project production rates, a list was created utilizing

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

the capabilities of the NumPy linspace library. This library proves instrumental in generating an array of
production days, which subsequently serves as input for the model function. This input, combined with
decline curve parameters, facilitates the forecasting of well production data. Moreover, this predictive
approach extends to estimating the field's reserves, encompassing a comprehensive analysis of production
trends and potential outcomes.
SciPy emerges as a crucial Python library that builds upon the foundation of NumPy, focusing on the
domains of scientific and technical computing. This augmentation of NumPy's capabilities is exemplified
through an extensive assortment of specialized functions tailored to address an array of scientific tasks.
Within the context of the current scope, SciPy was used for Optimization, which boasts a robust suite of
optimization algorithms meticulously designed to pinpoint the minima and maxima of functions, adeptly
navigating both constrained and unconstrained contexts. This proficiency proved invaluable in the context
of curve fitting, where the challenge lies in aligning historical production data with data synthesized through
curve fitting. The distinction between historical data and data generated by curve fitting underscores the
essence of optimization. When working with historical production data, the objective is often to find
the parameters that best align with the real-world observations, ultimately minimizing the discrepancies
between the actual values and the model's predictions. By harnessing SciPy's optimization capabilities, the
integration of historical data and curve fitting outcomes is refined, ultimately facilitating a more accurate
representation of the underlying phenomena and enhancing the predictive power of the model.
The curve_fit function in SciPy allows you to choose between different optimization methods to fit your
model to observed data. The step-by-step process for curve fitting is discussed below.
First step was aimed at formulating the model function. At this stage, a distinctive model function was
established, specifically employing a hyperbolic decline curve function. The parameters to be optimized b,
D & Q(O) were provided along with historical production data. Then in the second step, initial parameter
estimates and bounds were furnished. In this phase, preliminary approximations were supplied for the
model's parameters-namely, b, Q(O), and D—alongside defined boundaries. These initial estimates serve
as the launchpad for the optimization procedure, guiding the algorithm's exploration of the solution space.
It's important to note that without suitable initial parameter values, SciPy might encounter difficulties
in converging to a solution, underscoring the significance of providing accurate starting points for the
optimization process. During the third step, the Trust Region Reflective (TRF) method was adopted for
the optimization process. This method demonstrates optimal performance when the parameters under
consideration are subject to constraints, particularly in the form of bounds that define their allowable ranges.
Alternative optimization techniques, including the Levenberg-Marquardt (LM) method and the Dogbox
method, were also explored during the experimentation phase. However, the LM method encountered
convergence issues, leading to suboptimal outcomes.
Both the TRF and Dogbox methods, on the other hand, yielded results that met the desired criteria without
convergence issues. Next in the fourth step, following the successful optimization of the curve fit function,
the outcome from the curve_fit process provided the values for the parameters b, D, and Q(O) specific to
each well. Subsequently, these outcomes were methodically integrated into a freshly generated DataFrame
named "decline curve parameters." This automated procedure effectively streamlines the arrangement and
display of the achieved results. Then in the fifth step, a visual analysis was conducted by plotting historical
6 SPE-217288-MS

production data against forecasted production data in the Business Intelligence (BI) environment aimed at
assessing the level of congruence between the two datasets.
Furthermore, a new feature has been introduced to enable manual adjustment of decline parameters.
This feature allows users to customize the starting fit date, specify the number of days for forecasting
cumulative oil estimation, and define the fraction of the validation dataset for assessing the quality of the
match. This adjustment functionality was implemented to iteratively fine-tune the parameters manually,
with the objective of achieving an improved alignment between the historical and forecasted production
data. The specified initial values for the curve fitting parameters were b = 0.5, D = 0.005 1/day, and Q(O)

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

= 10,000 Barrels per day. These parameters were also subject to bounds during the fitting process: b is
bounded between 0 and 1 (inclusive), D is bounded between 0 and 0.01 (inclusive), and Q(O) is bounded
between 10 and 10000. In the final step, the parameters optimized through the curve fitting process were
integrated into the model function. to forecast the oil production rates and to estimate ultimate recovery
from well and reserves from asset. The entire DCA automation workflow, enacted within the in the BI
environment for DCA automation is presented below in Fig. 1.

Figure 1—Python optimization workflow for decline curve analysis in BI environment

Case Study: Equinor Volve Dataset

Research work conducted above was carried out on Volve oil field in North Sea as the study area. It was
believed to be formed by the collapse of adjacent salt ridges during the Jurassic period. Oil was discovered
in the field back in 1993 in the middle Jurassic Hugin sandstone formations. In this study, data from four
producer wells viz. NO 15/9-F-1 C, NO 15/9-F-12 H, NO 15/9-F-14 H & NO 15/9-F-15 D was used as
proof of concept to test the decline curve analysis solution.
Wells with extensive production histories were specifically selected for the purpose of production
forecasting and to test the Decline Curve Analysis (DCA) program. In the initial phase, a curve-fitting
algorithm was rigorously examined by adjusting the starting point of the fit. The selection of the best-fit
curve was based on minimizing associated errors. This phase is referred to as the "training dataset phase".
Once the parameters were successfully determined, they underwent validation in the "validation dataset
phase" using the remaining production data.
Achieving a satisfactory curve fit in the validation dataset bolstered confidence in the chosen parameters
(b, D, and Q(O)) for the decline curve analysis. Subsequently, the "forecasting dataset phase" was executed
to estimate the Estimated Ultimate Recovery (EUR) based on the forecasted timeframe.

Discussion of Results
To verify optimization algorithm, different test cases were run with real life production data available.
Results presented below from Fig. 2 to Fig. 5 indicate variability of b, D & Q(O) values with satisfactory
match between observed data and model data.
SPE-217288-MS 7

Using the decline curve parameters for each well, type curves were generated as shown in Fig. 6, for each
of the zones, which could be helpful in future development decisions of the field.

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

Figure 2—Case 4: DCA match for Well D, matching parameters: b=0.55, D=0.0014, Q(O)=600

Figure 3—Case 2: DCA match for Well B, matching parameters: b=0.11, D=0.0008, Q(O)=2200
8 SPE-217288-MS

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

Figure 4—Case 3: DCA match for Well C, matching parameters: b=0.01, D=0.0002, Q(O)=1250

Figure 5—Case 4: DCA match for Well D, matching parameters: b=1, D=0.0008, Q(O)=3000
SPE-217288-MS 9

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

Figure 6—Type curves generated for zone wise performance.

The research workflow was employed to align historical production data from wells in the Volve dataset
and to calculate reserves for the well. Fig. 7 to 10 represent the match between observed and modelled data
and Fig. 11 to 14 represent the cumulative oil with days based on forecasted rate from model.

Figure 7—Case Study: NO 15/9-F-1 C DCA

 Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023
SPE-217288-MS

Figure 8—Case Study: NO 15/9-F-12 H DCA

Figure 9—Case Study: NO 15/9-F-14 H DCA

10
 Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023
11

Figure 11—Case Study: NO 15/9-F-1 C Cumulative Oil Production Forecast

Figure 10—Case Study: NO 15/9-F-15 D DCA
SPE-217288-MS
12 SPE-217288-MS

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

Figure 12—Case Study: NO 15/9-F-12 H Cumulative Oil Production Forecast

Figure 13—Case Study: NO 15/9-F-14 H Cumulative Oil Production Forecast

SPE-217288-MS 13

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

Figure 14—Case Study: NO 15/9-F-1 C Cumulative Oil Production Forecast

Conclusion
In this paper, we showcased the python based automated DCA solution and discussed the accuracy aspects.
Based on the test cases run using automated DCA solution, below are the key conclusions.
By harnessing automation, the need for manual labor is significantly diminished, expediting the analysis
process. This strategic integration empowers engineers to allocate their energies towards interpretation
and decision-making, alleviating the burden of repetitive tasks and enhancing the overall efficiency of the
workflow. Automated DCA ensures consistent analysis across multiple wells and datasets, reducing the
risk of human error. This approach can be helpful for a field scale DCA matching exercise reducing the
overall manual efforts and reduce the time involved in updating the production forecast, which enables
faster performance monitoring.
Python's inherent versatility empowers analysts to tailor and expand their Decline Curve Analysis (DCA)
workflows to precisely align with specific business requisites. The results can be integrated into the real
time reservoir management digital dashboards for interactive visualization and controlled evaluation. Usage
of cloud-based data analytics can be used for real time processing of production decline and further assist
command for real time reservoir management at a field scale and, further will also help in reducing the cost
of running various softwares and the efforts spent on integrating the results from these platforms. Various
smart flow meters and internet of things (IoT) based devices at the production field can help in amalgamation
of high frequency data from various wells and further be pushed to the cloud-based production ecosystem.
The future trajectory of this endeavor encompasses its evolution into a more comprehensive framework
capable of accommodating diverse production scenarios, including transient conditions and unconventional
reservoirs.
It should be noted that methodology employed by authors assumes operating conditions of the well,
including factors such as artificial lift system, secondary recovery mechanisms, back pressure, contributing
14 SPE-217288-MS

net pay etc should remain relatively unchanged and the well's boundary conditions should remain consistent
and unchanged, adding new wells nearby may change the decline trend.

Acknowledgements
The authors want to thank the management of SLB for their permission to perform this research work and
publish this paper, along with their leadership, support, and encouragement for undertaking the task of
automating the existing workflows.
The authors would like to thank Equinor AS, the former Volve license partners ExxonMobil Exploration

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

& Production Norway AS and Bayerngas Norge AS, or their successors or permitted assignees, for
permission to use the Volve data set, and to all who have contributed to work using these data sets. Readers
are directed to data.equinor.com for more information about the Volve data set and license terms of use. In
this paper, the dataset has been used purely for research purposes only to explore different data analytics
workflows that could be developed using python. The authors would like to express their gratitude for
Equinor and other Volve data license partners for providing the data set for this research work.

References:
Arps, J. J. 1945. Analysis of Decline Curves. Trans AIME 160 (1): 228–247. SPE-945228-G. https://
doi.org/10.2118/945228-G
Castiñeira, David, Mondal, Anirban, and Sebastien Matringe. "A New Approach for Fast Evaluations of Large Portfolios
of Oil and Gas Fields." Paper presented at the SPE Annual Technical Conference and Exhibition, Amsterdam, The
Netherlands, October 2014. doi: https://doi.org/10.2118/170989-MS
Carpenter, Chris. "AI-Based Decline-Curve Analysis Manages Reservoir Performance." J Pet Technol 72 (2020): 58–59.
doi: https://doi.org/10.2118/0920-0058-JPT
Chavali, V. B. K., and W. J. Lee. "Fast Probabilistic Forecasting of Oil Production using Monte Carlo Simulations
on Data Driven Acquisition of Decline-Curve Parameter Distributions." Paper presented at the SPE/AAPG/SEG
Unconventional Resources Technology Conference, Houston, Texas, USA, July 2021. doi: https://doi.org/10.15530/
urtec-2021-5696
Chen, Jianping, and J. L. Prouty. "Automated Decline-Curve Analysis Using Robust (LAV) Nonlinear Parameter
Estimation Under MS-Windows™ 3.X." Paper presented at the Petroleum Computer Conference, New Orleans,
Louisiana, July 1993. doi: https://doi.org/10.2118/26249-MS
E. Jones, T. Oliphant, P. Peterson. et al. 2001. SciPy: Open-source scientific tool for Python, http://www.scipy.org/
Equinor, 2018, Volve data village dataset: released under a license based on CC BY 4.0, https://data.equinor.com/"Volve
Field Data Village Download - Data 2008-2016." Volve Field Data Village Download Data 2008-2016, https://
www.equinor.com/en/what-we-do/digitalisation-in-our-dna/volve-fielddata-village-download.html.
Falcon, Gerardo E. Guillen. "Flowing Material Balance and Decline Curve Analysis: A Hybrid Approach for Better EUR
and Decline Profile Predictions for CSG Wells." Paper presented at the SPE/AAPG/SEG Asia Pacific Unconventional
Resources Technology Conference, Virtual, November 2021. doi: https://doi.org/10.15530/AP-URTEC-2021-208381
Fetkovich, M.J. 1980. Decline Curve Using Type Curves, J Pet Technol (June 1980) 1065.
Holditch, S. A. 2006. Tight Gas Sands. J Pet Technol 58 (6): 86–93. SPE-103356-MS. http://dx.doi.org/10.2118/103356-
JPT
Jordan, Colin Lyle, Koochak, Roozbeh, Roberts, Martin, Nalonnil, Ajay, and Mike Honeychurch. "A Holistic Approach
to Big Data and Data Analytics for Automated Reservoir Surveillance and Analysis." Paper presented at the SPE Asia
Pacific Oil & Gas Conference and Exhibition, Adelaide, Australia, October 2022. doi: https://doi.org/10.2118/210757-
MS
Kianinejad, Amir, Kansao, Rami, Maqui, Agustin, Kadlag, Rahul, Hetz, Gill, Ibrahima, Fayadhoi, Suicmez, Vural, Darabi,
Hamed, and David Castineira. "Artificial-Intelligence-Based, Automated Decline Curve Analysis for Reservoir
Performance Management: A Giant Sandstone Reservoir Case Study." Paper presented at the Abu Dhabi International
Petroleum Exhibition & Conference, Abu Dhabi, UAE, November 2019. doi: https://doi.org/10.2118/197142-MS
Kuznetsov, Dan, Fernandes, Glen, Daly, Michael, and Kai Ma. "Statistical Decline Curve Analysis for Automated
Forecasting of Production from Coalbed Methane Wells." Paper presented at the SPE Asia Pacific Oil and Gas
Conference and Exhibition, Brisbane, Australia, October 2018. doi: https://doi.org/10.2118/191985-MS
Mogollón, Leonardo, Sinchitullo, Joseph, and Joel Cahuas. "Improving Tight Oil Reserves Forecast Using 3-Segment
Decline Curves Analysis in a Peruvian Northwest Basin." Paper presented at the SPE/AAPG/SEG Asia Pacific
SPE-217288-MS 15

Unconventional Resources Technology Conference, Virtual, November 2021. doi: https://doi.org/10.15530/AP-

URTEC-2021-208362
Nuaim, Ahmed, Madani, Husameddin, Seemann, Dennis, and Keith Holdaway. "Automating Decline Curve Analysis in
an Integrated Reservoir Management Portal." Paper presented at the SPE Digital Energy Conference and Exhibition,
The Woodlands, Texas, USA, April 2011. doi: https://doi.org/10.2118/143701-MS
Palacio, J.C. and Blasingame, T.A. 1993. Decline-Curve Analysis with Type Curves - Analysis of Gas Well Production
Data. Paper Presented at the Low Permeability Reservoirs Symposium, Denver, Colorado, USA 12-14 April.
SPE-25909-MS. http://dx.doi.org/10.2118/25909-MS
Purvis, D. C., and Heidi Kuzma. Evolution of Uncertainty Methods in Decline Curve Analysis Paper presented at the

Downloaded from http://onepetro.org/SPEURCE/proceedings-pdf/24SA02/2-24SA02/D022S004R002/3306506/spe-217288-ms.pdf by Colin Jordan on 12 November 2023

SPE/IAEE Hydrocarbon Economics and Evaluation Symposium, Houston, Texas, USA, May 2016. SPE-179980-MS.
doi: https://doi.org/10.2118/179980-MS
Snow, Cameron, and Roland Gunther. "Automating Petrophysics and Decline Curves Analysis for Performance Prediction
at the Basin-Scale: Application to the Powder River Basin." Paper presented at the SPE/AAPG/SEG Unconventional
Resources Technology Conference, Virtual, July 2020. doi: https://doi.org/10.15530/urtec-2020-3084