Applied Energy 375 (2024) 124103

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Assessment of geothermal energy potential from abandoned oil and gas wells
in Alberta, Canada✩
Mohammad Zolfagharroshan a , Minghan Xu a , Jade Boutot b , Ahmad F. Zueter a ,
Muhammad S.K. Tareen a , Mary Kang b , Agus P. Sasmito a ,∗
a
Department of Mining and Materials Engineering, McGill University, 3450 University, Frank Dawson Adams Bldg., Montreal, QC H3A0E8, Quebec, Canada
b Department of Civil Engineering, McGill University, 817 Sherbrooke St W, Macdonald Engineering Bldg., Montreal, QC H3A0C3, Quebec, Canada

GRAPHICAL ABSTRACT

ARTICLE INFO ABSTRACT

Keywords: Repurposing abandoned oil and gas wells (AOGWs) into geothermal wells provides an effective solution for
Geothermal energy developing a sustainable energy resource and reducing the carbon footprint. It involves generating green
Abandoned petroleum wells electricity and heat, thereby transforming expensive, unused assets into profitable ones. This current research
Monte-Carlo analysis
conducts a statistical assessment of the potential for geothermal energy extraction from AOGWs in Alberta,
Heat conduction
Canada. An efficient conjugate model of a geothermal system is developed for a double-pipe coaxial heat
exchanger configuration that simulates transient heat extraction and is validated against a pilot test. Then,
a Monte Carlo approach is employed to generate 10,000 samples representing abandoned wells in Alberta,
which serve as inputs for simulating each well over two decades. The simulation results reveal that nearly
80% of the wells show positive potential for geothermal energy extraction. Nearly 0.5% of wells with an
average outlet temperature of 109 ◦ C qualify for electricity generation via the organic Rankine cycle. The
majority are suitable for high to low-grade heating, with average outlet temperatures of 74 ◦ C, 49 ◦ C and
27 ◦ C. A parametric study on inlet water temperature reveals that increasing the inlet temperature by steps of
5 degrees can reduce the produced energy from 20% to 60%. The results underscore the promising potential
of repurposing AOGWs for geothermal energy.

✩ The short version of the paper was presented at ICAE2023, Doha, Qatar, Dec 3–5, 2023. This paper is a substantial extension of the short version of the
conference paper.
∗ Corresponding author.
E-mail address: [email protected] (A.P. Sasmito).
https://doi.org/10.1016/j.apenergy.2024.124103
Received 30 April 2024; Received in revised form 6 July 2024; Accepted 30 July 2024
Available online 7 August 2024
0306-2619/© 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-
nc/4.0/).
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

1. Introduction
Nomenclature

𝑎 Inner radius of ground (m) The energy sector in many countries is undergoing decarbonization
𝑏 Outer radius of ground (m) as part of their net-zero carbon emission strategies [1,2]. Reducing the
use of fossil fuels necessitates replacing them with reliable, sustainable
𝑐𝑝 Specific heat (J kg−1 K −1 )
resources [3,4]. Geothermal energy is a renewable resource that offers
𝐷, 𝑑 Pipe diameter (m)
continuous and sustainable power, unaffected by daily or seasonal vari-
𝐷ℎ Hydraulic diameter (m)
ations, unlike solar and wind [5,6]. The slow pace of geothermal energy
𝐸 Energy (GWh) development is primarily due to high drilling costs [7,8]. However,
𝑓 Friction factor (–) repurposing abandoned oil and gas wells (AOGWs) into geothermal
𝐺 Green function (–) wells is one of the suggested solutions to address upfront drilling
𝑔 Gravitational acceleration (m s−2 ) expenses for geothermal energy development [9]. This approach also
𝐻 Eigenfunction (–) addresses the environmental threats posed by AOGWs and mitigates
ℎ Heat transfer coefficient (W m−2 K −1 ) the challenging tasks associated with well plugging and abandonment
𝐽0 , 𝐽1 Bessel function of the first kind in the zeroth of non-producing petroleum wells [10,11]. The American National
and first orders (–) Petroleum Council defines abandoned hydrocarbon wells as wells that
𝑘 Thermal conductivity (W m−1 K −1 ) meet one of these conditions: lack recent production (e.g., inactive,
temporarily closed), lack a recent operator (i.e., orphaned), or have
𝐿 Length (m)
been permanently plugged and abandoned [12]. AOGWs pose highly
𝑁𝑢 Nusselt number (–)
uncertain risks, such as greenhouse gas emissions and water resource
𝑃𝑟 Prandtl number (–) contamination [13]. This becomes crucial, noting that the number of
𝑄 Heat transfer rate (W) AOGWs worldwide is estimated to be 20–30 million [14,15], with
𝑟 Radial coordinate (m) 400,000 in Canada [16]. Overall, repurposing AOGWs for geothermal
𝑅𝑒 Reynolds number (–) energy extraction appears to offer a multifaceted solution that warrants
𝑇 Temperature (K) more attention [17,18].
𝑡 Time (s) The defined projects and field demonstrations globally on con-
𝑌0 , 𝑌1 Bessel function of the second kind in the verting and repurposing hydrocarbon wells into geothermal wells are
zeroth and first orders (–) increasing, indicating a promising future for this approach [19–21].
𝑍 Depth, 𝑧-coordinate (m) The feasibility field studies mostly focused on co-production of oil
and water [9,12] mostly in mature fields such as Huabei oil field
𝑚̇ Mass flow rate (kg s−1 )
(China) [22], Laurel field (USA) [12], Los Angeles Basin (USA) [23]
Greek Symbols where the water cut is often high (e.g., over 97%). More diverse
cases can be found in the literature, such as projects involving the
𝜆𝑛 Eigenvalues (–)
re-injection of extracted brine into aquifers for geothermal energy
𝜇 Dynamic viscosity (Pa s) (Texas, USA [24]), the development of enhanced geothermal systems
𝜌 Mass density (kg m−3 ) (France [10]), and the utilization of AOGWs for geothermal energy
𝜏 Integration variable for t (–) storage (Illinois, USA [25]). The first closed-loop field demonstration of
𝜃 Shifted temperature (–) an abandoned well was implemented in Hungary by installing a mini
𝜉 Integration variable for r (–) heat plant at Kiskunhalas [9]. A deep well that was drilled in the 1960s
was retrofitted to act as a double-pipe coaxial heat exchanger (DPCHE),
Superscripts and subscripts demonstrating the potential to produce 0.5 MW of heat, entirely green
i, in In and suitable for heating a floor area of 20,000–30,000 square meters.
∞ Ambient There are few published details about this site, although the operator
claims that such technology can be applied to the large inventory of
f, l Liquid
9000 existing abandoned wells in Hungary [26].
geo Geothermal
Closed-loop heat extraction configurations are viewed as more reli-
g Ground able options than open-loop systems due to the high risk of mineral
v Vapor scaling and corrosion [27,28] from the direct contact of water and
w Wall reservoir rock in open-loop systems [29,30]. Additionally, a closed
𝑜, 𝑜𝑢𝑡 Out system allows for greater control over operational conditions and has
zero environmental impact, although it may produce less heat com-
Abbreviations
pared to open systems [31,32]. Various studies have proposed models
AOGW Abandoned oil and gas well for converting abandoned wells into closed-loop geothermal systems
CFD Computational fluid dynamics using U-tube and double-pipe coaxial heat exchanger configurations.
DPCHE Double pipe coaxial heat exchanger The presented frameworks share the common approach of solving the
H-G High-grade energy equation in the ground and the mass, momentum, and energy
equations in the wellbore, although employing different methodologies.
HTF Heat transfer fluid
Davis and Michaelides [33] conducted a numerical 1D steady-state
L-G Low-grade
study of a double-pipe coaxial heat exchanger for a typical abandoned
M-G Medium-grade oil well in South Texas (with a depth of 3 km and a bottomhole
ORC Organic Rankine cycle temperature of 157 ◦ C), showing promising potential for generating
2–3 MW of electric power. Bu et al. [34] numerically modeled the
same setup for transient heat extraction from a well with a depth of
4 km, considering geothermal gradients of 25 and 45 ◦ C/km, with
an inlet water temperature of 30 ◦ C over a period of 2 months. The
findings suggested that a minimum distance of 40 m between two wells

2
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

is necessary. Cheng et al. [14] simulated a 6 km deep abandoned well in 2. Methodology

China over 1000 days, varying ground thermal conductivity, formation
heat capacity (𝜌𝑐𝑝 ), and mass flow rate, using isobutane as the working This section is divided into three parts. First, the governing equa-
fluid. They found that the flow rate had the most significant impact tions related to the physics of the problem and the boundary conditions
on stabilization time. Investigating the effects of working fluid, depth, of the ground and wellbore, which are the main computational domains
and geothermal gradient in AOGWs, Cheng et al. [35] found that wells as shown in Fig. 1(a) and (b), are presented. Second, the process of
shallower than 3 km, even at high geothermal gradients of 50 ◦ C/km, preparing the model inputs for statistical analysis is explained. Finally,
exhibit low power generation efficiency. a step-by-step solution algorithm is presented, along with a flowchart.
Templeton et al. [36] developed a model based on
three-dimensional Fourier’s law using commercial software. They were 2.1. Analytical space-marching model
among the first to verify their model against previously developed
frameworks and attributed the deviations in their model to simplifica- This section begins with presenting the exact solution of the tran-
tions present in earlier studies. Wight and Bennett [37] developed a 1D sient energy equation in the ground. Heat conduction in the ground is
steady-state model of a wellbore focusing on geothermal heat extraction considered the dominant heat transfer mechanism, with the effects of
convection and radiation assumed to be negligible. In this scenario, the
from AOGWs, without incorporating a ground model. They determined
heat conduction equation in 3D cylindrical coordinates is expressed as
that water mass flow rates between 2.5 and 15 kg/s could achieve
follows [40]:
a minimum outlet temperature of 130 ◦ C, suitable for connection to ( )
( ) ( )
an organic Rankine cycle (ORC) for electricity generation. Their study 𝜕𝑇 1 𝜕 𝜕𝑇 1 𝜕 𝜕𝑇 𝜕 𝜕𝑇
𝜌𝑐𝑝 = 𝑘𝑟 + 𝑘 + 𝑘 (1)
𝜕𝑡 𝑟 𝜕𝑟 𝜕𝑟 2
𝑟 𝜕𝜙 𝜕𝜙 𝜕𝑧 𝜕𝑧
considered well depths ranging from 4.2 to 6.8 km. Kharseh et al. [19]
modeled a U-tube heat exchanger configuration in an abandoned where 𝑟, 𝜙, 𝑧 are coordinates, 𝑇 is temperature, 𝑡 indicates time, and 𝑘,
oil well in Qatar using a commercial software package, employing 𝜌 and 𝑐𝑝 are ground’s thermal conductivity, density, and specific heat
Fourier’s three-dimensional heat conduction and assuming infinite capacity, respectively.
ground. They identified optimal conditions for maximizing electricity By assuming uniform ground properties in all directions, and consid-
generation, reporting a required mass flow rate exceeding 3.7 kg/s and ering heat transfer to be much more significant in the radial direction
a temperature difference between inlet and outlet of 15–17 ◦ C. Hu than in the vertical direction, the given equation can be simplified to a
et al. [38] simulated a single abandoned well as a DPCHE with a depth 1D form as follows:
of 3.5 km using COMSOL Multiphysics in the Hinton sedimentary basin 1 𝜕𝑇 𝜕2 𝑇 1 𝜕𝑇
= + (2)
(Alberta, Canada). They incorporated transient thermodynamic proper- 𝛼 𝜕𝑡 𝜕𝑟2 𝑟 𝜕𝑟
ties and verified their model using Ramey’s analytical approach over a where 𝛼 is thermal diffusivity defined as 𝑘∕(𝜌𝑐𝑝 ).
25-year simulation period, proposing an outlet water temperature of As depicted in Fig. 1(b), the ground, where the energy equation
40 ◦ C under optimal operational conditions. needs to be solved, exhibits a hollow cylindrical geometry. For this
As more attention is directed towards abandoned hydrocarbon case, two boundary conditions are required: one at the inner boundary
wells, there is an increasing demand for reliable mathematical frame- (𝑟 = 𝑎) and another at the outer boundary (𝑟 = 𝑏).
works and computing tools to model the coupled ground-wellbore At the inner side, a time-dependent convective boundary is consid-
system. The reliability of early developed models is questionable due ered, where the heat transfer fluid exchanges energy with the ground.
to their lack of verification and validation. In recent works, while At the outer side, a constant temperature boundary condition is applied.
some studies have focused on verification, validation against field data The boundary conditions are expressed mathematically as:
remains limited. Another unaddressed gap in the literature is the lack 𝜕𝑇 |
−𝑘 | = ℎ[𝑇∞ (𝑡) − 𝑇 (𝑟 = 𝑎, 𝑡)] (3)
of comprehensive criteria for classifying geothermal wells based on 𝜕𝑟 |𝑟=𝑎
quantified long-term heat extraction potential rather than initial ground
temperature. Most studies in the literature simulate limited scenarios 𝑇 (𝑟 = 𝑏, 𝑡 = 0) = 𝑇0 (4)
(e.g., single wellbore, single field) and durations; some are steady-state where ℎ is the convective coefficient of heat transfer, 𝑇∞ is heat transfer
models focusing solely on the wellbore, thus decoupling the ground fluid temperature, 𝑎 and 𝑏 represent the inner and outer diameters of
in simulations. In another group of studies, geothermal potential and the hollow cylinder. 𝑇0 is the initial ground temperature. The initial
classifications are primarily based on the Lindal diagram [39], which ground condition at time 𝑡 = 0 is provided to the model as a function
categorizes the application of geothermal energy based on the resource of the geothermal gradient.
temperature. Moreover, the potential of abandoned hydrocarbon wells The non-homogeneity of the convective boundary is mitigated by
in many geographical regions, including Canada, remains unassessed. introducing a linear temperature shift, expressed as:
Thus, the current study aims to address the aforementioned gaps by
employing a validated conjugate model to simulate a multitude of 𝜃(𝑟, 𝑡) = 𝑇 (𝑟, 𝑡) − 𝑇∞ (𝑡 = 0) = 𝑇 (𝑟, 𝑡) − 𝑇∞ (0) (5)
AOGWs for long-term heat extraction operations in Alberta, Canada. The Green function solves the shifted homogeneous heat transfer
In this paper, first, an efficient reduced-order analytical model is problem. Thus, the equation takes the following new form:
developed to solve the energy equation in the ground, coupled with
1 𝜕𝜃 𝜕 2 𝜃 1 𝜕𝜃
the repurposed abandoned well configured as a double-pip coaxial heat = + (6)
𝛼 𝜕𝑡 𝜕𝑟2 𝑟 𝜕𝑟
exchanger. Then, the model is validated against a geothermal field test
conducted for closed-loop geothermal heat extraction, sharing the same 𝜕𝜃 | ℎ ℎ
| − 𝜃(𝑎, 𝑡) = − (𝑇∞ (𝑡) − 𝑇∞ (0)) = 𝑔(𝑡) (7)
geometry and multiphysics phenomena. In the next step, 10,000 Monte 𝜕𝑟 |𝑟=𝑎 𝑘 𝑘
Carlo samples representing AOGWs in Alberta are generated based on
𝜃(𝑏, 𝑡) = 0 (8)
ground’s thermophysical properties, geothermal data, and well depth
data from the province of Alberta in Canada. These samples serve as where 𝑔(𝑡) is defined for simplicity. Eq. (6) is solved using Green
inputs for the model to simulate 20 years of heat extraction, following function, yielding the final expression of:
a parametric study that varies the inlet water temperature as a critical 𝑏 𝑡
operational parameter. 𝜃(𝑟, 𝑡) = (𝑇0 − 𝑇∞ (0)) 𝐺(𝑟, 𝜉, 𝑡) 𝑑𝜉 − 𝛼 𝑔(𝜏)𝐺(𝑟, 𝜉, 𝑡 − 𝜏) 𝑑𝜏 (9)
∫𝑎 ∫0

3
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Fig. 1. (a) A not-to-scale schematic of the configuration for repurposing a geothermal well in this study is presented. The configuration entails a double-pipe heat exchanger,
where the cold fluid is injected through the annular area and produced through the central insulated pipe. (b) The boundary conditions of the ground at the inner and outer sides
are mathematically expressed, with the ground temperature remaining unchanged beyond a distance of 50 m from the central axis of the wellbore.

where 𝐺 is the Green function defined as: as a function of inlet temperature [42]. Reynolds number is calculated
as:
𝜋2 ∑
∞ 𝜆4𝑛 𝐽12 (𝜆𝑛 𝑏)
𝐺(𝑟, 𝜉, 𝑡) = 𝜉𝐻𝑛 (𝑟)𝐻𝑛 (𝜉) exp(−𝛼𝜆2𝑛 ) (10) 4𝑚̇
2 𝑛=1
𝐵𝑛 𝑅𝑒 = (16)
𝜋𝐷ℎ 𝜇𝑓
and 𝐵𝑛 is:
where 𝑚̇ is water mass flow rate and 𝜇𝑓 is the viscosity of water.
[ℎ ]2 ( ℎ2 ) Friction factor in the annular area is computed as:
𝐵𝑛 = 𝜆2𝑛 𝐽 (𝜆 𝑎) + 𝜆𝑛 𝐽1 (𝜆𝑛 𝑎) − 𝜆2𝑛 + [𝜆𝑛 𝐽1 𝑏]2 (11)
𝑘 0 𝑛 𝑘2
𝑓𝑎𝑛𝑛 = (1.8 log10 (𝑅𝑒∗ ) − 1.5)−2 (17)
and 𝐻𝑛 is formulated as:
and 𝑅𝑒∗ is:
[ℎ ] ℎ [ ] [ ]
𝐻𝑛 (𝑟) = 𝑌 (𝜆 𝑎) + 𝜆𝑛 𝑌1 (𝜆𝑛 𝑎) 𝐽0 (𝜆𝑛 𝑟) − [ 𝐽0 (𝜆𝑛 𝑎) + 𝜆𝑛 𝐽1 (𝜆𝑛 𝑎)]𝑌0 (𝜆𝑛 𝑟) 1 + (𝐷𝑖𝑛 ∕𝐷𝑜𝑢𝑡 )2 ln(𝐷𝑖𝑛 ∕𝐷𝑜𝑢𝑡 ) + 1 − (𝐷𝑖𝑛 ∕𝐷𝑜𝑢𝑡 )2
𝑘 0 𝑛 𝑘 𝑅𝑒∗ = 𝑅𝑒 (18)
(12) [1 − (𝐷𝑖𝑛 ∕𝐷𝑜𝑢𝑡 )]2 ln(𝐷𝑖𝑛 ∕𝐷𝑜𝑢𝑡 )
Determining the pumping power to estimate the energy consump-
where the eigenvalues are calculated using below expression:
tion required for geothermal heat extraction is essential in the design
[ ℎ ] [ ℎ ]
−𝐽1 (𝜆𝑛 𝑏) 𝜆𝑛 𝑌1 (𝜆𝑛 𝑎)+ 𝑌0 (𝜆𝑛 𝑎) +𝑌1 (𝜆𝑛 𝑏) 𝜆𝑛 𝑌1 (𝜆𝑛 𝑎)+ 𝐽0 (𝜆𝑛 𝑎) = 0 (13) process. Therefore, Darcy–Weisbach equation [43] is used to calculate
𝑘 𝑘 the frictional pressure loss as:
The focus now shifts to detailing the governing equations for fluid
𝜌𝐿𝑣2
flow inside the wellbore. Water is injected into the well through the 𝛥𝑃 = 𝑓 (19)
2𝐷
annular area as the heat transfer fluid and returns to the surface
through the central pipe. The central pipe is assumed to be thermally where 𝑣 is velocity, 𝑔 is gravitational acceleration, and 𝑓 and 𝐷 are fric-
insulated to prevent any heat exchange between the incoming and tion factor and hydraulic diameter, respectively. Moody chart is used
for the inner pipe friction factor calculation, assuming typical stainless
returning flows.
steel roughness (𝜖 = 0.15 mm). The pumping power is calculated as the
The convective coefficient of heat transfer, denoted as ℎ in Eq. (3),
product of the pressure drop and the flow rate, divided by the assumed
is a crucial parameter to calculate, and it is determined as follows:
90% pump efficiency.
𝑘𝑓
ℎ = 𝑁𝑢 (14)
𝐷ℎ 2.2. Monte-Carlo modeling statistical analysis
where 𝑁𝑢 is Nusselt number, 𝑘𝑓 is the thermal conductivity of the
working fluid and 𝐷ℎ is the hydraulic diameter. The hydraulic diameter In this section, the selection of input parameters representing ther-
for concentric circular pipes is defined as the difference between the mophysical, geothermal, and geometrical characteristics of abandoned
inner and outer diameters of the pipes (also called tubing and casing), wells in Alberta, Canada, for use in the statistical study, is discussed
denoted as 𝐷ℎ = 𝐷𝑜𝑢𝑡 − 𝐷𝑖𝑛 . Here, the working fluid properties and the in detail. To simulate transient heat extraction from a single wellbore
geometrical characteristics are considered to be known. Thus, Nusselt using the model described in the previous section, two sets of input
number has to be calculated. parameters are required: the thermophysical properties of the ground
Gnielinski [41] proposed a Nusselt correlation for turbulent flows and the geometry of the wellbore.
inside the annular area of concentric pipes with 𝑅𝑒 > 4000, 𝐷ℎ ∕𝐿 ≤ 1, As per energy equation, thermal diffusivity (𝛼) of the ground has to
and 0.1 ≤ 𝑃 𝑟 ≤ 100 as follows: be known. This includes ground density (𝜌), thermal conductivity (𝑘)
[ ] and specific heat capacity (𝑐𝑝 ). The geothermal gradient (𝑔geo ), which
(𝑓𝑎𝑛𝑛 ∕8)(𝑅𝑒 − 1000)𝑃 𝑟 𝐷
𝑁𝑢 = 1 + ( ℎ )2∕3 (15) represents the increase in ground temperature with depth, is another
1 + 12.7(𝑓𝑎𝑛𝑛 ∕8)1∕2 (𝑃 𝑟2∕3 − 1) 𝐿 crucial input parameter, and its effect emerges in the initial ground
where 𝑓𝑎𝑛𝑛 is friction factor, 𝑅𝑒 is Reynolds number, 𝑃 𝑟 is Prandtl condition. Furthermore, the dimensions of the wellbore depth, central
number, and 𝐿 is pipe length. Prandtl number of water is determined tube, and outer pipe sizes have to be known.

4
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Table 1 2.3. Solution algorithm

The range of input parameters and their distribution types for generating Monte Carlo
samples are presented.
In the previous two sections, the governing equations, model inputs,
Parameter Min Max Distribution
and assumptions were explained in detail. Hence, in this section, the so-
𝑘 (W m−1 K−1 ) 0.9 5.69 Uniform
lution algorithm that links the previous sections together is explained.
𝜌 (kg m−3 ) 1600 3300 Uniform
𝑐𝑝 (kJ kg−1 K−1 ) 0.76 0.91 Uniform Fig. 3 shows the flowchart of the proposed framework. Assuming
𝑔𝑔𝑒𝑜 (◦ C m−1 ) 0.02 0.039 Uniform the Monte Carlo samples have already been generated, the stepwise
𝑍 (m) 500 7750 Biased solution algorithm is as follows:

1. Operational parameters including inlet temperature of the heat

transfer fluid (𝑇𝐻𝑇 𝐹 ) and its mass flow rate (𝑚̇ 𝐻𝑇 𝐹 ) are set;
These properties may vary across different geographical locations
2. Dimensionless numbers including Reynolds number (𝑅𝑒),
due to variations in lithology and the depth of the depleted oil and
Prandtl number (𝑃 𝑟), friction factor for the inner side and
gas reservoir. Here, a constant value for the pipe sizes (diameters) annular area (𝑓 ), and Nusselt number (𝑁𝑢), are calculated;
is assumed, while the remaining five are considered critical input 3. For the first abandoned well or sample (𝑖 = 1), the model reads
parameters, which are: (1) ground thermal conductivity, (2) ground the five critical inputs including ground properties, geothermal
density, (3) ground specific heat capacity, (4) geothermal gradient, (5) gradient, and well depth;
well depth. 4. Mesh is created considering 𝑁𝑥 divisions in the radial direction
Monte Carlo statistical analysis offers flexibility in handling complex and 𝑁𝑦 divisions axially, and initial ground conditions as per
systems, accounts for uncertainty, provides a comprehensive explo- Eq. (20) are set as a function of the geothermal gradient (𝑔𝑔𝑒𝑜 )
ration of potential outcomes, and aids in decision-making. However, and depth (𝑍);
the computational efficiency and sensitivity to input parameters are 5. Transient heat conduction equation is solved for the first gridline
regarded as drawbacks of this approach. In the current study, the (𝑁𝑦 = 1) to compute the radial ground temperature profile
aim is to model 10,000 abandoned wells and subsequently conduct 4 throughout the operational duration. Then, 𝑇𝐻𝑇 𝐹 is updated
additional sets of simulations, each varying one operational parameter. using Newton’s cooling law as the fluid exchanges heat with the
This results in a total of 50,000 simulations across individual wells over wellbore wall;
a 20-year operational period. 6. Marching over depth continues downward and the updated
The first challenge described is addressed in this study by using a 𝑇𝐻𝑇 𝐹 is employed at the convective boundary as the ambient
computationally efficient reduced-order model of the ground and well- temperature until reaching the final gridline;
bore. In response to the second challenge, it should first be noted that 7. The final calculated heat transfer fluid temperature at the last
for simulating 10,000 wells, a random combination of the five critical gridline is considered the outlet temperature (𝑇𝐻𝑇 𝐹 (𝑜𝑢𝑡𝑙𝑒𝑡) ), from
parameters is referred to as a Monte Carlo sample or an abandoned which the heat transfer rate is determined;
well. To generate such samples, at the first step it is essential to know 8. The simulations are completed upon modeling the final well
at least the minimum and maximum values of the desired parameters. (𝑖 = 10,000).
Such information is available from geological maps for thermophysical The significance of temporal and spatial mesh parameters in multi-
properties [44], and geoScout database for geothermal gradient [45] physics problems is well recognized. The impact of these factors on both
and well depth [46]. In the next step, ideally, it is desired to know accuracy and computational time is discussed in the model validation
the distribution of each parameter to generate the samples accordingly. section.
This ensures that more samples are generated around the peak or fewer
samples are generated from the lower frequencies (probabilities). 3. Model validation
The only available distribution among the five parameters is the
well depth data (𝑍). Thus, the mathematical expression for well depth The conjugate reduced-order model of the ground-wellbore, de-
distribution is determined using curve fitting and applied to gener- scribed in Section 2.1, is validated in this section. The results of a
ate depth values. For the remaining parameters, the uniform random field test conducted with the same geometrical setup and a similar
function from the NumPy library in Python is utilized. The range and purpose to our research have been selected to evaluate the performance
distribution type of each parameter are outlined in Table 1. Here, of our developed model. While the field test was conducted on a fresh
10,000 samples are generated to represent possible AOGWs. geothermal well, the geometry and multiphysics phenomena in the test
Fig. 2 visualizes the distribution of the generated Monte Carlo setup are identical to those in the targeted abandoned petroleum wells
samples. To provide a condensed overview of the generated samples, in the current study.
the density, thermal conductivity, and specific heat capacity are merged In the test setup, a double-pipe coaxial heat exchanger configuration
and presented as thermal diffusivity in Fig. 2(a). The distribution of is installed in the ground, with water injected through the annular
geothermal gradient and well depth samples is shown in Fig. 2(b) and space and withdrawn to the surface via the central tube as a closed-loop
(c). system, as shown in the schematic in Fig. 1(a). The main heat transfer
mechanism within the ground is conduction, while the water exchanges
The final important input parameter is the temperature profile of
heat with the ground through forced convection. Therefore, the prin-
the ground, which serves as the initial condition. For each individual
cipal thermophysical processes in the pilot test and the current study
well, a linear increase in ground temperature with depth, based on the
of 10,000 AOGWs are the same. Thus, at the initial stage, it is justified
geothermal gradient, is considered. Therefore, the temperature profile
to use the described mathematical model of transport phenomena in
is expressed as:
Section 2.1. The test is described below, followed by a discussion of
𝑇 (𝑧) = 𝑇𝑠 + 𝑔𝑔𝑒𝑜 × 𝑧 (20) the validation process.
A DPCHE geothermal setup with a depth of 500 meters was tested
where 𝑇𝑠 represents the temperature of the first ground layer and 𝑧 over a 456-hour period in a newly drilled geothermal well in Oita,
denotes the desired depth. 𝑇𝑠 is assumed to be the average value of Japan [48]. Water, circulating in the system with a mass flow rate
5.6 ◦ C, representing the shallow ground temperature in the province of of 1.66 kg/s and an inlet temperature of 70 ◦ C, was used during
Alberta after passing the active layer, as indicated by Tareen et al. [47]. the experiment. Operational parameters were selected to meet the

5
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Fig. 2. Distribution of Monte-Carlo samples used as inputs for the reduced-order model: (a) thermal diffusivity, (b) geothermal gradient, (c) well depth.

Table 2
The characteristics of the wellbore and ground during the geothermal heat extraction
experiment conducted in Oita, Japan are presented [48].
Material properties
1. Ground:
Density 2490 (kg m−3 )
Thermal conductivity 2.0 (W m−1 K−1 )
Specific heat capacity 1.1 (kJ kg−1 K−1 )
2. Water:
Density 998.2 (kg m−3 )
Thermal conductivity 0.6 (W m−1 K−1 )
Specific heat capacity 4.182 (kJ kg−1 K−1 )
Geometry
Depth 500 (m)
Casing OD 177.8 (mm)
Casing ID 161.7 (mm)
Tubing OD 114.3 (mm)
Tubing ID 103.88 (mm)
Operational conditions
Inlet water temperature 70 (◦ C)
Mass flow rate 1.66 (kg s−1 )

temperature profile of the ground, measured in the field, is curve-

fitted, and the resulting mathematical expression is incorporated into
the reduced-order model, as illustrated by a dashed-line in Fig. 4(a).
The measured outlet water temperatures from the field test are
shown alongside the model predictions in Fig. 4(b). Error values for
days 2, 5, 10, and 19 (the final day) of heat extraction are calculated
and compared. The model exhibits some inaccuracies during the initial
hours of heat extraction limited to the first 48 h. This is attributed to the
transient behavior of the system. Using the model for very early times
does not guarantee reliable results. It is important to note that in this
model, unlike numerical simulations, each time-step is independent of
the previous one, meaning such behavior cannot influence times before
Fig. 3. Flowchart illustrates the proposed solution algorithm. or after any given time-step. As the system reaches a semi-steady state,
the model’s performance significantly improves and closely aligns with
the observed data for the rest of the operation period. In particular,
the error value is 2.2% after 48 h, which reduces to 0.45% by 120 h.
specific requirements of electricity generation scenarios, given the high The error value slightly fluctuates and reaches 0.65% at 240 h, which
initial ground temperature. The outlet water temperature was measured may be attributed to measurement errors observed in the field data.
within the period. Pokhrel et al. [48] employed computational fluid Eventually, the error value reaches 0.005% at the end of the recorded
operation time.
dynamics (CFD) to model this geothermal system. Their CFD model
In the validation process, it is essential to employ a sufficient
was validated with minimal error and closely matched field data in pre-
number of eigenvalues and axial layers in the developed model. The
dicting outlet water temperature. They used the measured temperature selection of different temporal time-step values does not affect the
profile as shown in Fig. 4 (a) [49] as initial ground conditions, while results. Therefore, a daily time-step size is considered for validation
other inputs such as material properties, geometrical, and operational and the rest of the simulations. In Table 3, the deviations in the final-
parameters representing the system are listed in Table 2. The initial time measured outlet water temperature when varying the number of

6
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Fig. 4. (a) Comparison between the initial ground temperature measured in the field and the input provided to the model for validation purposes. (b) Results of the model
validation process for outlet water temperature.

Table 3
The deviation of the model in predicting outlet water temperature as the number of eigenvalues and axial layers changes is presented as well as the run time. For model validation,
100 eigenvalues and 100 axial layers are used.
Eigenvalues no. Deviation Axial layers no. Deviation Execution time (s)
2 9.2% 5 2.3% 180
10 2.6% 20 0.4% 735
20 0.3% 50 0.14% 1824
50 0% 100 0.005% 3696
100 – 500 0.005% 17 989

eigenvalues and axial layers, along with the required execution time, 4.1. Evaluation of geothermal energy potential from AOGWs in Alberta,
are presented. For eigenvalues, no change is observed once the number Canada
reaches 50. Since the execution time remains unchanged, it is increased
to 100 to ensure reliability. On the other hand, in the model validation, The findings from the simulation of 10,000 abandoned wells under
100 axial layers are used. Increasing the axial layers from 100 to 500 constant operational conditions, referred to as the base case analysis,
enhances the accuracy slightly but multiplies the execution time by are thoroughly examined in this section. The base case operational
conditions for heat extraction over 20 years for each individual well
4.87.
include a water flow rate of 1.66 kg/s and an inlet water temperature
In conducting a statistical analysis like the current study, where of 15 ◦ C. The casing and tubing sizes are constant for all the cases
50,000 cases are simulated, run time is crucial. To reduce the execution and are the same as values reported in Table 2. As a result, a constant
time of modeling an abandoned well by 95% while accepting an error Reynolds number of 45,822 and a corresponding Nusselt number of 265
of 2.3%, five axial layers are chosen for the simulations conducted in are obtained.
the upcoming sections. The reported error value is an underestimation, In order to classify the abandoned wells based on the application
making the output results slightly conservative. of extracted thermal power, outlet water temperature is employed as
the main criterion. Wells that have outlet water temperatures (𝑇out )
exceeding 90 ◦ C are regarded as suitable for generating electricity
4. Results and discussion using a geothermal-driven organic Rankine cycle (ORC). Outlet water
temperatures between 15 ◦ C to 90 ◦ C are considered suitable for
heating applications. This range is divided into three equal intervals:
In this section, the results of the assessment of geothermal potential
15 ≤ 𝑇out < 40 ◦ C for low-grade heating (i.e., heat pump applications),
from AOGWs in Alberta are analyzed using the described solution 40 ≤ 𝑇out < 65 ◦ C for medium-grade heating, and 65 ≤ 𝑇out < 90 ◦ C for
algorithm. The wellbore and ground characteristics are based on the high-grade heating.
10,000 Monte Carlo samples generated in Section 2.2. For the oper- Initially, the results indicate that 20.02% of the total wells have out-
ational parameters, a mass flow rate of 1.66 kg/s and an inlet water let water temperatures lower than the inlet water temperature, whereas
temperature of 15 ◦ C is used for the base analysis. Each individual well 79.98% exhibit higher outlet water temperatures. The first parameters
is simulated over 20 years of operation. It should be noted that the to review are the Monte Carlo percentiles listed in Table 4. The 𝑃10 ,
flow rate used is the same as that in the field test [50] in the model 𝑃50 , and 𝑃90 values of outlet water temperature and thermal power
validation section, yielding the same Reynolds number. are presented to clarify technical uncertainties for decision-makers
The inlet temperature used in the field test (i.e., 70 ◦ C) is relatively considering prospective outcomes. The net energy values are closely
aligned with the gross energy values due to the minimal pressure loss
high since the site is located near a volcanic region, unlike many oil
from friction, resulting in similar trends for both.
and gas reservoirs that often have lower bottom-hole temperatures.
A 𝑃10 value of 15.23 ◦ C for outlet water temperature and 0.258 GWh
Therefore, to adopt a more realistic and widespread approach, an inlet
for net thermal energy indicates that by retrofitting an abandoned well
water temperature of 15 ◦ C is selected for the base case. Ideally, into a geothermal well, there is a 90% probability that these parameters
optimizing the inlet water temperature based on the initial ground will exceed the given values over a 20-year period. As per 𝑃90 values,
temperature is necessary, depending on the application of the produced there is a 10% chance that the outlet temperature and thermal power
energy. In the parametric study section, the inlet temperature is varied will exceed 33.71 ◦ C and 22.39 GWh, respectively, which represent the
from 5 ◦ C to 25 ◦ C to further investigate its impact on outlet water optimistic estimates. The median scenario, 𝑃50 which is more likely to
temperature and extracted thermal energy. occur, indicates an outlet water temperature of 17.67 ◦ C and a thermal

7
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Table 4 The properties of 7469 wells within the low-grade heating category
Monte Carlo percentiles of 7,998 abandoned wells identified with positive potential for
are presented in Table 6. The outlet water temperature range of 15
geothermal energy after initial screening.
to 40 ◦ C is divided into 5-degree increments to improve resolution for
Parameter 𝑃10 𝑃50 𝑃90
this densely populated group of wells. In this range, starting from the
𝑇out (◦ C) 15.23 17.67 33.71
highest outlet temperatures at a depth of nearly 3 km, with an average
𝐸Gross (GWh) 0.275 3.20 22.44
𝐸Net (GWh) 0.258 3.18 22.39
progressive lowering of depth by 0.47 km, the ground temperature
decreases by 19.5%, outlet water temperature decreases by 18.2%, and
produced thermal energy decreases by 42.7%.
Fig. 6 presents the box plots of outlet water temperature (𝑇out ), net
power of 3.18 GWh, where there is an equal chance of production being thermal energy (𝐸𝑁𝑒𝑡 ), and heat transfer rate per unit length (𝑄∕𝑍) for
either less or more than these values. each of the well groups, further highlighting the range and distribution
The mentioned percentile values indicate a high potential for heat- of the output results. The range and quartiles for each box in Fig. 6 are
ing, particularly for heat pump applications, due to the greater prob- presented in the Table 7.
ability of achieving small temperature increments. This statement is In Fig. 6(a) and (b), as the outlet water temperature and pro-
further supported by illustrating the distribution of outlet water tem- duced thermal energy increase, the length of the box denoting the
perature (𝑇out ), gross thermal energy (𝐸Gross ), and net thermal energy interquartile range (between 𝑄1 and 𝑄3 ) also increases, indicating
(𝐸Net ) produced over 20 years of operation, plotted against input greater variation in wells producing higher temperatures. Hence, op-
parameters namely, ground thermal conductivity (𝑘), ground specific erators might need to be more vigilant in their strategies when dealing
heat capacity (𝑐𝑝 ), ground density (𝜌), geothermal gradient (𝑔geo ), and with wells that have higher outlet water temperatures. The minimum
well depth (𝑍), as shown in Fig. 5. value for net energy in the low-grade group indicates that there is
The visible trend in Fig. 5 shows that depth is directly proportional a possibility of consuming more energy than is produced in some
to the output parameters. It can be seen that for ORC electricity cases. However, it is observed that such cases are limited in number.
generation, a minimum depth of 4.43 km is required, while high-grade Screening the wells based on the ratio of pumping power to thermal
heating requires a minimum depth of 3.48 km. Medium-grade heating energy produced shows that among the 7469 wells in the low-grade
requires a minimum depth of 2.4 km, and for heat pump applications, category, this ratio is below 10% for 7006 wells. If the threshold
the minimum depth is 530 m. The next visible trend is observed in increases to 30%, the number of wells meeting the condition becomes
the geothermal gradient at higher outlet water temperatures, with 83% 7316. Thus, around 150 wells fall out of this range, which, in practice,
of candidate wells for ORC and 68% of high-grade heating candidates means that a higher portion of pumping power is required for heat
exhibiting a minimum geothermal gradient of 0.03 ◦ C/m. There is a
extraction that may not be economically justified.
wider distribution observed for lower outlet water temperatures, which
According to Fig. 6(c), the minimum heat transfer rates per unit
complicates the identification of a definitive trend for geothermal
length for each group of wells are as 86.91 W/m for ORC, 60.12 W/m
gradient.
for high-grade heating, and 34.47 W/m for medium-grade heating. The
The focus now shifts to ground thermophysical properties, where
minimum value of 𝑄∕𝑍 for low-grade heating is influenced by wells
it is observed that in 96% of cases, the thermal conductivity of rocks
that minimally increase the outlet water temperature. However, based
for ORC application exceeds 2.5 (W m−1 K−1 ), indicating that sites
on 𝑄1 value, it can be seen that 25% of the 7469 wells can deliver less
predominantly composed of rock types such as shale, mudstone, and
than 5.46 W/m. Examining the median (𝑄2 ), it is observed that the
siltstone may not be ideal for ORC electricity generation. In the high-
average heat transfer rate per unit length (𝑄∕𝑍) increases sharply by
grade and medium-grade heating ranges, 87% and 84% of the cases
434% from low-grade to medium-grade heating and then increases on
further illustrate this trend, with data points visibly spreading towards
average by 30%, with an expected average 𝑇out rise of 30 ◦ C up to wells
higher thermal conductivity values. The region where most high-grade
in ORC category.
and medium-grade cases tend to cluster corresponds to abandoned
The reported energy and outlet temperatures are based on the final
oil and gas sites characterized by formations such as limestone and
dolomite, which exhibit high thermal conductivity values approaching time-step of transient heat extraction after 20 years. Consequently,
the upper limit of 5.69 (W m−1 K−1 ). due to the quasi-linear evolution of the heat transfer rate profile,
For the specific heat capacity, the cases with the highest outlet the reported thermal power values are conservative estimates. Fig. 7
temperature values are primarily located near the lower bound, with illustrates the yearly temporal evolution of outlet water temperature
approximately equal numbers distributed around the median value and heat transfer rate over the 20-year period for selected samples,
of 835 J/kg K−1 in both halves. The distribution of ground density featuring different outputs from ORC to low-grade wells. Comparing the
shows that cases with densities above and below the median value of reported final values while considering the quasi-linear profile reveals
2450 kg/m3 are roughly equal across ORC, high-grade, and medium- an average underestimation of 2 to 8% according to Table 8.
grade heating scenarios. Thus, no clear trend is evident for ground
specific heat capacity and density. 4.2. Effect of inlet water temperature
According to the criterion described for classifying AOGWs based
on outlet water temperature (𝑇out ), 0.49% of the wells (49 wells) In this section, the inlet water temperature varies from 5 to 25 ◦ C at
exhibit potential for geothermal-driven ORC. For high-grade heating 5-degree increments to investigate the impact on outputs. Thus, there
applications, 90 wells qualify, followed by 390 wells in the medium- is a total of 50,000 data points for each of the output parameters
grade range, with the majority, comprising 7469 wells, falling within including outlet water temperature (𝑇out ), and heat transfer rate (𝑄).
the low-grade range. Evaluating 𝑇out values shows that for 𝑇𝑖𝑛 = 5◦ C and 10 ◦ C, 100% of
The average system characteristics and outputs of the wells qualified the wells demonstrate a value exceeding 𝑇in . As the inlet temperature
for ORC electricity generation, high-grade and medium grade heating increases from 15 to 25 ◦ C, this ratio declines to 79.98%, 54.51%, and
are presented in Table 5. Noting that depth is the most correlated 37.07% across the wells. The substantial change occurs in reducing the
parameter with 𝑇out , it is observed that for every 1.05 km reduction number of low-grade wells from around 9500 wells to 3000 wells as
in depth from ORC to medium-grade heating, the bottom-hole tempera- 𝑇𝑖𝑛 rises from 5 to 25 ◦ C. The variation in the number of wells in other
ture (𝑇bht ) is, on average, 24% colder, and the outlet water temperature groups is smaller, with increases of approximately 3 for ORC, 20 for
(𝑇out ) decreases at a rate of 33%. Moreover, there is an average reduc- high-grade heating, and 200 for medium-grade heating.
tion in thermal energy produced by 43%, which represents the decline Fig. 8 shows quartiles 1 to 3, indicating the distribution of aban-
in temperature difference between the inlet and outlet. doned wells across different groups based on their application types

8
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Fig. 5. Outlet water temperature, gross thermal energy, and net thermal energy from the base case analysis are plotted against the thermophysical properties of the ground (a),
geothermal gradient (b), and well depths (c).

Table 5
Average thermophysical, geothermal, and geometrical characteristics, along with output parameters and the number of wells in each category, are presented.
Scenario 𝛼 𝑔geo 𝑍 𝑇bht (◦ C) 𝑇out (◦ C) 𝐸Gross (GWh) 𝐸Net (GWh) Well no.
(m2 s−1 ) (◦ C m−1 ) (m)
ORC 1.94 × 10−6 0.0341 5690 199.34 109.90 113.85 113.74 49
H-G heating 2.16 × 10−6 0.0320 4628 150.54 74.81 71.75 71.66 90
M-G heating 2.04 × 10−6 0.0319 3489 115.10 48.92 40.69 40.63 390

Table 6
Average properties of 7,469 wells suitable for low-grade heating, divided into five groups based on outlet water temperature.
Range (◦ C) 𝛼 𝑔geo 𝑍 𝑇bht 𝑇out 𝐸Gross (GWh) 𝐸Net (GWh) Well no.
(m2 s−1 ) (◦ C m−1 ) (m) (◦ C) (◦ C)
35–40 1.98 × 10−6 0.0306 2989 95.76 37.29 26.74 26.68 192
30–35 1.90 × 10−6 0.0307 2672 86.35 32.40 20.87 20.82 309
25–30 1.92 × 10−6 0.0310 2257 74.34 27.26 14.71 14.66 596
20–25 1.74 × 10−6 0.0304 1884 61.93 22.15 8.57 8.54 1202
15–20 1.61 × 10−6 0.0301 1110 38.27 16.58 1.9 1.88 5170

Fig. 6. Box plots of (a) outlet water temperature, (b) thermal power, and (c) heat transfer rate per unit length for each group, defined as ORC for electricity generation, high-grade
(H-G) heating, medium-grade (M-G) heating, and low-grade (L-G) heating applications.

9
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Table 7
Range and quartiles of the outputs for four groups of wells, defined based on outlet water temperature: ORC for electricity generation, high-grade (H-G) heating, medium-grade
(M-G) heating, and low-grade (L-G) heating applications.
Parameter 𝑇out (◦ C) 𝐸Net (GWh) 𝑄∕𝑍 (W/m)
Scenario Min 𝑄1 𝑄2 𝑄3 Max Min 𝑄1 𝑄2 𝑄3 Max Min 𝑄1 𝑄2 𝑄3 Max
ORC 90.11 99.32 108.08 117.06 153.61 90.02 101.05 111.56 122.32 166.12 86.91 107.60 119.07 125.47 151.30
H-G heating 65.07 69.48 73.40 79.60 89.69 59.95 65.28 69.96 77.4 89.50 60.12 79.88 94.98 103.31 123.29
M-G heating 40.06 43.24 47.39 53.24 64.94 30.01 33.80 38.78 45.80 59.84 34.47 60.15 70.15 78.20 107.40
L-G heating 15.01 15.72 17.33 21.34 39.95 −0.01 0.84 2.77 7.57 29.87 5.2 × 10−3 5.46 13.12 25.82 74.79

Fig. 7. The evolution of outlet water temperature (a - d) and heat transfer rate (e - h) for three selected samples in each group of wells.

Table 8 the extracted thermal power decreases on average for ORC by 4.6% for
The difference between real and conservative assumption of thermal energy (KWh)
every 5-degree increment in temperature. Similarly, the decline rates
over 20 years of operation. These values are based on the integration of the curves
presented in Fig. 7.
for high-grade, medium-grade, and low-grade systems are 8%, 12%,
Group label Difference
and 10%, respectively.
Fig. 9 illustrates the variation of net thermal energy produced over
ORC electricity Well #209 4.16%
Well #773 5.48% time and the cumulative pumping power required over time for each
Well #1749 6.05% group of wells as the inlet water temperature changes. Increasing the
High-grade heating Well #344 5.13% inlet water temperature decreases the total thermal power produced
Well #769 6.65% in all groups of wells, as indicated by the black bars. On average,
Well #924 8.00% this reduction is 33% for every 5 ◦ C increments in inlet temperature.
Medium-grade heating Well #119 6.52% Meanwhile, the average required pumping power for ORC and high-
Well #165 7.91% grade heating groups experiences minimal change, largely since most
Well #254 5.41%
of these wells are located at greater depths and the number of wells in
Low-grade heating Well #3 2.83%
these groups does not change significantly. The average well depths in
Well #30 4.46%
Well #55 5.20%
these groups vary between 50 and 60 m.
In medium-grade heating, the required pumping power slightly
decreases with increasing inlet water temperature, indicating the elim-
ination of some deep wells with colder ground temperatures. This
and their corresponding heat transfer rates. In the first row depicting change is minimal, limited to 6%, resulting from variations in average
quartiles of outlet water temperatures versus inlet water temperature, depth up to 160 m. However, as previously explained, the increase in
the ORC, high-grade, and medium-grade wells exhibit minimal changes inlet temperature decreases the number of wells categorized as low-
as the inlet water temperature varies. These changes are limited to a grade heating, resulting in the elimination of more shallow wells. This
maximum of 1–2 ◦ C across the same figure. However, this variation is change significantly impacts the required pumping power, increasing it
more noticeable in low-grade heating, where more wells are excluded by 51%. The reason is the increase in average well depth, which rises
as the inlet water temperature increases. The same trend persists when from 1.2 km to 1.9 km as 𝑇in increases from 5 to 25 ◦ C.
transitioning from quartile 1 to quartile 3, with the figures shifting up-
wards by an average of 10–12 ◦ C. In the second row, the impact of inlet 5. Conclusion
water temperature on the heat transfer rate becomes more apparent,
as an increase in inlet water temperature decreases the temperature This study evaluates the geothermal energy potential from aban-
difference, resulting in lower heat transfer rate values. Consequently, doned petroleum wells in Alberta, Canada, using statistical analysis. A

10
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

Fig. 8. Box charts divided based on four 𝑇𝑜𝑢𝑡 ranges, each corresponding to different 𝑇𝑖𝑛 values from 5 to 25 ◦ C.

Fig. 9. Average produced thermal energy and required pumping power required for each group of wells are shown.

computationally efficient model is developed for a double-pipe heat ex- • A minor portion, approximately 0.5% of the 10,000 simulated
changer configuration, validated, and utilized as a multiphysics model wells, show potential for electricity generation, while the vast
to quantify the geothermal potential. The investigation focuses on majority, about 79.5%, are suitable for heating.
simulating 10,000 abandoned wells over 20 years and assessing the • To generate electricity from abandoned wells, typical require-
impact of inlet water temperature as a key operational parameter. ments include a geothermal gradient higher than 0.034 ◦ C/m, a
The key findings from the assessments are summarized as below: depth of 5.6 km, and a bottom-hole temperature of 199 ◦ C, which
results in an outlet temperature of 110 ◦ C.
• The base analysis in this study reveals a promising geothermal • A linear correlation is observed between well depth and outlet
potential exists for nearly 80% of the simulated abandoned wells. water temperature. As the depth decreases from 7.7 km to 3 km,

11
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

thermal energy production drops by 43% per km. This decline References
rate also applies to wells located below 3 km for every 0.5 km
reduction in depth. [1] Zhao N, You F. Can renewable generation, energy storage and energy ef-
ficient technologies enable carbon neutral energy transition? Appl Energy
• Increasing the inlet water temperature from 5 to 25 ◦ C reduces 2020;279:115889.
the number of wells qualified for low-grade heating to one-third, [2] O’Sullivan M, Gravatt M, Popineau J, O’Sullivan J, Mannington W, McDow-
while the change in the number of wells producing water at ell J. Carbon dioxide emissions from geothermal power plants. Renew Energy
2021;175:990–1000.
higher outlet temperatures remains negligible.
[3] Pfenninger S, Keirstead J. Renewables, nuclear, or fossil fuels? Scenarios for
• Increasing the inlet water temperature in 5-degree increments re- Great Britain’s power system considering costs, emissions and energy security.
duces the heat transfer rate for ORC applications by an average of Appl Energy 2015;152:83–93.
23%. This effect gradually increases to 50%–60% for high-grade [4] Chapman AJ, McLellan BC, Tezuka T. Prioritizing mitigation efforts considering
co-benefits, equity and energy justice: Fossil fuel to renewable energy transition
to low-grade heating applications. pathways. Appl Energy 2018;219:187–98.
[5] McTigue JD, Wendt D, Kitz K, Gunderson J, Kincaid N, Zhu G. Assess-
While the current study quantified the geothermal potential of ing geothermal/solar hybridization–Integrating a solar thermal topping cycle
AOGWs, future research could focus on optimizing inlet water temper- into a geothermal bottoming cycle with energy storage. Appl Therm Eng
2020;171:115121.
ature and flow rate for heating and electricity generation, considering [6] Pokhrel S, Amiri L, Poncet S, Ghoreishi-Madiseh SA. Reduced order 1+ 3D
ground temperature. Moreover, future research could investigate how numerical model for evaluating the performance of solar borehole thermal energy
passive phase change tools, such as two-phase closed thermosyphons, storage systems. J Energy Storage 2023;66:107503.
[7] Lukawski MZ, Silverman RL, Tester JW. Uncertainty analysis of geothermal well
can enhance the efficiency of heat extraction from AOGWs at relatively
drilling and completion costs. Geothermics 2016;64:382–91.
shallow depths. Given the low ground temperatures in shallow wells, [8] Ashena R. Analysis of some case studies and a recommended idea for geothermal
this study eliminated wells with depths below 500 m. Considering energy production from retrofitted abandoned oil and gas wells. Geothermics
the inventory of wells at shallow depths, using phase change heat 2023;108:102634.
[9] Santos L, Taleghani AD, Elsworth D. Repurposing abandoned wells for
transfer to significantly increase the convection coefficient of heat
geothermal energy: Current status and future prospects. Renew Energy
transfer appears to be an attractive choice. Additionally, addressing 2022;194:1288–302.
issues related to data availability, assessing socio-economic impacts, [10] Kurnia JC, Shatri MS, Putra ZA, Zaini J, Caesarendra W, Sasmito AP. Geothermal
and conducting studies on methane mitigation are intriguing aspects energy extraction using abandoned oil and gas wells: Techno-economic and
policy review. Int J Energy Res 2022;46(1):28–60.
of repurposing AOGWs into geothermal wells. [11] El Hachem K, Kang M. Methane and hydrogen sulfide emissions from abandoned,
active, and marginally producing oil and gas wells in Ontario, Canada. Sci Total
Environ 2022;823:153491.
CRediT authorship contribution statement [12] Jello J, Baser T. Utilization of existing hydrocarbon wells for geothermal system
development: A review. Appl Energy 2023;348:121456.
[13] Williams JP, Regehr A, Kang M. Methane emissions from abandoned oil and gas
Mohammad Zolfagharroshan: Writing – original draft, Visualiza- wells in Canada and the United States. Environ Sci Technol 2020;55(1):563–70.
tion, Validation, Software, Methodology, Investigation, Formal analy- [14] Cheng W-L, Li T-T, Nian Y-L, Wang C-L. Studies on geothermal power generation
sis, Data curation, Conceptualization. Minghan Xu: Writing – review using abandoned oil wells. Energy 2013;59:248–54.
[15] Cheng W-L, Liu J, Nian Y-L, Wang C-L. Enhancing geothermal power generation
& editing, Validation, Methodology, Formal analysis, Data curation.
from abandoned oil wells with thermal reservoirs. Energy 2016;109:537–45.
Jade Boutot: Writing – review & editing, Methodology, Investigation, [16] Bowman LV, El Hachem K, Aljabre O, Kang M. Methane emissions from
Formal analysis, Data curation. Ahmad F. Zueter: Writing – review & abandoned and suspended oil and gas wells in alberta and saskatchewan. In:
editing, Validation, Methodology. Muhammad S.K. Tareen: Writing GeoConvention. 2022, p. 4.
[17] Fan H, Zhang L, Wang R, Song H, Xie H, Du L, et al. Investigation on geother-
– review & editing, Investigation, Data curation. Mary Kang: Writing mal water reservoir development and utilization with variable temperature
– review & editing, Supervision, Resources, Project administration, regulation: a case study of China. Appl Energy 2020;275:115370.
Funding acquisition, Conceptualization. Agus P. Sasmito: Writing – re- [18] Ebrahimi D, Kolahi M-R, Javadi M-H, Nouraliee J, Amidpour M. A brief survey
on case studies in geothermal energy extraction from abandoned wells. In:
view & editing, Supervision, Resources, Project administration, Funding
Utilization of thermal potential of abandoned wells. Elsevier; 2022, p. 75–96.
acquisition, Conceptualization. [19] Kharseh M, Al-Khawaja M, Hassani F. Optimal utilization of geothermal
heat from abandoned oil wells for power generation. Appl Therm Eng
2019;153:536–42.
Declaration of competing interest [20] Wang Z, Ning Z, Guo W, Zhan J, Zhang Y. Study of fracture monitoring and
heat extraction evaluation in geothermal reservoir modified by abandoned well
pattern: Numerical models and case studies. Energy 2024;296:131144.
The authors declare that they have no known competing finan- [21] Kurnia JC, Putra ZA, Muraza O, Ghoreishi-Madiseh SA, Sasmito AP. Numer-
cial interests or personal relationships that could have appeared to ical evaluation, process design and techno-economic analysis of geothermal
influence the work reported in this paper. energy extraction from abandoned oil wells in Malaysia. Renew Energy
2021;175:868–79.
[22] Liu X, Falcone G, Alimonti C. A systematic study of harnessing low-temperature
geothermal energy from oil and gas reservoirs. Energy 2018;142:346–55.
Data availability
[23] Bennett K, Li K, Horne R, et al. Power generation potential from coproduced
fluids i the los angeles basin [Ph.D. thesis], CA, USA: Stanford University
Data will be made available on request. Stanford; 2012.
[24] Akhmadullin I. Utilization of co-produced water from oil production: Energy gen-
eration case. In: SPE health, safety, security, environment, & social responsibility
conference. SPE; 2017, p. D031S017R003.
Acknowledgments
[25] Jello J, Khan M, Malkewicz N, Whittaker S, Baser T. Advanced geothermal
energy storage systems by repurposing existing oil and gas wells: A full-scale
The first and fifth authors give thanks McGill Engineering Doc- experimental and numerical investigation. Renew Energy 2022;199:852–65.
[26] ThinkGeoEnergy. First closed-cycle geothermal heat plant set up in Hungary.
toral Award (MEDA). The second author acknowledges the support 2024, https://www.thinkgeoenergy.com/first-closed-cycle-geothermal-heat-
from Vanier Canada Graduate Scholarship (Vanier CGS) funding. The plant-set-up-in-hungary/. [Accessed 18 June 2024].
third author gives thanks to the McGill Vadasz Scholars Program and [27] Zolfagharroshan M, Khamehchi E. A rigorous approach to scale formation and
deposition modelling in geothermal wellbores. Geothermics 2020;87:101841.
the Canada Graduate Scholarship-Doctoral (CGS D) from the Natural
[28] Zolfagharroshan M, Khamehchi E. Mineral scaling impact on petrophysical
Sciences and Engineering Research Council of Canada (NSERC). The properties of reservoir rock in a geothermal field located in Northwestern Iran.
authors are also grateful for funding from NSERC (RGPIN-2021-02901). SPE J 2024;29(02):1029–44.

12
M. Zolfagharroshan et al. Applied Energy 375 (2024) 124103

[29] Radioti G, Sartor K, Charlier R, Dewallef P, Nguyen F. Effect of undisturbed [40] Cole K, Beck J, Haji-Sheikh A, Litkouhi B. Heat conduction using Greens
ground temperature on the design of closed-loop geothermal systems: A case functions. CRC Press; 2010.
study in a semi-urban environment. Appl Energy 2017;200:89–105. [41] Gnielinski V. Turbulent heat transfer in annular spaces—a new comprehensive
[30] Yang H, Li J, Zhang H, Jiang J, Guo B, Gao R, et al. Thermal behavior prediction correlation. Heat Transf Eng 2015;36(9):787–9.
and adaptation analysis of a reelwell drilling method for closed-loop geothermal [42] Hefni MA, Xu M, Zueter AF, Hassani F, Eltaher MA, Ahmed HM, et al. A 3D
system. Appl Energy 2022;320:119339. space-marching analytical model for geothermal borehole systems with multiple
[31] Qin CC, Loth E. Isothermal compressed wind energy storage using abandoned heat exchangers. Appl Therm Eng 2022;216:119027.
oil/gas wells or coal mines. Appl Energy 2021;292:116867. [43] Agson-Gani PH, Zueter AF, Xu M, Ghoreishi-Madiseh SA, Kurnia JC, Sasmito AP.
[32] Dai C, Li J, Shi Y, Zeng L, Lei H. An experiment on heat extraction from a Thermal and hydraulic analysis of a novel double-pipe geothermal heat ex-
deep geothermal well using a downhole coaxial open loop design. Appl Energy changer with a controlled fractured zone at the well bottom. Appl Energy
2019;252:113447. 2022;310:118407.
[33] Davis AP, Michaelides EE. Geothermal power production from abandoned oil [44] Hamilton WN, Langenberg CW, Price MC, Chao DK. Geological map of alberta.
wells. Energy 2009;34(7):866–72. 1998, Alberta Energy and Utilities Board, EUB/AGS Map 236.
[34] Bu X, Ma W, Li H. Geothermal energy production utilizing abandoned oil and [45] Huang KY, Hickson CJ, Cotterill D, Champollion Y. Geothermal assessment of
gas wells. Renew Energy 2012;41:80–5. target formations using recorded temperature measurements for the Alberta No.
[35] Cheng W-L, Li T-T, Nian Y-L, Xie K. Evaluation of working fluids for geothermal 1 geothermal project. Appl Sci 2021;11(02):608.
power generation from abandoned oil wells. Appl Energy 2014;118:238–45. [46] Alberta Energy Regulator. Alberta Energy Regulator. 2024, URL https://www.
[36] Templeton J, Ghoreishi-Madiseh S, Hassani F, Al-Khawaja M. Abandoned aer.ca/. [Accessed 03 July 2024].
petroleum wells as sustainable sources of geothermal energy. Energy [47] Tareen MS, Zueter AF, Zolfagharroshan M, Xu M, Sasmito AP. Seasonal ground
2014;70:366–73. cold energy storage potential for data center cooling using thermosyphon: A
[37] Wight N, Bennett N. Geothermal energy from abandoned oil and gas wells using comparative study of five cities in Canada for carbon footprint reduction. J
water in combination with a closed wellbore. Appl Therm Eng 2015;89:908–15. Energy Storage 2024;84:111486.
[38] Hu X, Banks J, Wu L, Liu WV. Numerical modeling of a coaxial borehole heat [48] Pokhrel S, Sasmito AP, Sainoki A, Tosha T, Tanaka T, Nagai C, et al. Field-scale
exchanger to exploit geothermal energy from abandoned petroleum wells in experimental and numerical analysis of a downhole coaxial heat exchanger for
Hinton, Alberta. Renew Energy 2020;148:1110–23. geothermal energy production. Renew Energy 2022;182:521–35.
[39] Watson SM, Falcone G, Westaway R. Repurposing hydrocarbon wells for geother- [49] Zolfagharroshan M, Zueter AF, Tareen MS, Xu M, Sasmito AP. Two-phase closed
mal use in the UK: The onshore fields with the greatest potential. Energies thermosyphons (TPCT) for geothermal energy extraction: A computationally
2020;13(14):3541. efficient framework. Appl Therm Eng 2024;248:123205.
[50] Pokhrel S, Amiri L, Zueter A, Poncet S, Hassani FP, Sasmito AP, et al. Thermal
performance evaluation of integrated solar-geothermal system; a semi-conjugate
reduced order numerical model. Appl Energy 2021;303:117676.

13