ArticlePDF Available

Abstract and Figures

Methane leakage caused by well integrity failure was assessed at 28 abandoned gas wells and 1 oil well in the Netherlands, which have been plugged, cut and buried to below the ground surface (≥3 mbgl). At each location, methane concentrations were thoroughly scanned at the surface. A static chamber setup was used to measure methane flow rates from the surface as well as from 1 m deep holes drilled using a hand auger. An anomalously high flow rate from 1 m depth combined with isotopic confirmation of a thermogenic origin revealed ongoing leakage at 1 of the 29 wells (3.4%), that had gone undetected by surficial measurements. Gas fluxes at the other sites were due to shallow production of biogenic methane. Detailed investigation at the leaking well (MON-02), consisting of 28 flux measurements conducted in a 2 × 2 m grid from holes drilled to 1 and 2 m depth, showed that flux magnitude was spatially heterogeneous and consistently larger at 2 m depth compared to 1 m. Isotopic evidence revealed oxidation accounted for roughly 25% of the decrease in flux towards the surface. The estimated total flux from the well (443 g CH4 hr−1) was calculated by extrapolation of the individual flow rate measurements at 2 m depth and should be considered an indicative value as the validity of the estimate using our approach requires confirmation by modelling and/or experimental studies. Together, our findings show that total methane emissions from leaking gas wells in the Netherlands are likely negligible compared to other sources of anthropogenic methane emissions (e.g. <1% of emissions from the Dutch energy sector). Furthermore, subsurface measurements greatly improve the likelihood of detecting leakage at buried abandoned wells and are therefore essential to accurately assess their greenhouse gas emissions and explosion hazards.
Content may be subject to copyright.
Contents lists available at ScienceDirect
Journal of Contaminant Hydrology
journal homepage: www.elsevier.com/locate/jconhyd
Impact of groundwater ow on methane gas migration and retention in
unconsolidated aquifers
Gilian Schout
a,b,c,
, Niels Hartog
a,c
, S. Majid Hassanizadeh
a
, Rainer Helmig
d
, Jasper Grioen
b,e
a
Earth Sciences Department, Utrecht University, 3584 CB Utrecht, the Netherlands
b
Copernicus Institute of Sustainable Development, Utrecht University, 3584 CB Utrecht, the Netherlands
c
KWR Water Cycle Research Institute, 3433 PE Nieuwegein, the Netherlands
d
Institute for Modelling Hydraulic and Environmental Systems, Universität Stuttgart, Pfaenwaldring 61, 70569 Stuttgart, Germany
e
TNO Geological Survey of the Netherlands, 3584 CB Utrecht, the Netherlands
ABSTRACT
Methane leaking at depth from hydrocarbon wells poses an environmental and safety hazard. However, determining the occurrence and magnitude of gas migration
at ground surface is challenging, as part of the leaking gas is retained during upward migration. We investigated migration through unconsolidated sedimentary
aquifers using a two-phase, two-component (water and methane) ow and transport model constructed in DuMu
x
. A sensitivity analysis for migration through a 60 m
thick sandy aquifer showed that retention by dissolution can be signicant even with low groundwater Darcy velocities of 1 m.yr
1
. Retention was negligible in the
absence of groundwater ow. Besides groundwater velocity, both hydrogeological (permeability, entry pressure, pore-size distribution, and residual gas saturation)
and leakage conditions (depth, magnitude and spatial dimensions) determined model outcomes. Additional simulations with interbedded ner grained sediments
resulted in substantial lateral spreading of migrating gas. This delayed upward migration and enhanced retention in overlying sandy units where groundwater
velocities are highest. Overall, the results of this study show that for unconsolidated aquifer systems and the most commonly observed leakage rates (0.110 m
3
.d
1
),
signicant amounts of migrating methane can be retained due to dissolution into laterally owing groundwater. Consequently, resulting atmospheric methane
emissions above such leaks may be delayed with decades after the onset of leakage, signicantly reduced, or prevented entirely.
1. Introduction
The reliance on fossil fuel resources for the majority of the world's
energy supply has resulted in a large number of onshore hydrocarbon
wells, conservatively estimated at over 4 million (Davies et al., 2014).
In spite of eorts to maintain the vertically isolating function of geo-
logical formations that are penetrated when installing and operating
such wells, failure of the wellbore system is a commonly observed
problem (Davies et al., 2014) and can lead to leakage of hazardous
liquids and gases. Furthermore, research has shown that this risk con-
tinues or may develop even after the active life-time of wells and their
abandonment (Kang et al., 2014;Townsend-Small et al., 2016). Parti-
cularly, upward leakage of methane through anthropogenically opened,
unintended connections between hydrocarbon reservoirs and the
shallow subsurface has become a growing concern worldwide as it may
contribute to greenhouse gas emissions (Kang et al., 2014), deteriorate
water quality (Vengosh et al., 2014), and form an explosion hazard
(Chilingar and Endres, 2005). On top of that, leaky wells could serve as
pathways for the migration of other uids when the downhole condi-
tions are actively changed, for example when hydraulic fracturing is
carried out (Brownlow et al., 2016)orCO
2
is stored (Gasda et al., 2004)
in nearby wells. Thus, their presence may also hamper the safe im-
plementation of future uses of the subsurface.
To be able to quantify and mitigate these risks, the accurate de-
tection of gas leakage is vital. This is typically achieved by measure-
ments of either sustained casing pressure (SCP) or surface casing vent
ow (SCVF) at the wellhead (King and King, 2013). However, these
measurements may only reect a part of the total leakage ux, as a
signicant fraction of leaking gas can escape the wellbore system en-
tirely and enter the surrounding geology (Forde et al., 2019), which is
also referred to as gas migration. Lackey and Rajaram (2018) identied
three main mechanisms that may lead to gas migration: (1) gas cir-
cumvention, when gas migrates through a cemented outer annulus and
a section of low quality cement is overlain by a section of higher quality
cement, causing the gas to migrate outwards, (2) groundwater cross-
ow, when gas leaks through an uncemented outer annulus and is
transported into the surrounding formation by lateral groundwater
ow, and (3) SCP-induced gas migration, when the gas pressure at the
bottom of the surface casing exceeds the hydrostatic pressure leading to
gas overow.
As gas migrating outside the wellbore may become trapped, dis-
solved or degraded (Cahill et al., 2018), surcial or shallow
https://doi.org/10.1016/j.jconhyd.2020.103619
Received 6 September 2019; Received in revised form 17 January 2020; Accepted 23 January 2020
Corresponding author at: Earth Sciences Department, Utrecht University, 3584 CB Utrecht, the Netherlands.
E-mail address: g.schout@uu.nl (G. Schout).
Journal of Contaminant Hydrology 230 (2020) 103619
Available online 24 January 2020
0169-7722/ © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
T
groundwater wells may only detect leakage after long periods of time,
or possibly never. Measurements of soil gas migration are typically
carried out using surface ux chambers (Erno and Schmitz, 1996). In
the vadose zone, oxidation and dispersion of methane has indeed been
shown to be capable of masking leaking wells from being detected at
the surface entirely (McMahon et al., 2018;Schout et al., 2019). Fur-
thermore, a recent eld experiment where methane was released in a
shallow aquifer showed that the combined eects of trapping and dis-
solution of leaking gas in the saturated zone signicantly reduced the
amount of gas that reached the surface (Cahill et al., 2018). Lastly,
several eld-based studies have shown that anaerobic methane oxida-
tion coupled to either sulphate reduction (Van Stempvoort et al., 2005;
Wolfe and Wilkin, 2017) and/or reduction of iron and manganese
oxides (Schout et al., 2018;Woda et al., 2018) can lead to attenuation
of a migrating methane plume. With increasing depth and longer mi-
grating pathways leaking gas can become more severely aected by
these attenuation and retention processes, and in turn surcial or
shallow subsurface measurements become increasingly less reliable
tools for detecting gas migration. Indeed, the fate of methane from leaks
occurring at depths greater than 10 m was identied as a key knowl-
edge gap in studying gas migration (Cahill et al., 2019).
Given the complexity and related costs of measuring gas leakage in
eld or experimental settings, methane migration has also been as-
sessed by means of multiphase numerical ow and transport simula-
tions. Several studies have aimed to determine ranges of possible gas
migration ow rates from the reservoir depth to shallow aquifers
through high permeability pathways, such as faults or improperly ce-
mented annuli (Kissinger et al., 2013;Nowamooz et al., 2015;Reagan
et al., 2015;Schwartz, 2015). Tatomir et al. (2018) simulated migration
of methane at great depth (~1500 m) into an inclined, regional aquifer.
Breakthrough times of gaseous methane at various oset distances from
the leak origin were calculated, that could for example correspond to
conductive faults connecting the deep aquifer with an overlying
freshwater aquifer. Rice et al. (2018) investigated SCP-induced gas
migration through a shale formation present at the bottom end of the
surface casing into an overlying shallow aquifer. They showed that for
given source zone pressures, the permeability distribution and the
parametrization of the capillary pressure saturation and relative per-
meability functions of the shale formation controlled the ow rate of
methane into the overlying aquifer. Depending on these properties,
ow rates at the base of the aquifer can be slow, which may allow
methane contamination to go undetected.
Other studies assumed leakage to a shallow aquifer system occurred,
and focused on the migration and attenuation of methane therein. Roy
et al. (2016) coupled their multiphase model to a reactive transport
simulator and showed that in conned aquifers, anaerobic methane
oxidation coupled to sulphate reduction could attenuate a migrating
dissolved methane plume signicantly. This attenuating eect was even
larger in unconned aquifers, where aerobic methane oxidation was
also possible. Moortgat et al. (2018) simulated gas phase methane mi-
gration through an aquifer system characterized by the presence of
highly permeable pathways such as fractures and uvial channels. They
showed that rapid lateral gas migration over distances of several kilo-
meters is possible through these features, but that migration is much
less rapid in unfractured media. D'Aniello et al. (2019) simulated
leakage of methane from a geothermal well into a 2 m thick surcial
unconsolidated aquifer. Given the limited thickness and because mass
transfer of methane to the aqueous phase was ignored, retention of
methane in this aquifer was limited. Klazinga et al. (2019) carried out
2D simulations imitating the eld experiments by Cahill et al. (2017)
where methane gas was injected up to 10 m depth in a sandy aquifer.
Lateral migration of the gas phase was shown to be signicant even
over such a relatively shallow interval, as a result of anisotropic sedi-
ments, and the presence of low permeability layers. Furthermore, wider
plumes and larger groundwater ow velocities resulted in larger
amounts of methane that were retained in the aquifer.
The eect of methane retention by dissolution into laterally owing
groundwater has not been considered in detail in previous studies of gas
migration. In spite of the relatively low solubility of methane, dis-
solutive retention could be important in the shallow part of un-
consolidated groundwater systems, where groundwater velocities are
generally higher. Methane migration through unconsolidated aquifers
is also less likely to be dominated by quick gas phase ow through
preferential ow paths, which reduces the potential for dissolutive re-
tention. Other factors also possibly play a role, such as the increase in
methane aqueous solubility with increasing hydrostatic pressure (i.e.
depth) and the lateral spreading of the methane plume as a result of
anisotropy and low permeable layers. The interaction of these transport
processes and their inuences on gas migration were studied in a
parameter sensitivity analysis based on 3D, two-phase, two-component
numerical simulations across a range of realistic conditions. The im-
pacts of horizontally layered unconsolidated aquifer systems on me-
thane migration and retention were illustrated by simulation of a
number of additional scenarios, based on the geology encountered at
two recently identied leaking wellbore sites in the Netherlands. The
overall aim of this study was to determine whether, and if so, to what
extent subsurface methane migration through laterally owing
groundwater is impacted by methane dissolution.
2. Material and methods
2.1. Governing equations and constitutive relations
Numerical modelling was carried out using the open source multi-
physics simulation package DuMu
x
(Ackermann et al., 2017;Flemisch
et al., 2011). DuMu
x
is capable of calculating multiphase multi-
component ow and transport at the continuum scale (also referred to
as miscible two-phase ow or compositional ow). For this study, two
phases α(liquid and gas) and two components k(H
2
O and CH
4
) were
considered. The model accounts for mass transfer between the two
phases as well as the compressibility of both phases. Phase velocities are
calculated using Darcy's law. Furthermore, binary diusion is assumed,
and the diusive uxes in each phase are calculated according to Fick's
law. The fully coupled mass balance equation is then formulated as
follows:
∑∑
=∇
∇−
+
∙∇+
κg
ϕρXS
tρX k
μ
ρD X q
() ()
{}
α
αα
kα
α
αα
k
α
αα
α
αeff α α
kk
,(1)
where Φis the porosity, ρis the density, Xis the mass fraction, Sis the
saturation, κis the intrinsic permeability tensor, k
r
is the relative per-
meability, μis the dynamic viscosity, Pis the pressure, gis the gravity
vector, D
e
is the eective diusion coecient and qis a source or sink
term. A number of constitutive relationships are needed to close this
system of equations. Firstly, the sum of the saturations and that of the
mass fractions of the two components in each phase must equal 1:
∑∑
+= = =SS X X1, 1,
1
gl kg
k
klk(2)
where the subscripts g and w represent the gaseous and liquid phase,
respectively. The phase pressures are related by the capillary pressure
P
c
as given by:
−=
P
PPS()
.
glcg (3)
The formulations by Brooks and Corey (1964) were used for the
capillary pressure - saturation relationship and relative permeabilities:
=
P
SPS()
cg e
eλ
1
(4)
⎜⎟
==
++
kS k S S,(1)1
rl eλrg e eλ
2/ 3 2
21
(5)
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
2
where P
e
is the capillary entry pressure, S
e
is the eective saturation and
λis the pore size distribution index. S
e
is dened as:
=
−−
SSS
SS1
ellr
gr lr (6)
where S
lr
and S
gr
are the residual (or irreducible) liquid and gaseous
saturation, respectively. According to these formulations, the relative
permeability of the gas phase is 0 as long as S
g
<S
gr
. Hence, the re-
sidual saturations in each cell have to reach at least up to the
(threshold) residual gas saturation before a neighboring cell can be
invaded by the gas phase (Fig. S1). A regularization of the Brooks-Corey
P
c
(S
g
)relationship was used for S
e
> 1 and S
e
< 0.01. To avoid nu-
merical issues with having an innite gradient of the P
c
(S
g
)curve, the
slope of the curve when approaching these limits was extended up to
S
l
= 1 and S
l
= 0, respectively (Fig. S1).
Both water and methane components can be present in either phase.
Concentrations in each phase are assumed to be at thermodynamic
equilibrium, meaning that mass transfer between the two phases occurs
instantaneously at every time step in each control volume. Methane
solubility is determined according to Henry's law and is temperature
and pressure dependent, with the Henry's coecient based on the
IAPWS formulations (Fernández-Prini et al., 2003). A freshwater
aquifer is assumed and therefore the eects of variable salinities on
methane solubility are not considered. The solubility of methane also
determines the phase state of the model, with a gas phase appearing
once the mole fraction of methane (x
lCH4
)exceeds that of the equili-
brium (solubility) mole fraction, and vice versa. Phase appearance and
disappearance are accompanied by a primary variable switch: from P
l
and S
g
when both phases are present to P
l
and x
lCH4
when only the
liquid phase is present. Energy transport is not simulated, and the
model is considered to be in thermal equilibrium. For the spatial dis-
cretization a vertex centered nite volume method (box method)is
used and a fully implicit, backward Euler method for the temporal
discretization (Helmig, 1997).
The density of the gaseous phase follows the ideal gas law. Viscosity
is calculated as described in Reid et al. (1987). In the range of tem-
perature and pressure relevant for this study, the density and viscosity
of methane gas calculated in this manner were shown to be nearly equal
to more complex equations of state (Kissinger et al., 2013). The density
and viscosity of the liquid phase are both pressure and temperature
dependent, according to the IAPWS denitions (Wagner and Pruss,
2002). The inuence of the phase composition on the liquid phase
properties is not taken into account, since even at the greatest depth
considered in this study (480 m of water column) the maximum CH
4
mass concentration is negligible compared to the aqueous phase density
(~0.1%). The diusion coecient of water in the gaseous phase is
determined according to the method in Fuller et al. (1966) and the
diusion coecient of methane in the aqueous phase according to Reid
et al. (1987). Mechanical dispersion was not implemented in the model.
In multiphase systems, the dispersion coecient is saturation depen-
dent, which would result in a much more complex system of equations
(Helmig, 1997) and excessive runtimes. Computations were performed
on a workstation and parallelized over 16 processors, resulting in
runtimes of up to 2 days for the sensitivity analysis and 4 days for the
layered scenarios.
2.2. Geological context and conceptual model
The range of hydrogeological conditions for which gas migration
was studied are representative for unconsolidated sedimentary
groundwater systems globally, but were inspired by the conditions that
prevail in the subsurface of the Netherlands. The Netherlands is one of
the major oil and gas producing nations in Europe with around 2500
onshore oil and gas wells (Schout et al., 2019). A survey of 986 gas
wells by the State Supervision of Mines (SodM) revealed some form of
well barrier failure at 227 (23%) of these wells (SodM, 2019). Ob-
servations of gas bubbles in ooded well cellars showed that leakage of
thermogenic gas occurred at at least 13 wells (1.3%). Hydro-
geologically, the country is characterized by the presence of shallow
sandy aquifer units of Plio-Pleistocene age. These aquifers can be either
phreatic or, in the northern and western part of the country, can be
conned by overlying Holocene deposits (Fig. S2). A succession of
thick, marine clays of Neogene and Paleogene age form the bedrock
below the aquifer units in nearly the entire country (de Vries, 2007).
Typically, the surface casing of oil and gas wells would at least extend
down into these clays, the depths of which can be up to around 500 m
below surface in the oil and gas producing areas (Kombrink et al.,
2012). The modelled domain is a hypothetical rectangular block of a
sandy aquifer within such a unconsolidated sedimentary sequence
(Fig. 1a). Gas migration into the base of the model is assumed. Con-
ceptually, it could be caused by a number of possible failure scenarios
occurring below the simulated part of the aquifer, including SCP-in-
duced gas migration or gas circumvention.
2.3. Domain discretization, boundary and initial conditions
The dimensions of the simulated aquifer were set to 100 × 60 × 60
m (XYZ), suciently large so that no boundary eects on gas migration
occurred. Owing to symmetry in the y-axis, only half of this domain is
simulated. Initial numerical testing showed that model outcomes sta-
bilized when using a grid size smaller than 1x1x1 m. Hence, a grid size
of 0.5 × 0.5 × 0.5 m was used, resulting in a total of 1,440,000 cells.
Fig. 1. Conceptual model example (a) and numerical implementation of the modelled section (b).
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
3
The simulation time is 20 years, during which injection of CH
4
is
continuous. For the CH
4
ux at the inlet (mol.m
2
.s
1
) a Neumann
type boundary was used over 8 cells from x = 40 to 42 m (2 × 1 m
2
), or
what would be 2 × 2 m
2
area when accounting for the half of the
domain that is not modelled (Fig. 1b). The remainder of the bottom face
of the model and the two lateral faces parallel to the direction of
groundwater ow are treated as no-ow boundaries for both compo-
nents. The two lateral faces perpendicular to groundwater ow are
assigned Dirichlet conditions with a pressure equal to the hydrostatic
pressure, albeit with a minor increase in pressure on one side to induce
groundwater ow. Lastly, the top face of the model is assigned a no-
ow condition for H
2
O and an outow condition for CH
4
. This allows
methane to escape the model freely through the top boundary but keeps
water owing strictly horizontally. Conceptually, this corresponds to
cases where the gas either escapes to the atmosphere or continues to
migrate upwards to overlying layers, depending on the depth of the top
of the domain. The model is initially fully water saturated (S
g
= 0) and
no dissolved methane is present anywhere. A thermal gradient of
31.3 °C.km
1
is imposed in all simulations, equivalent to the average
thermal gradient found in the Netherlands (Verweij et al., 2018), on top
of the yearly average ambient air temperature of 10 °C. The density and
viscosity of water inside the model and at its lateral boundaries are
determined during an initialization phase that precedes the actual
model run.
2.4. Parameter space sensitivity analysis
A total of 17 simulations were run, together encompassing a para-
meter space representative of the expected conditions for gas migration
through unconsolidated sandy aquifers (Table 1). Groundwater ow
velocities up to 100 m.yr
1
were considered, as observed for regional
groundwater ow in the Netherlands (Bloemendal and Hartog, 2018).
The thickness of the model is 60 m. For the base case a depth of 60 m
was chosen for the bottom of the model domain, with depths of 240 and
480 m also considered. Preliminary simulations showed that model
outcomes are insensitive to the temperature gradient however, as si-
mulations ran with a constant temperature of 10 °C yielded virtually
equal outcomes. Hence, variations in the geothermal gradient were not
considered.
Methane ow rates at the inlet were varied between 0.1 and 10 m
3
CH
4atm
.d
1
(i.e. the volumetric CH
4
ow rate at atmospheric pressure
and 10 °C) as the majority of reported SCVF rates fall within this range.
For example, 68% of wells with SCVF in British Columbia, Canada, had
ow rates below 1 m
3
.d
1
and 25% between 1 and 10 m
3
.d (Wisen
et al., 2019). Similarly, the vast majority of reported SCVF rates in
Alberta, Canada, are also below 10 m
3
.d
1
(Dusseault et al., 2014). It
should be noted that very large SCVF rates exceeding 1000 m
3
.d
1
have also been reported (e.g. Nowamooz et al., 2015), but were not
considered in this study. Imposed leakage rates were sustained for the
full 20 year simulation period. The assumption of a constant leakage
rate over long time periods was also made by Rice et al., 2018. Field
measurements of leakage from abandoned wells in Pennsylvania that
remained virtually constant over a 3 year time period support this as-
sumption (Kang et al., 2016), as do measured SCPs that sustained over
measurements periods up to nearly 10 years (Lackey and Rajaram,
2018).
While properties of sandy aquifers are well known for single-phase
problems, experimental studies where multiphase properties are de-
termined are sparse. To reduce the number of possible scenarios, three
sets of experimentally determined values for an air-water system were
used for the porosity, permeability, entry pressure, and pore size dis-
tribution (Clayton, 1999). As a base case assumption, the properties of a
coarse grained uvial sand (D50 of 0.61 mm) were taken (Table 1). The
eect of migration through ner grained sands was considered by im-
plementing the values determined for a hydraulic ll(D50 of
0.16 mm) and a clayey alluvium(D50 of 0.09 mm). Hereafter, we refer
to these as a ne sand and very ne sand, respectively, according to the
Wentworth scale of grain size classications (Wentworth, 1922). Given
that the transverse permeability is not known, an anisotropy factor (k
h
/
k
v
) of 5 was assumed in all simulations. Variations in anisotropy with a
k
h
/k
v
of 1 and 10 were simulated in scenarios 14 and 15, respectively
(Table 1). A residual gas saturation of 1% was assumed. The sensitivity
of model outcomes to the residual gas saturation was investigated in
scenarios 12 and 13, with values of 0% and 15%, respectively. The
residual wetting phase saturation was set to 10% for all simulations. In
scenarios 16 and 17 the inlet boundary size was changed (to 1 × 1 m
2
and4×4m
2
), while maintaining the total gas ow rate over the
boundary.
2.5. Setup of two layered case studies based on real sites
For the sensitivity analysis, a simplied, homogeneous, sandy
aquifer was considered. However, in unconsolidated sedimentary ba-
sins these aquifers are typically alternated by layers of both coarser and
ner grained sediments, ranging from gravel to clay. To investigate the
eect of such stratication on gas migration, additional simulations
Table 1
Summary of relevant input parameters used for the 17 simulated scenarios in the sensitivity analysis.
Scenario [#]
a
Q
gin
[m
3
.d
1
]q
w
[m.yr
1
] Depth [m] Porosity []k
h
[m
2
]k
v
[m
2
]P
e
[Pa] λ[m
1
]S
gr
[] Note
1 0.1 1 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.01 base case
2 0.1 060 0.38 5.3E-11 1.1E-11 1766 1.50 0.01
3
b
0.1 060 0.38 5.3E-11 1.1E-11 1766 1.50 0.01 x
iCH4
/x
maxCH4
= 95%
411 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.01
5 0.1 1 60 0.45 7.8E-12 1.6E-12 3924 1.20 0.01 Fine sand
6 0.1 1 60 0.33 2.7E-11 5.4E-12 2256 0.15 0.01 Very ne sand
711060 0.38 5.3E-11 1.1E-11 1766 1.50 0.01
81 100 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.01
910 100 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.01
10 11240 0.38 5.3E-11 1.1E-11 1766 1.50 0.01
11 11480 0.38 5.3E-11 1.1E-11 1766 1.50 0.01
12 0.1 1 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.00
13 0.1 1 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.15
14 0.1 1 60 0.38 5.3E-11 5.3E-11 1766 1.50 0.01 k
h
/k
v
=1
15 0.1 1 60 0.38 5.3E-11 5.3E-12 1766 1.50 0.01 k
h
/k
v
=10
16 0.1 1 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.01 Area inlet: 1 × 1 m
2
17 0.1 1 60 0.38 5.3E-11 1.1E-11 1766 1.50 0.01 Area inlet: 4 × 4 m
2
Bold numbers indicate parameters that are varied with respect to the base case scenario.
a
Volumetric CH
4
ow rate over inlet at atmospheric pressure and 10 °C due to domain symmetry only half of this ow rate is applied in the model.
b
Initial methane concentration at 95% of the depth-dependent solubility mole fraction (x
iCH4
/x
maxCH4
= 95%).
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
4
were carried out based on the hydrogeology observed at two locations
where methane leakage was recently shown to occur. At the rst lo-
cation, near the village of Sleen in the east of the Netherlands, gas
migration resulted from a blowout that occurred in 1965 (Schout et al.,
2018). At the second location, in a village called Monster in the west of
the Netherlands, gas migration was detected above a fully decommis-
sioned cut-and-buried gas well (Schout et al., 2019). This leak was more
recently closed oby the responsible operator, in accordance with
Dutch law. Hydrogeologically, these two locations are quite dierent as
they are situated at opposite ends of the Netherlands (Fig. S2). Hence,
they serve as two distinct case studies used to investigate the eects of
horizontal layering on gas migration. However, it is important to note
here that the goal is not to reproduce exactly the leakage conditions at
these sites, as required information about the depth and magnitude of
the leaks is not known.
Lithological and permeability data were retrieved from the publicly
available national hydrogeological model REGIS II (Vernes and Doorn,
2005). REGIS II divides the subsurface into sandy, complex, and clayey
layers. Complex layers typically consist of successions of more and less
permeable sediments that are taken together as one regional layer. As a
result, they often have a large anisotropy with a much lower vertical
than horizontal permeability. Where not given, anisotropy was assumed
to be 5 for sandy layers and 10 for clay layers. Porosities were assumed
to be 30% for sandy layers, 35% for complex layers, and 40% for clay
layers. The Brooks-Corey pore size distribution index was assumed to be
1.5 for sandy layers, following the experimentally determined value for
the sand used in the base case scenario. Accordingly, a smaller pore size
distribution index was chosen for the clayey (0.75) and for complex
(0.5) layers, reecting the larger variation in pore sizes that may be
expected for such lithologies. The Leverett J-function (Eq. (7)) was used
to scale the entry pressure based on the ratio between porosity and
permeability. Separate reference values were used for the complex,
sandy and clayey units (Table S1).
=
SpSkϕ() () /
wcw(7)
The bedrock at the Sleen site can be considered the top of the Breda
Formation at 126 m depth. The case study modelled after this site was
therefore constructed from this interface up to the surface. At the
Monster location, the thick clay unit below the base of the Maassluis
formation at 231 m depth was considered as the bedrock. Furthermore,
the surcial Holocene deposits (55 m thick) are not included in the
model as REGIS II does not provide data on them. The modelled section
is therefore 176 m thick. The resulting hydrogeological input data used
for both case studies is shown in Fig. 2. For each case two simulations
were carried out, one with a groundwater head gradient of 25 cm.km
1
and one with 100 cm.km
1
, on top of the hydrostatic pressure. A ow
rate of 10 m
3
CH
4atm
.d
was assumed for all simulations. Otherwise,
the boundary and initial conditions were kept equal to those used in the
sensitivity analysis. The width of the model was extended to 80 m to
avoid gas phase pools below interfaces of entry pressure or permeability
reaching the boundaries of the model. Given the extended size of these
models and the more complex hydrogeology, grid renement was
slightly reduced to 1 × 1 × 1 m to improve run times.
3. Results
First, an assessment of the potential for methane retention in the
subsurface is briey presented, and the relative magnitude of forces
relevant to methane migration for the unconsolidated aquifers under
study. Then, the results of the sensitivity analysis are presented, fol-
lowed by the results of hydrogeologically layered case studies.
3.1. Dimensional analysis of methane retention and migration
Within the range of temperature and pressure considered in this
study, that result from a maximum depth of 480 m below water table,
methane aqueous solubility and gas phase density increase from 31 to
1112 mg.L
1
and 0.69 to 32.3 kg.m
3
, respectively (Fig. S3). While
methane mass density as a gas phase is much greater than its solubility
at equal depths, the mass stored in the aqueous phase is roughly equal
at gas phase saturations of 4% (Fig. 3). For gas saturations exceeding
4%, retention in the gas phase starts to rapidly exceed aqueous reten-
tion. At a gas saturation of 4% and a depth of 60 m below the water
table, the cumulative storage in both phases is ~0.2 kg.m
3
(Fig. 3). In
comparison, the ow rate in our base case scenario is 0.1 m
3
CH
4atm
.d
1
, equal to a methane mass ow rate of 0.07 kg.d
1
or ux of
~0.02 kg.m
2
.d
1
(given the inlet area of 4 m
2
). Therefore, at this
depth, it would require 10 days to saturate 1 m
3
of aquifer with me-
thane, assuming a gas saturation of 4%. Although this is just a rst
approximation, which notably does not take into account the replen-
ishment of available groundwater for methane to dissolve in due to
groundwater ow, it conrms that there is indeed potential for sub-
surface methane retention to signicantly aect the monitorability of
gas migration originating at depth.
Following the denitions in Kopp (2009) a dimensional analysis of
the balance of forces in the system was carried out. Preliminary testing
showed that for the gas migration velocities encountered in our sensi-
tivity analysis, viscous forces were insignicant compared to both
gravity and capillary forces. Hence, the system is characterized by the
dimensionless Bond number:
==
Bo capillary forces
gravitational forces
p
ρρgl()
cr
wg
cr (8)
where p
cr
and l
cr
are the critical pressure and length, respectively, and g
is the gravity constant (9.81 m.s
2
). The critical pressure is dened as
the capillary pressure drop over the saturation front length and is
therefore roughly equal to entry pressure. Typically, for advection-
driven ow systems, the critical length is taken to be equal to the length
of the saturation front width (Kopp, 2009). When assuming a critical
length equal to the discretization length (0.5 m), capillary forces equal
gravitational forces for entry pressure values of 5 kPa (Fig. S4). This
shows that for the three sands considered in the sensitivity analysis,
which have a maximum entry pressure of 3.9 kPa (Table 1), gravita-
tional forces likely dominate. However, the complex and clayey layers
in the hydrogeologically layered case studies have a maximum entry
pressure of 7.8 kPa (Fig. 2). Depending on the actual width of the sa-
turation front, capillary forces likely exceed gravitational forces for
these clayey sediments.
3.2. Sensitivity analysis
3.2.1. Retention by CH
4
dissolution
In the base case scenario (Table 1) the upward, buoyancy-driven
migration of the gas plume from the base of the simulated aquifer to its
top (60 m interval) takes 1.65 years (Fig. 4a and Table 2). Due to the
coarse grain size and relatively low methane ux, migration is mostly
vertical and the gas phase stays within the column overlying the inlet
boundary. This observation is in line with the ndings of the dimen-
sional analysis, and conrms that gravitational forces indeed dominate
in this case. Maximum gas phase saturations remain low at only 2.4%.
As the gas phase migrates, the surrounding water column is saturated
throughout, and methane saturated groundwater is advectively trans-
ported away from the gas phase plume, allowing more gaseous methane
to be dissolved (Fig. 5). Migration time reduces to 1.14 years when the
Darcy velocity in the aquifer is reduced to zero (scenario 2) and just
0.35 years if in addition the initial concentration of methane
throughout the aquifer is raised from 0 to 95% of the solubility (sce-
nario 3). Therefore, methane dissolution, from the bottom to the top of
a 60 m thick sandy aquifer with a groundwater Darcy velocity of only
1 m.yr
1
, causes the migration of a 0.1 m
3
CH
4atm
.d
1
leak to occur
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
5
1.3 years more slowly than it would have without dissolution.
The dierence in results of the rst three scenarios is more pro-
nounced when looking at the percentage of inowing methane that
exits the model through the top boundary (Q
out
/Q
in
). For scenario 3,
Q
out
/Q
in
reaches 82% in a year and goes to 100% after approximately
5 years (Fig. 4a). Ultimately, this fraction even slightly exceeds 100% as
some of the initially dissolved methane also passes through the top
boundary (Fig. 4a). For scenario 2, when both the initial methane
concentration and groundwater velocity are zero, Q
out
/Q
in
increases
gradually with time to 91% after 20 years. The remaining 9% is trapped
in the aquifer by dissolution and subsequent diusion away from the
gas phase plume. With time, this fraction would steadily increase fur-
ther as the aquifer saturates with methane and the rate of diusion
slows down. However, for the base case scenario these outcomes are
entirely dierent, as Q
out
/Q
in
stabilizes after 4 years when just 10% of
the imposed methane ow rate migrates on through the top boundary.
This shows that dissolution in combination with advective transport
exerts a much stronger control on gas migration than dissolution with
diusive transport, even for groundwater with a Darcy velocity of only
1 m.yr
1
.
3.2.2. Impact of methane ow rate
The relative impact of dissolutive retention is signicantly reduced
when increasing the methane ow rate through the inlet by a factor of
10, from 0.1 to 1 m
3
CH
4atm
.d
1
(scenario 4). While maximum gas
phase saturations increased to 4.6%, migration is still primarily vertical
and migration time reduced from 1.65 year in the base case scenario to
just 0.16 year (Table 2). Therefore, a smaller fraction of migration
methane can be dissolved and retained in the aquifer, given that the
groundwater velocity was kept equal, and Q
out
/Q
in
stabilizes after just
2 years at 85% compared to 4 years and 10% in the base case scenario
(Fig. 4a).
3.2.3. Impact of sand properties
Due to the smaller permeability, larger entry pressure and smaller
pore size distributions of the ner grained sands simulated in scenarios
5 and 6 (Table 1), gas phase propagation patterns are changed con-
siderably with respect to the base case (Fig. 5). Particularly for the case
of a very ne sand (scenario 6), with a 10 times smaller pore size
Fig. 2. Hydrogeological input data used for construction of the two layered case studies. In the permeability plot, the solid line (k
v
) and dashed line (k
h
) show the
vertical and horizontal permeability, respectively. In the Darcy velocity plot, the solid line (q
25
) and dashed line (q
100
) show the modelled Darcy velocities resulting
from the cases with groundwater head gradients of 25 and 100 cm.km
1
, respectively.
Fig. 3. Potential mass density of methane as a gas phase (green lines) or in the
aqueous phase (blue lines) as a function of depth, for three dierent values of
the gas phase saturation (S
g
). Temperature is based on a thermal gradient of
31.3 °C km
1
and a porosity of 38% was assumed, equal to that of the base case
scenario. (For interpretation of the references to colour in this gure legend, the
reader is referred to the web version of this article.)
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
6
distribution index (0.15 versus 1.5 in the base case), the gaseous plume
becomes more rounded (Fig. 5a). This also leads to some minor up-
stream migration of the gas phase, and is the result of an increase in
capillary force, which acts in all directions equally, relative to the
gravitational force, which acts only vertically upwards. In the ne sand
case (scenario 5), both vertical and horizontal permeabilities are ~7
times lower than in the base case. Due to this increased resistance to
ow, the maximum gas phase saturation becomes higher than in the
base case: 3.0% versus 2.4%. The gas phase can be seen to migrate
slightly along with groundwater ow in both scenarios with ner
average grain sizes (Fig. 5b). Ultimately, the change in shape and de-
creased speed of upward propagation allows more methane to be dis-
solved in both cases, given that Darcy velocities were kept equal. This
causes the gas phase propagation to stagnate at 6.5 m and 18.5 m below
the top of the model domain for the ne sand and the very ne sand
cases, respectively (Table 2). Hence, Q
out
/Q
in
remains 0. The distance
with which methane has been transported at the base of the model is
28 m in the base case scenario, 24 m in scenario 5 and 34 m in scenario
6, calculated from the inlet at x = 42 m to where the concentration
equals half that of the solubility at 60 m depth (204 mg.L
1
). The
dierence between these distances is caused by the dierences in por-
osity, resulting in varying eective groundwater velocities. It should be
noted that due to diusion, the distance where methane concentrations
are suciently large to be readily detected in groundwater samples
(~0.1 mg.L
1
) is almost twice as large (Fig. 5b).
3.2.4. Impact of groundwater ow velocity
Gas migration resulting from a 1 m
3
CH
4atm
.d
1
ow rate is not
impacted greatly when raising groundwater velocity from 1 to
10 m.yr
1
(scenarios 4 and 7). Migration time increases from 0.16 to
0.25 year and Q
out
/Q
in
is reduced from 85% to 76% (Table 2). How-
ever, when groundwater velocity is further increased to 100 m.yr
1
(scenario 8), the gas plume movement is severely impacted and mi-
gration stagnates at 53.5 m below the top of the aquifer, having moved
up just 6.5 m (Fig. 4b). The impact of a 100 m.yr
1
groundwater ow
velocity was also assessed for an even higher methane ow rate of
10 m
3
CH
4atm
.d
1
(scenario 10). The higher ow rate results in larger
maximum gas saturations (6.2%) and much more rapid gas phase
Fig. 4. Vertical propagation of the gas phase plume (left) and fraction of inowing methane that exits (Q
out
/Q
in
) the domain through the top boundary (right).
Results for scenarios 14 are shown in gure a, scenarios 4, 7, 8 and 9 in gure b, and scenarios 4, 10 and 11 in gure c. Lines are labelled according to the
corresponding scenario numbers in Table 1, with simulated variations with respect to the base case indicated in gure legends.
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
7
migration, as the top of the aquifer was reached in 0.06 years. However,
Q
out
/Q
in
stabilized at just 2%. This shows that even large gaseous ow
rates can be severely impacted by dissolutive retention, in spite of a
very rapid upward migration of the gas phase.
3.2.5. Impact of aquifer depth
The depth of the aquifer was increased 4 and 8 times in scenarios 11
and 12, respectively, resulting in an increased depth of the base of the
aquifer of 240 m and 480 m depth (aquifer thickness remains 60 m).
Simulations were run with a methane ow rate of 1 m
3
CH
4atm
.d
1
, and
are thus compared to that of scenario 4 (Table 1). The average methane
solubility over the resulting depth intervals is 4.8 and 8.6 times higher
than in the base case scenario, and the average gas phase density in-
creases 5.3 and 10.9 times (Fig. S3). Both aqueous and gaseous
Table 2
Percentage of inowing methane that exits through the top boundary of the model (Q
out
/Q
in
) with increasing time, for the 17 scenarios (Table 1) considered in the
sensitivity analysis.
Scenario [#] Time to top aquifer [yrs] max depth gas plume [m] Q
out
/Q
in
[%]
0.25 [yrs] 0.5 [yrs] 1 [yrs] 2 [yrs] 3 [yrs] 5 [yrs] 7.5 [yrs] 10 [yrs] 15 [yrs] 20 [yrs]
1 1.65 0.0 0.0 0.0 4.6 8.8 9.9 10.1 10.1 10.2 10.2
2 1.14 0.0 0.0 0.0 55.5 68.7 80.9 85.4 87.7 90.1 91.4
3 0.35 0.1 42.7 82.1 93.4 96.8 99.3 100.5 101.2 102.1 102.7
4 0.16 47.3 72.9 82.9 84.7 85.0 85.1 85.1 85.1 85.2 85.2
5 Not reached 6.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
6 Not reached 18.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
7 0.25 0.0 68.0 73.5 74.9 75.3 75.4 75.5 75.5 75.5 75.5
8 Not reached 53.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
9 0.06 2.1 2.2 2.2 2.2 2.2 2.3 2.4 2.4 2.4 2.4
10 0.70 0.0 0.0 25.4 42.8 44.7 45.1 45.2 45.2 45.2 45.2
11 1.45 0.0 0.0 0.0 10.8 15.3 16.4 16.6 16.6 16.4 16.1
12 1.09 0.0 0.0 0.0 13.1 15.4 15.5 15.5 15.6 15.6 15.6
13 Not reached 4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
14 0.80 0.0 0.0 11.4 21.9 23.2 23.3 23.3 23.4 23.5 23.5
15 3.19 0.0 0.0 0.0 0.0 0.0 2.0 2.2 2.2 2.3 2.3
16 1.12 0.0 0.0 0.0 16.8 19.0 19.6 19.7 19.7 19.8 19.8
17 Not reached 17 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Fig. 5. Cross sections of the gas phase saturation (green) and dissolved CH
4
mass concentration (blue) after simulated times of 1 year (a) and 10 years (b). Results are
shown for the case of a coarse sand, ne sand and very ne sand, corresponding to scenarios 1, 5 and 6 (Table 1). Vertical dashed lines indicate the relative position of
the CH
4
inlet boundary at the base of the model at 60 m depth. (For interpretation of the references to colour in this gure legend, the reader is referred to the web
version of this article.)
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
8
retention capacity is therefore greatly increased at these depths and
model outcomes are indeed signicantly changed, as migration time to
the top of the aquifer increased from 0.16 year to 0.7 year at 240 m
depth and 1.45 year at 480 m depth (Table 2). By extension, Q
out
/Q
in
stabilized later and decreased from 85% to 45% at 240 m depth and to
16% at 480 m depth (Fig. 4).
3.2.6. Impact of the residual gas saturation
In scenario 12 the residual gas saturation was lowered to 0%.
Migration time to the top of the aquifer decreased to 1.09 years, com-
pared to 1.65 years in the base case scenario. Q
out
/Q
in
increased from
10 to 16%. This shows that even very small variations in residual gas
saturation impact gas retention and migration. In scenario 13 the re-
sidual gas saturation was raised to 15%. As the residual gas saturation
acts as a sort of threshold value, below which ow cannot start, this
causes gas phase saturations to become much larger than in the base
case scenario, and thus more gas is stored in the porous medium.
Migration of the gas phase plume slows down considerably and had
nearly stagnated after 20 years of simulation at a depth of 4 m below
the top of the aquifer (Table 2).
3.2.7. Impact of anisotropy
Variations in anisotropy were modelled by raising the vertical per-
meability to equal the horizontal permeability in scenario 14 (k
h
/
k
v
= 1) and lowering it to 1/10th of the horizontal permeability in
scenario 15 (k
h
/k
v
= 10, Table 1). The expected higher and lower gas
migration speed results in the gas reaching the top of the aquifer in only
0.80 years for the lower anisotropy case and in 3.19 years in the higher
anisotropy case, compared to 1.65 in the base case scenario (Table 2).
In turn, the change in gas migration velocity aects the amount of
methane that is dissolved and Q
out
/Q
in
after 20 years of leakage
changes to 24% and 2% for scenarios 14 and 15, respectively.
3.2.8. Impact of the area over which the inux occurs
The results of earlier simulations and the dimensional analysis de-
monstrate that for the ow rates considered in this study and parameter
values typical of sandy aquifers, buoyancy driven, vertical gas migra-
tion dominates. As a consequence, the degree of dissolutive retention is
not only related to the groundwater velocity, ow rate, and sediment
properties, but also depends on the area of the inlet. This determines
the leakage ux (L
3
.L
2
.d
1
) and hence the amount of water that is
available for methane to dissolve in. To investigate the importance of
this eect, the inlet boundary size was reduced from 2 × 2 m
2
to 1 × 1
m
2
in scenario 16, while keeping Q
in
equal. Upward migration velocity
increased considerably and gas reached to the top of the aquifer in
1.12 years while Q
out
/Q
in
after 20 years increased to 20%, compared
1.65 years and 10% in the base case. When increasing the inlet area to
4×4m
2
(scenario 17), migration velocity was signicantly reduced
such that stagnation occurred at 17 m below the top of the aquifer
(Table 2).
3.3. Layered case studies based on real sites
3.3.1. Case study 1: Sleen site
The modelling domain of the rst case study consists of a simplied
hydrogeology with 7 main layers, cumulatively 126 m thick (Fig. 2).
The lowest layer is characterized by a high anisotropy factor of 125 (k
h
/
k
v
) and low vertical permeability of 4.210
14
m
2
. This complex layer is
overlain by 5 sandy sections with slightly varying permeabilities. No-
tably, the third layer from the bottom is a coarse sand and has the
highest horizontal permeability of 1.110
10
m
2
. The sandy units are
only interbedded by a 1 m thick complex layer with a low vertical
permeability from 1 to 2 m depth. Two scenarios were carried out using
this parametrization, one with a constant head gradient of 0.25 m.km
1
and the other with 1 m.km
1
. Given the constant head gradient,
groundwater velocities in the coarse sandy layer are highest, with Darcy
velocities of 8.6 and 34.2 m.yr
1
in the rst and second scenario, re-
spectively (Fig. 2).
The low k
v
in the highly anisotropic bottom layer causes signicant
lateral spreading of the methane plume to occur, with gas phase sa-
turations greater than 6% close to the inlet. In the 0.25 m.km
1
head
gradient scenario, this spreading occurs more or less symmetrically in
each direction (Fig. 6). In the 1.00 m.km
1
scenario the plume is tilted
more strongly in the direction of groundwater ow. Gas migration
through this 49 m thick layer takes roughly 3.2 years in both scenarios.
In comparison, migration through the remaining 77 m thick modelling
domain occurs in only 1.9 years in the rst scenario (Table S2). Dis-
solution of gaseous methane in the horizontally owing groundwater is
highly signicant, as evidenced by the stagnation of the gas phase
plume in the high permeability coarse sand layer in the 1.00 m.km
1
scenario (Fig. 6). The high groundwater velocities here, combined with
the lateral spreading of the methane plume that occurs in the under-
lying layers, causes the 10 m
3
CH
4atm
.d
1
ow rate to be completely
dissolved once the plume reaches a depth of around 52 m. In the
0.25 m.km
1
scenario, Q
out
/Q
in
stabilizes after 20 years at 41% (Table
S2). The radius of the gaseous plume at the top of the simulated section
(which in this case represents the groundwater table) reaches a max-
imum value of around 20 m, and is slightly elongated in the direction of
groundwater ow.
3.3.2. Case study 2: Monster site
Similar to the rst case study, the lowest layer of the second case
study consists of a complex unit with a low vertical permeability and
high anisotropy. However, the overlying stratigraphy is notably dif-
ferent than in the rst case study and is characterized by the presence of
5 low permeability, high entry pressure clay layers that alternate the
Fig. 6. Gas phase saturation and dissolved CH
4
mass concentration after 6 years
of a 10 m
3
CH
4atm
.d
1 leakage rate for the rst case study site (Sleen). Top
gures shows the results of the scenario with a head gradient of 0.25 m.km
1
,
lower gure for the scenario with a head gradient of 1 m.km
1
. Horizontal
dashed lines delineate the dierent hydrogeological layers (Fig. 2), with darker
shaded layers representing the clayey and complex layers that have a higher
entry pressure and a lower permeability than the sandy layers.
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
9
sandy aquifers (Fig. 2). Particularly the rst and second clay layers from
the bottom, at depths of 155143 m and 119110 m, have much lower
vertical permeabilities (~410
15
m
2
). Accordingly, they are also as-
signed higher entry pressures (~7.8 kPa) than the sandy units, and the
values considered in the sensitivity analysis. The sandy aquifers are also
slightly less permeable than in the rst case study, resulting in Darcy
velocities of ~0.7 m.yr
1
and ~2.7 m.yr
1
for the low and high head
gradient scenario's, respectively. The modelled section is 176 m thick,
and the top of the model is at 55 m depth. Gas migrating through the
top boundary would enter the overlying Holocene deposits, which were
not considered in our simulations.
As in the rst case study, the complex unit at the base of the model
causes substantial lateral spreading (Fig. 7). After around 3 years, the
migrating gas phase encounters the rst clay layer which results in
signicant gas phase pooling. Invasion of the clay layer only occurs
after gaseous saturations exceed 6%. At that saturation, the combined
capillary and gravitational forces were large enough to overcome the
entry pressure barrier. In the low head gradient scenario, this gaseous
pool grows to saturations exceeding 10% and a diameter of roughly
70 m. In the high head gradient scenario, less pooling occurs due to the
larger amount of methane that is already dissolved in the underlying
units. However, the pool is more extensive in the direction of
groundwater ow. Similar behavior is observed at the 2nd clay layer,
but entry pressures assigned to the remaining three clay layers were not
sucient to cause signicant pooling. In the low head gradient sce-
nario, the gas phase plume reaches the top of the modelled domain after
16.2 years. Even at the end of the 50 year simulation time, Q
out
/Q
in
had
not fully stabilized reaching a value of 46.3% (Table S2). Due to the
relatively slow upward migration, pooling of gas below the clay layers,
and low groundwater velocities in the clay layers, the dissolved me-
thane plume takes on a pine tree-like shape (Fig. 7). In the high head
gradient scenario, gas phase migration stagnates at 103 m depth, or
48 m below the top of the modelling domain. Notably, the center of the
gaseous plume at that depth can be seen to have shifted by roughly
30 m in the direction of groundwater ow. This is caused by the im-
posed hydraulic gradient, which results in slightly lower water pres-
sures downstream. Given that the entry pressure is dened as P
e
=P
g
-
P
l
, the migrating gas will preferentially invade cells those cells and the
plume shifts in the direction of groundwater ow.
4. Discussion
4.1. Upward migration and retention of methane gas
For the simulated homogeneous sandy aquifers and leakage ow
rates up to 1 m
3
CH
4atm
.d
1
, gas migration was shown to be primarily
buoyancy driven and vertical (Fig. 5), as expected based on an analysis
of the relative magnitude of gravitational and capillary forces (Fig. S4),
and as observed by other researchers for weakly isotopic sandy porous
media (Klazinga et al., 2019). A limited amount of up or down gradient
ow of gas only occurred for two simulations with more ne grained
sediment properties (Fig. 5), due to their lower vertical permeability
and an increase in capillary forces. Gas phase saturations remained low
and reached up to around 5%, depending on the imposed ow rate and
permeability. At these saturations, the potential mass storage of me-
thane in the aqueous phase is either larger than or equal to the storage
in the gaseous phase (Fig. 3), even without considering the additional
dissolution capacity with groundwater ow.
Gaseous retention has been shown to be signicant in simulations of
methane migrating from gas reservoirs towards shallow aquifers, when
assuming large residual gas saturations of up to 30% (Kissinger et al.,
2013). However, there is little consensus on appropriate values to use
for simulating gas migration through unconsolidated aquifers. While
Klazinga et al. (2019) used a single value of 10%, Rice et al. (2018)
chose a negligibly small value, arguing that gas cannot be trapped
during processes with strictly increasing gaseous saturations (i.e. drai-
nage), which is the case in our numerical simulations. However, gas
migration through real-world, heterogeneous sediments is likely to be a
more dynamic process. As a result, intermittent periods of imbibition
may still occur, particularly when leakage at the leakage point is not
entirely continuous. In spite of this uncertainty, model outcomes
showed relatively little sensitivity to a change in residual gas saturation
from 1% to 0% (scenario 12, Table 2), because gas migration is more
strongly controlled by retention in the aqueous phase at those con-
centrations. On the contrary, an imposed residual saturation of 15%
exerted a strong control on upward gas migration leading to a much
more slowly developing gas plume (scenario 13, Table 2).
4.2. Retention of methane by dissolution
The sensitivity analysis showed that dissolutive retention of me-
thane in unconsolidated sedimentary aquifers can play a major role in
limiting upward gas migration. Even low groundwater ow velocities
(1 m.yr
1
) resulted in signicant methane retention (Fig. 4a). On the
contrary, retention was negligible in the absence of groundwater ow.
This shows that it is primarily driven by advective transport of dis-
solved methane away from the gas phase plume, rather than diusive
transport. Hence, retention is proportional to the groundwater velocity
Fig. 7. Gas phase saturation and dissolved CH
4
mass concentration after
15 years of a 10 m
3
CH
4atm
.d
1 leakage rate for the second case study site
(Monster). Top gures show the results of the scenario with a head gradient of
0.25 m.km
1
, lower gures for the scenario with a head gradient of 1 m.km
1
.
Horizontal dashed lines delineate the dierent hydrogeological layers (Fig. 2),
with darker shaded layers representing the clayey and complex layers that have
a higher entry pressure and a lower permeability than the sandy layers.
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
10
and methane solubility. The eect of retention on methane migration
was therefore also much less signicant for a simulation where the
initial aqueous CH
4
concentration was raised to 95% solubility. In this
scenario, migration time through the 60 m thick aquifer was just
0.35 years, compared to 1.65 years in the base case with fully un-
saturated groundwater (Fig. 4a). This also highlights that upward me-
thane gas migration can be more rapid in areas with pre-occurring
concentrations of dissolved methane. However, natural methane con-
centrations near solubility are rarely found in shallow groundwater. As
an example, the methane solubility at just 60 m depth below the water
table (~200 mg.L
1
, Fig. S3) already exceeds the maximum con-
centration ever observed in Dutch shallow groundwater (~120 mg.L
1
,
Cirkel et al., 2015). Hence, dissolutive retention can still be signicant
even for locations with relatively high methane concentrations.
Besides groundwater velocity and initial methane concentrations,
model outcomes indicate a strong control of the sediment properties
and the magnitude of the leak, as they determine the velocity and di-
mensions of the upward migrating gas phase. For a given groundwater
velocity, a more dilute or more slowly migrating gas plume allows for a
larger fraction of the total ux to dissolve. Sediment properties that
limit upward migration include the vertical permeability, the residual
gas saturation, and the capillary pressure-saturation and relative per-
meability relations. Using the Brooks-Corey model, the latter two are
dened by the capillary entry pressure and the pore size distribution
index. Indeed, for two simulated scenarios with sediment properties
associated with more nely grained sediments (D50 0.16 and 0.09 mm,
compared to 0.61 mm in the base case), methane migration resulting
from a 0.1 m
3
CH
4atm
.d
1
ow rate stagnated entirely due to the
complete dissolution of the migrating gas plume at depth (Fig. 5).
Model outcomes also show that the dimension of the inlet over
which the leakage is imposed also has a strong inuence. Here, we
assumed that gas migrated into the model domain over a 2 × 2 m
2
area,
and that the origin of the leak was somewhere below the bottom of the
model domain, for example caused by gas circumvention or SCP-in-
duced leakage (Fig. 1). This also explains why retention was greatly
enhanced in two simulated case studies where the sandy aquifers where
alternated by sediment beds of lower permeability (Fig. 2). Besides
their greater thickness and depth, such layers cause signicant
spreading of the gaseous plume to occur, which in turn enhances dis-
solutive retention signicantly (Figs. 6 and 7). Flow rates of 10 m
3
CH
4atm
.d
1
were either completely dissolved, causing migration to
stagnate well below the top of the modelled domain, or took several
decades to fully develop (Table S2). Signicant lateral spreading of
migrating methane was also shown to occur for SCP-induced leakage
originating in a low permeable shale below a shallow aquifer (Rice
et al., 2018). In such conditions, our results show that more permeable
overlying layers can become strong barriers to upward migration due to
dissolutive retention. Conversely, the impact of dissolutive retention
would be much smaller when methane migration is focused through
high permeable pathways such as faults or fractures.
As methane solubility is strongly depth dependent (Fig. S3), dis-
solutive retention is likely to be even more signicant when gas mi-
gration occurs at greater depths than those considered in this study.
However, the eect of increasing solubility due to increasing pressures
may be counteracted somewhat by increasing salinities, which decrease
methane solubility. Also, groundwater is generally more stagnant at
depth than in surcial aquifers. As shown in this work, dissolutive re-
tention is limited in the absence of groundwater ow. The complex
interaction of these processes, and their eect on the fate of gas mi-
gration occurring at depths greater than those considered in this study
(up to 480 mbgl), would be a relevant subject for further research.
4.3. Gaseous pooling and permeability clay layers to gas migration
While signicant gas phase pooling occurred below the low
permeable clay layers (k
v
up to minimum of 410
15
m
2
,P
e
up to
7.8 kPa) in the 2nd case study, they were ultimately permeable to the
migrating gas (Fig. 7). Invasion of a higher entry pressure layer requires
a large enough gas accumulation in the underlying layer such that the
combined capillary and gravitational forces exceed the entry pressure
barrier. For large enough entry pressure values and horizontal barriers,
such conditions would not be obtained as the accumulating gas pool
spreads out laterally into a thin, pancake like shape. High resolution
permeability measurements have shown that thin, low permeability
clay lenses can be present within larger clayey formations, with per-
meabilities that are orders of magnitude lower than the average per-
meability of the unit (Rogiers et al., 2014). Given their low perme-
ability and presumably higher entry pressure, these lenses would likely
act as impermeable barriers to gas phase ow, essentially turning the
entire formation into a cap rock. However, the ubiquity and lateral
extent of these lenses is poorly understood. Therefore, further research
into how heterogeneity of clay deposits impacts gas migration would be
required.
4.4. Continuum modelling of gas migration
The low observed saturations call into question whether the injected
gas actually forms a continuous phase, and hence whether the widely-
used Darcy's law approach is applicable for modelling gas migration
through unconsolidated aquifers. Air sparging experiments have shown
that, for unconsolidated sediments, the injected gas phase migrates as
continuous gaseous channels when grain sizes are 1 mm. For larger
grain sizes the ow pattern gradually changes to discontinuous bubble
ow (Brooks et al., 1999). The grain sizes considered in this study fall in
the former category, and hence channel ow may be expected to
dominate, although the ow type is also dependent on the ux size.
Furthermore, whether such migration takes the form of a dendritic
network of small channels or a smaller amount of larger gas channels is
poorly understood (Brooks et al., 1999). Continuous dewatered chan-
nels through which the gas phase dominantly migrates also occurred in
the eld experiments of gas migration by Cahill et al. (2018). Observed
dissolved methane concentrations below theoretical solubility were
attributed to this ow pattern, as the available contact area over which
mass transfer between phases can occur is smaller, and the collected
groundwater samples aggregate both water that has been in contact
with a gas channel and water that has not.
As this process is not captured in our simulations, the resulting
amount of dissolution may be overestimated. On the other hand, me-
chanical dispersion was not included in our simulations. Excluding
dispersion actually leads to an underestimation of the dissolutive ca-
pacity, as dispersion would spread out both the gas phase and the
dissolved methane more rapidly, allowing for more contact with the
water and quicker transport of dissolved methane away from the gas
phase plume. Ultimately, the suitability of continuum models for
modelling gas migration should be further analyzed through re-
production of experimental results. A promising experimental method
which could be employed for this purpose was recently published by
Van De Ven and Mumford (2018), who injected gaseous CO
2
at the
bottom of a 2D ow cell and used pH as a proxy to visually track dis-
solved CO
2
concentrations. Reproducing such experiments with nu-
merical simulation would be a good model validation strategy.
4.5. Implications for methane monitoring
From this study a number of important implications for monitoring
of gas migration and leakages in the vicinity of oil and gas wells can be
derived. Notably, surcial expressions of gas leakage originating at
depth in unconsolidated sedimentary basins may take several years to
manifest in settings dominated by relatively coarse grained sands, and
may take at least up to several decades when migrating gas also en-
counters interbedded low permeable layers. Surcial detection of
leaking oil and gas wellbores may therefore only become possible long
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
11
after the onset of leakage, and potentially after wells have already been
abandoned. This is particularly relevant as in many oil and gas jur-
isdictions decommissioned wells are cut and buried below the ground
surface (Davies et al., 2014;Schout et al., 2019), in which case direct
measurements of well integrity failure at the wellhead are no longer
possible. Furthermore, for example in the Netherlands there are cur-
rently no regulations in place that require operators to monitor the
locations of gas wells for extensive periods post-decommissioning.
Relying on surcial ux measurements or near-surface groundwater
wells may lead to the underestimation of the actual occurrence and
magnitude of gas leakage, and groundwater contamination at depth
may remain unnoticed. Hence, measuring dissolved gas molecular and
isotopic compositions in groundwater wells in close proximity to po-
tential leakage points is likely a more reliable method. When surface
expressions do occur, the dissolution of methane may still be such that
measured uxes at the surface only represent a fraction of the ow rate
at the leakage point. Lastly, our simulations show that regional
groundwater ow may cause the center of a migrating gas plume to be
transported several tens of meters downstream from the well before
reaching the surface. Recent eld studies have employed search radii of
15 m (Schout et al., 2019) and 20 m (Forde et al., 2019) from the co-
ordinates of gas wells when carrying out surcial ux measurements.
While likely suitable in most cases, widening the investigated area in
the direction of groundwater ow may need to be considered de-
pending on the conditions.
5. Conclusions
Oil and gas well failure leading to gas migration in the subsurface
poses both an environmental and safety hazard. Eective mitigation of
these hazards and the proper assessment of their impact is reliant on the
successful detection and quantication of leaks. In this study, upward
methane migration through unconsolidated aquifers was analyzed
using two-phase, two-component (H
2
O and CH
4
) numerical simulations
in DuMu
x
. Results show that the retention of migrating methane due to
dissolution into laterally owing groundwater can become signicant
at groundwater Darcy velocities as low as 1 m.yr
1
. Retention was
shown to be dependent on a complex interaction between the lateral
groundwater velocity, the depth of the leak, and the velocity and shape
of the upwardly migrating gas plume. The latter is in turn a function of
the leakage ux and sediment properties (permeability, capillary
pressure-saturation and relative permeability relations, and residual gas
saturation). Across a range of conditions representative of un-
consolidated aquifers and leak sizes up to 1 m
3
CH
4atm
.d
1
, the time it
took before the gas plume propagated through to the top of the mod-
elled domain varied from less than 2 months to 5 years. In some sce-
narios, the total methane leakage rate was dissolved and transported
laterally, causing the gas plume to stagnate at depth.
Subsurface methane retention was even more pronounced in addi-
tional simulations of migration through stratied sequences based on
the hydrogeological conditions at two leaking gas wells in the
Netherlands, consisting of a number of alternating sedimentary units
ranging in grain size from clays to coarse sands. Under such conditions,
depending on the imposed groundwater head gradient, ow rates of up
to 10 m
3
CH
4atm
.d
1
were either entirely retained in subsurface or took
several decades to fully develop. Not only the presence of ne grained
and low permeable layers formed barriers to upward gas migration, but
also high permeable sands with large groundwater velocities. These
allow for a larger dissolutive capacity, particularly when the migrating
gas phase has been spread out laterally by underlying low permeability
units. Overall, the results of this study show that for the most commonly
observed methane leakage rates (0.110 m
3
.d
1
), unconsolidated
aquifer systems with lateral groundwater ow can retain signicant
amounts of migrating methane due to dissolution. Consequently, re-
sulting atmospheric methane emissions above such leaks may be de-
layed with decades after the onset of leakage, signicantly reduced or
prevented entirely. Therefore, groundwater contamination and future
explosion hazards may go unnoticed.
Declaration of Competing Interest
None
Acknowledgements
This work is part of the research program Shale Gas and Water
with project number 859.14.001, which is nanced by the Netherlands
Organization for Scientic Research (NWO). We thank the German
Research Foundation (DFG) for providing nancial support for a visit to
Stuttgart University in March of 2019 through the SFB1313, project
number 327154368. Furthermore, we thank Dennis Gläser of Stuttgart
University for extensive assistance with setting up the DuMu
x
simula-
tions and many helpful discussions about numerical simulation of
multiphase ow and transport in general.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.jconhyd.2020.103619.
References
Ackermann, S., Beck, M., Becker, B., Class, H., Fetzer, T., Flemisch, B., et al., 2017.
DuMuX 2.11.0. https://doi.org/10.5281/ZENODO.439488.
Bloemendal, M., Hartog, N., 2018. Analysis of the impact of storage conditions on the
thermal recovery eciency of low-temperature ATES systems. Geothermics 71,
306319. https://doi.org/10.1016/j.geothermics.2017.10.009.
Brooks, R.H., Corey, A.T., 1964. Hydraulic Properties of Porous Media. Hydrology
Papers. Colorado State Universityhttps://doi.org/10.13031/2013.40684.
Brooks, M.C., Wise, W.R., Annable, M.D., 1999. Fundamental changes in in situ air
sparging ow patterns. Groundw. Monit. Remediat. 19 (2), 105113. https://doi.org/
10.1111/j.1745-6592.1999.tb00211.x.
Brownlow, J.W., James, S.C., Yelderman, J.C., 2016. Inuence of hydraulic fracturing on
overlying aquifers in the presence of leaky abandoned wells. Groundwater 54 (6),
781792. https://doi.org/10.1111/gwat.12431.
Cahill, A.G., Steelman, C.M., Forde, O., Kuloyo, O., Emil Ru, S., Mayer, B., et al., 2017.
Mobility and persistence of methane in groundwater in a controlled-release eld
experiment. Nat. Geosci. https://doi.org/10.1038/ngeo2919.
Cahill, A.G., Parker, B.L., Mayer, B., Mayer, K.U., Cherry, J.A., 2018. High resolution
spatial and temporal evolution of dissolved gases in groundwater during a controlled
natural gas release experiment. Sci. Total Environ. 622623, 11781192. https://doi.
org/10.1016/j.scitotenv.2017.12.049.
Cahill, A.G., Beckie, R., Ladd, B., Sandl, E., Goetz, M., Chao, J., et al., 2019. Advancing
knowledge of gas migration and fugitive gas from energy wells in Northeast British
Columbia, Canada. Greenh. Gases: Sci. Technol. 18 (2019), 118. https://doi.org/10.
1002/ghg.1856.
Chilingar, G.V., Endres, B., 2005. Environmental hazards posed by the Los Angeles Basin
urban oilelds: an historical perspective of lessons learned. Environ. Geol. https://
doi.org/10.1007/s00254-004-1159-0.
Cirkel, G., Hartog, N., De La, B., Gonzalez, L., Stuyfzand, P., 2015. Methaan in Ondiep
Nederlands Grondwater: Verbinding Met De Diepe Ondergrond? H2O. Retrieved
from. https://goo.gl/7GCtkS.
Clayton, W.S., 1999. Eects of pore scale dead-end air ngers on relative permeabilities
for air sparging in soils. Water Resour. Res. 35 (10), 29092919. https://doi.org/10.
1029/1999WR900202.
DAniello, A., Fabbricino, M., Ducci, D., Pianese, D., 2019. Numerical Investigation of a
Methane Leakage from a Geothermal Well into a Shallow Aquifer.
Groundwaterhttps://doi.org/10.1111/gwat.12943. (September).
Davies, R.J., Almond, S., Ward, R.S., Jackson, R.B., Adams, C., Worrall, F., et al., 2014. Oil
and gas wells and their integrity: implications for shale and unconventional resource
exploitation. Mar. Pet. Geol. 56, 239254. https://doi.org/10.1016/j.marpetgeo.
2014.03.001.
de Vries, J.J., 2007. Groundwater. In: Wong, T.E., Batjes, D.A.J., de Jager, J. (Eds.),
Geology of the Netherlands. Royal Netherlands Academy of Arts and Sciences,
Amsterdam, pp. 354. Retrieved from. https://www.press.uchicago.edu/ucp/books/
book/distributed/G/bo5456650.html.
Dusseault, M.B., Jackson, R.E., MacDonald, D., 2014. Towards a road map for mitigating
the rates and occurrences of long-term wellbore leakage. Georma 169. Retrieved
from. http://georma.com/wp-content/uploads/2015/05/lwp-nal-report_
compressed.pdf.
Erno, B., Schmitz, R., 1996. Measurements of soil gas migration around oil and gas wells
in the Lloydminster area. J. Can. Pet. Technol. 35 (7), 3746. https://doi.org/10.
2118/96-07-05.
Fernández-Prini, R., Alvarez, J.L., Harvey, A.H., 2003. Henrys constants and vapor-liquid
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
12
distribution constants for gaseous solutes in H2O and D2O at high temperatures. J.
Phys. Chem. Ref. Data 32 (2), 903916. https://doi.org/10.1063/1.1564818.
Flemisch, B., Darcis, M., Erbertseder, K., Faigle, B., Lauser, A., Mosthaf, K., et al., 2011.
DuMux: DUNE for multi-{phase,component,scale,physics,...} ow and transport in
porous media. Adv. Water Resour. 34 (9), 11021112. https://doi.org/10.1016/j.
advwatres.2011.03.007.
Forde, O.N., Mayer, K.U., Hunkeler, D., 2019. Identication, spatial extent and dis-
tribution of fugitive gas migration on the well pad scale. Sci. Total Environ. 652,
356366. https://doi.org/10.1016/j.scitotenv.2018.10.217.
Fuller, E.N., Schettler, P.D., Giddings, J.C., 1966. A new method for prediction of binary
gas-phase diusion coecients. Ind. Eng. Chem. 58 (5), 1827. https://doi.org/10.
1021/ie50677a007.
Gasda, S.E., Bachu, S., Celia, M.A., 2004. Spatial characterization of the location of po-
tentially leaky wells penetrating a deep saline aquifer in a mature sedimentary basin.
Environ. Geol. 46 (67), 707720. https://doi.org/10.1007/s00254-004-1073-5.
Helmig, R., 1997. Multiphase Flow and Transport Processes in the Subsurface. A
Contribution to the Modeling of Hydrosystems. Environmental Engineering. Springer
Retrieved from. https://www.springer.com/gp/book/9783642645457.
Kang, M., Kanno, C.M., Reid, M.C., Zhang, X., Mauzerall, D.L., Celia, M.A., et al., 2014.
Direct measurements of methane emissions from abandoned oil and gas wells in
Pennsylvania. Proc. Natl. Acad. Sci. 111 (51), 1817318177. https://doi.org/10.
1073/pnas.1408315111.
Kang, M., Christian, S., Celia, M.A., Mauzerall, D.L., Bill, M., Miller, A.R., et al., 2016.
Identication and characterization of high methane-emitting abandoned oil and gas
wells. Proc. Natl. Acad. Sci. 113 (48), 1363613641. https://doi.org/10.1073/pnas.
1605913113.
King, G.E., King, D.E., 2013. Environmental risk arising from well-construction failure
dierences between barrier and well failure, and estimates of failure frequency across
common well types, locations, and well age. SPE Prod. Oper. 28 (4), 323344.
https://doi.org/10.2118/166142-PA.
Kissinger, A., Helmig, R., Ebigbo, A., Class, H., Lange, T., Sauter, M., et al., 2013.
Hydraulic fracturing in unconventional gas reservoirs: risks in the geological system,
part 2: modelling the transport of fracturing uids, brine and methane. Environ.
Earth Sci. 70 (8), 38553873. https://doi.org/10.1007/s12665-013-2578-6.
Klazinga, D.R., Steelman, C.M., Cahill, A.G., Walton, K.M., Endres, A.L., Parker, B.L.,
2019. Methane gas transport in unconned aquifers: a numerical sensitivity study of
a controlled release experiment at CFB Borden. J. Contam. Hydrol. 225, 103506.
https://doi.org/10.1016/j.jconhyd.2019.103506.
Kombrink, H., Doornebal, J.C., Duin, E.J.T., Den Dulk, M., Van Gessel, S.F., Ten Veen,
J.H., Witmans, N., 2012. New insights into the geological structure of the
Netherlands: results of a detailed mapping project. Neth. J. Geosci./ Geol. Mijnb. 91
(4), 419446. https://doi.org/10.1017/S0016774600000329.
Kopp, A., 2009. Evaluation of CO2 Injection Processes in Geological Formations for Site
Screening. Stuttgart University.
Lackey, G., Rajaram, H., 2018. Modeling gas migration, sustained casing pressure, and
surface-casing-vent ow in onshore oil and gas wells. Water Resour. Res. 126.
https://doi.org/10.1029/2018WR024066.
McMahon, P.B., Thomas, J.C., Crawford, J.T., Dornblaser, M.M., Hunt, A.G., 2018.
Methane in groundwater from a leaking gas well, Piceance Basin, Colorado, USA. Sci.
Total Environ. 634, 791801. https://doi.org/10.1016/j.scitotenv.2018.03.371.
Moortgat, J., Schwartz, F.W., Darrah, T.H., 2018. Numerical modeling of methane
leakage from a faulty natural gas well into fractured tight formations. Groundwater
56 (2), 163175. https://doi.org/10.1111/gwat.12630.
Nowamooz, A., Lemieux, J.M., Molson, J., Therrien, R., 2015. Numerical investigation of
methane and formation uid leakage along the casing of a decommissioned shale gas
well. Water Resour. Res. 51 (6), 45924622. https://doi.org/10.1002/
2014WR016146.
Reagan, M.T., Moridis, G.J., Keen, N.D., Johnson, J.N., 2015. Numerical simulation of the
environmental impact of hydraulic fracturing of tight/shale gas reservoirs on near-
surface groundwater: background, base cases, shallow reservoirs, short-term gas, and
water transport. Water Resour. Res. 51 (4), 25432573. https://doi.org/10.1002/
2014WR016086.
Reid, R., Prausnitz, J., Poling, B., 1987. The Properties of Gases and Liquids, (Fourth Edi).
McGraw-Hill, New York.
Rice, A.K., McCray, J.E., Singha, K., 2018. Methane leakage from hydrocarbon wellbores
into overlying groundwater: numerical investigation of the multiphase ow processes
governing migration. Water Resour. Res. https://doi.org/10.1002/2017WR021365.
Rogiers, B., Winters, P., Huysmans, M., Beerten, K., Mallants, D., Gedeon, M., et al., 2014.
High-resolution saturated hydraulic conductivity logging of borehole cores using air
permeability measurements. Hydrogeol. J. 22 (6), 13451358. https://doi.org/10.
1007/s10040-014-1144-y.
Roy, N., Molson, J., Lemieux, J.M., Van Stempvoort, D., Nowamooz, A., 2016. Three-
dimensional numerical simulations of methane gas migration from decommissioned
hydrocarbon production wells into shallow aquifers. Water Resour. Res. 52 (7),
55985618. https://doi.org/10.1002/2016WR018686.
Schout, G., Hartog, N., Hassanizadeh, S.M., Grioen, J., 2018. Impact of an historic
underground gas well blowout on the current methane chemistry in a shallow
groundwater system. Proc. Natl. Acad. Sci. 115 (2), 296301. https://doi.org/10.
1073/pnas.1711472115.
Schout, G., Grioen, J., Hassanizadeh, S.M., Cardon de Lichtbuer, G., Hartog, N., 2019.
Occurrence and fate of methane leakage from cut and buried abandoned gas wells in
the Netherlands. Sci. Total Environ. 659, 773782. https://doi.org/10.1016/j.
scitotenv.2018.12.339.
Schwartz, M.O., 2015. Modelling the hypothetical methane-leakage in a shale-gas project
and the impact on groundwater quality. Environ. Earth Sci. 73 (8), 46194632.
https://doi.org/10.1007/s12665-014-3787-3.
SodM, 2019. De integriteit van onshore putten in Nederland. Den Haag Retrieved from.
https://www.sodm.nl/documenten/rapporten/2019/02/07/de-integriteit-van-
onshore-putten-in-nederland.
Tatomir, A., Mcdermott, C., Bensabat, J., Class, H., Edlmann, K., Taherdangkoo, R.,
Sauter, M., 2018. Conceptual model development using a generic features, events,
and processes (FEP) database for assessing the potential impact of hydraulic frac-
turing on groundwater aquifers. Adv. Geosci. 45, 185192. https://doi.org/10.5194/
adgeo-45-185-2018.
Townsend-Small, A., Ferrara, T.W., Lyon, D.R., Fries, A.E., Lamb, B.K., 2016. Emissions of
coalbed and natural gas methane from abandoned oil and gas wells in the United
States. Geophys. Res. Lett. 43 (5), 22832290. https://doi.org/10.1002/
2015GL067623.
Van De Ven, C.J.C., Mumford, K.G., 2018. Visualization of gas dissolution following
upward gas migration in porous media: technique and implications for stray gas. Adv.
Water Resour. 115, 3343. https://doi.org/10.1016/j.advwatres.2018.02.015.
Van Stempvoort, D., Maathuis, H., Jaworski, E., Mayer, B., Rich, K., 2005. Oxidation of
fugitive methane in ground water linked to bacterial sulfate reduction. Ground Water
43 (2), 187199. https://doi.org/10.1111/j.1745-6584.2005.0005.x.
Vengosh, A., Jackson, R.B., Warner, N., Darrah, T.H., Kondash, A., 2014. A critical review
of the risks to water resources from unconventional shale gas development and hy-
draulic fracturing in the United States. Environ. Sci. Technol. 48 (15), 83348348.
https://doi.org/10.1021/es405118y.
Vernes, R.W., van Doorn, T.H., 2005. Van Gidslaag naar Hydrogeologische Eenheid;
Toelichting op de totstandkoming van de dataset REGIS II. NITG-05-038-B. TNO,
Utrecht.
Verweij, J.M., Nelskamp, S.N., Ten Veen, J.H., De Bruin, G., Geel, K., Donders, T.H., 2018.
Generation, migration, entrapment and leakage of microbial gas in the Dutch part of
the Southern North Sea Delta. Mar. Pet. Geol. 97 (July), 493516. https://doi.org/10.
1016/j.marpetgeo.2018.07.034.
Wagner, W., Pruss, A., 2002. The IAPWS formulation 1995 for the thermodynamic
properties of ordinary water substance for general and scientic use. J. Phys. Chem.
Ref. Data 31 (2), 387535. https://doi.org/10.1063/1.1461829.
Wentworth, C.K., 1922. A scale of grade and class terms for clastic sediments author ( s ):
Chester K. Wentworth published by: the University of Chicago Press Stable URL:
http://www.jstor.org/stable/30063207. J. Geol. 30 (5), 377392. https://doi.org/
10.1086/622910.
Wisen, J., Chesnaux, R., Werring, J., Wendling, G., Baudron, P., Barbecot, F., 2019. A
portrait of wellbore leakage in northeastern British. In: Proceedings of the National
Academy of Sciences of the United States of America, pp. 110. https://doi.org/10.
1073/pnas.1817929116.
Woda, J., Wen, T., Oakley, D., Yoxtheimer, D., Engelder, T., Castro, M.C., Brantley, S.L.,
2018. Detecting and explaining why aquifers occasionally become degraded near
hydraulically fractured shale gas wells. Proc. Natl. Acad. Sci. https://doi.org/10.
1073/pnas.1809013115.
Wolfe, A.L., Wilkin, R.T., 2017. Evidence of sulfate-dependent anaerobic methane oxi-
dation within an area impacted by coalbed methane-related gas migration. Environ.
Sci. Technol. 51 (3), 19011909. https://doi.org/10.1021/acs.est.6b03709.
G. Schout, et al. Journal of Contaminant Hydrology 230 (2020) 103619
13
... It was also found that the surface flux to the atmosphere is not representative of the lateral extent of groundwater contamination. Schout et al. (2020) conducted a three-dimensional simulation to study the effect of the flow of groundwater on the migration of methane leaked from oil and gas wells and its retention in aquifers. They indicated that the leakage flux measured at the surface underestimates the subsurface leakage. ...
... In this section, we present numerical simulations of gas migration to quantify the effect of capillary pressure. The gas-water capillary pressure is defined as P c = p g -p w = p e S −1/λ (Brooks & Corey, 1964;Schout et al., 2020) where S = (S w -S wr )/(1-S gr -S wr ), p e is the capillary entry pressure, λ is the pore size distribution parameter, S gr MOHAMMADI ET AL. ...
Article
Full-text available
Steam injection into oil sands reservoirs for bitumen extraction leads to the in situ generation of gases (mainly methane). Many oil sands formations have permeable cap rocks that may allow these gases to migrate to the overlying formations and pose environmental challenges to the groundwater. We present a detailed analysis of the long‐term fate of the in situ generated gases over the course of thermal oil recovery operations. The impacts of capillary barriers on upward migration of the in situ generated gases are quantified. Our findings reveal that the upward migration of gases over a 100‐year period in formations with continuous flow barriers with a permeability of less than ∼1 × 10⁻¹⁸ m² is negligible. However, for formations with discontinuous flow barriers, the migrated gas can potentially leak to the surface depending on the capillary entry pressure of the preferential pathways (e.g., sand layers). In this case, the upward migration is primarily controlled by high permeability pathways. It was found that the migrated gas is effectively trapped by the residual and dissolution trapping mechanisms, and the gas migration to the overlying formations is prevented or significantly reduced by capillary pressure of sand layers. The results indicate that the migrated gas to the shallow groundwater is primarily comprised of methane and is free of CO2 and H2S. These results provide valuable insights into the long‐term fate of these gases in thermal oil recovery operations. They also offer potential opportunities for the development of regulatory frameworks and screening considerations for caprock/seal integrity/risk assessments.
... Under conditions of higher groundwater velocity, a greater volume of water will come in contact with a stationary or slow-moving body of free-phase gas over any given time interval, increasing the amount of gas that dissolves (Klazinga et al., 2019;Schout et al., 2020). At sites with low sustained gas pressure or where the source has been terminated, the free-phase gas flow rate may actually be exceeded by the groundwater velocity. ...
... At sites with low sustained gas pressure or where the source has been terminated, the free-phase gas flow rate may actually be exceeded by the groundwater velocity. In these instances, upward free-phase gas flow paths can be altered or arrested entirely by dissolution (Schout et al., 2020;Van De Ven and Mumford, 2020c). • Multicomponent mass transfer: The rate of free-phase gas dissolution can also be affected by the presence of other gases that are contained in the free-phase gas mixture and/or dissolved in the groundwater (e.g., atmospheric gases like nitrogen or oxygen, and microbial gases including methane and carbon dioxide). ...
Article
Full-text available
Releases of stray natural gas into groundwater from oil and gas exploration, production, or storage can pose a risk to groundwater users and landowners in the form of a fire or explosion hazard. The acute nature of stray gas risk differs from the long-term health risks posed by the ingestion or inhalation of other petroleum hydrocarbons (e.g., benzene). Stray gas also exhibits different fate and transport behaviors in the environment than petroleum hydrocarbon contaminants. These differences include the co-occurrence of dissolved and free gas phases, the potential for rapid and extensive transport of free-phase gas through preferential pathways, and the resulting highly variable and discontinuous spatial distribution of free-phase gas. While there is extensive guidance on response actions for releases of other hydrocarbon contaminants such as benzene, there is relatively little practical information available in regulatory guidance or technical literature regarding appropriate response measures for the investigation and remediation of stray gas impacts. This paper describes key considerations in the physical, chemical, and hydrogeological characteristics of stray gas releases and implications for the improved investigation and mitigation of associated risks.
... The parameters of gas and water relative permeability curves, provided in Table 2, were selected based on the previously fitted data for the most similar injection sites in Alberta (Bachu 2013). We also used the van Genuchten capillary pressure model (1980) (Schout et al., 2020) to describe the capillarity effects in the flow model as given by: ...
Article
In this study, we report comprehensive sensitivity analysis using artificial intelligence (AI) and machine learning techniques to determine the dominant operating and formation statistical parameters affecting the transient behavior of CO2 injection wellbore (i.e., wellbore pressure and temperature). This study focuses on the CO2 injection tests undertaken at the Field Research Station (FRS) injection site in Alberta, Canada. The CO2 wellbore transient behavior is simulated using a fully coupled nonisothermal transient multiphase wellbore-reservoir flow model (Jafari Raad et al., 2021). Radial Basis Function (RBF) Neutral Network and Polynomial Regressions are employed to characterize the dependency of the monitoring functions to the model parameters. The results show that the bottom-hole pressure (BHP) shows maximum sensitivity to the injection rate and minimum sensitivity to the injection pressure. The injection rate and the permeability of the target layer show the highest non-linearity effects. For a constant injection rate, the permeability of the target layer plays the most important controlling effect on the wellbore pressure. The results show that the injection pressure and permeability of the formation layer below the target formation represent the minimum non-linearity and interaction effects. This study provides an improved understanding of operating and reservoir parameters controlling the CO2 wellbore monitoring practices in CO2 injections.
... The capillary pressure is defined as P c = p nw − p w = p e w-nw S −1/λ where S = (S w − S wr )/ (1 − S nwr − S wr ), p e w-nw is the capillary entry pressure of the wet and nonwet phases (i.e., water−oil or oil−gas), λ is the pore size distribution parameter, S nwr is the residual nonwetting phase saturation, and S wr is the residual wetting phase saturation. 29,30 Three rock types are defined, including flow barriers (e.g., shale), high permeability pathways (e.g., sand), and the oil sand formation. The flow barriers (e.g., shale) saturation functions are defined using the typical values of S wr = 0.40, S or = 0.15, S gr = 0.05, p e o-g = p e w-o = 500 kPa (p e g-w = 1000 kPa), and λ = 1. ...
... For example, while the distribution of dissolved CH 4 plume in Scenario 2 is channeled and highly ramified, it is more localized in Scenario 4, accumulating at relatively high concentrations near the injection points with smooth and uniform patterns (Figures S7b and S7d in Supporting Information S1). Such localized and uniform distributions, as frequently reported in previous studies (Roy et al., 2016;Schout et al., 2020), result in an underestimation of dissolved CH 4. ...
Article
Full-text available
We investigate mechanisms that enhance lateral methane (CH4) plume migration in shallow aquifers that exhibit complex and multiscale sedimentary architecture. We show how heterogeneity in capillary pressure characteristics related to the sedimentary architecture causes gaseous CH4 to spread over larger areas by retarding, deviating, or blocking upward buoyancy‐driven CH4 migration. Simplifying or ignoring capillary pressure heterogeneity thus leads to overestimation of leaked CH4 to the atmosphere, and underestimation of mobile gaseous CH4 in aquifer. We show, both qualitatively and quantitatively, that meter‐scale sedimentary stratification contributes more to CH4 plume migration than the millimeter‐ and centimeter‐scale strata comprising them. Results indicate that the extent of gaseous CH4 leakage, and its associated impacts on groundwater quality and global warming, cannot be accurately assessed unless the sedimentary architecture and resulting heterogeneity in capillary pressure are represented.
... The gas phase methane migrates from a leakage source to a shallow aquifer by the upward buoyancy while the dissolved methane migrates more slowly via groundwater flow (Cahill et al., 2018;Taherdangkoo et al., 2020a). Previous modeling studies of water-quality impacts associated with hydrocarbon development have employed multi-phase flow models to simulate methane migration from the deep subsurface to freshwater aquifers along permeable pathways (Cahill et al., 2018;Kissinger et al., 2013;Schout et al., 2020;Taherdangkoo et al., 2020a). ...
Article
The uncontrolled release of methane from natural gas wells may pose risks to shallow groundwater resources. Numerical modeling of methane migration from deep hydrocarbon formations towards shallow systems requires knowledge of phase behavior of the water-methane system, usually calculated by classic thermodynamic approaches. This study presents a Gaussian process regression (GPR) model to estimate water content of methane gas using pressure and temperature as input parameters. Bayesian optimization algorithm was implemented to tune hyper-parameters of the GPR model. The GPR predictions were evaluated with experimental data as well as four thermodynamic models. The results revealed that predictions of the GPR are in good correspondence with experimental data with a MSE value of 3.127 × 10⁻⁷ and R² of 0.981. Furthermore, the analyses showed that the GPR model exhibits an acceptable performance comparing with the well-known thermodynamic models. The GPR predicts the water content of methane over widespread ranges of pressure and temperature with a degree of accuracy needed for typical subsurface engineering applications.
... by fugitive gas remains limited. Numerical simulation of fugitive-gas migration has provided insights into: (a) the role low-permeability layers may play to hinder vertical movement of free-phase gas and enhance lateral accumulation (Klazinga et al., 2019); and (b) the role of gas dissolution on delaying upward gas migration and therefore retaining gas in sandy aquifers (Schout et al., 2020). Fugitive gas dissolution is largely controlled by the free-phase gas configuration and geometry in the subsurface (herein referred to as source architecture) and groundwater flow Van De Ven & Mumford, 2020a, 2020b. ...
Article
Full-text available
Plain Language Summary The impacts of fugitive gas from leaky oil and gas wells remain poorly understood but include greenhouse gas emissions to the atmosphere and degradation of potable groundwater resources. To assess migration patterns, environmental impacts, and fate of fugitive gas, we released natural gas in a shallow groundwater system, representative of much of the Western Canada Sedimentary Basin (WCSB) where extensive petroleum resource development takes place and ∼0.6% of wells have been confirmed as releasing natural gas into the subsurface. Our results indicate that for clay‐rich surficial sediments typical of the WCSB, fugitive gas is likely to remain trapped in the subsurface and to slowly dissolve in groundwater in an irregular manner over a long period. The different constituents that form natural gas dissolved at different rates, changing their molecular ratios; stable‐carbon isotope ratios remained stable. Our findings suggest that where glacial till forms the surficial geology, subsurface fugitive gas released from leaky oil and gas wells remains trapped in the ground, limiting emissions of greenhouse gases to the atmosphere but potentially impacting groundwater.
... We also employed the van Genuchten capillary pressure model (1980) (Schout et al., 2020) to capture the capillarity effects in the reservoir flow model. ...
Article
We investigate the dynamics of a CO2 injection wellbore using a fully-coupled non-isothermal transient multiphase wellbore-reservoir flow model. The focus of this study is on the transient behavior of the wellbore parameters in injection tests undertaken at the CaMI Field Research Station (FRS) in Alberta, Canada. Wellbore transient pressure and temperature and subsequent CO2 phase behavior through the wellbore were analyzed, and the impacts of the reservoir parameters on the wellbore transient flow behavior are characterized. Our results show that the presence of gas-liquid transient flow in the wellbore and the subsequent liquid CO2 build-up in the wellbore significantly increases the bottom hole pressure. The results show that the mean permeability of the target formation layer represents the most important controlling effect on the wellbore transient functions. We used the concepts and understanding developed through the wellbore flow dynamics characterization and the sensitivity analysis to redesign injection tests to control the CO2 wellbore transient flow. We also performed history-matching of the injection tests to examine various combinations of porosity and permeability multipliers in an attempt to match the historical pressure and temperature data obtained during the CO2 injection tests. The outcomes of this study provide an improved understanding of the transient wellbore behavior during CO2 injection over certain pressure-temperature ranges. Results developed here may be used to monitor the injection performance, well and reservoir responses under different injection scenarios.
... Methane dissolved in groundwater tends to be oxidized, controlled by redox conditions, leading to potential degradation of groundwater quality. A proportion of leaked methane can be retained in groundwater due to dissolution and might slowly escape to the atmosphere (Schout et al., 2020). Methane is a potent greenhouse gas with a global warming potential of 34 times higher than CO 2 over a 100-year horizon (Cahill et al., 2017;Rice et al., 2018). ...
Article
The upward migration of methane from natural gas wells associated with fracking operations may lead to contamination of groundwater resources and surface leakage. Numerical simulations of methane transport in the subsurface environment require knowledge of methane solubility in the aqueous phase. This study employs machine learning (ML) algorithms to predict methane solubility in aquatic systems for temperatures ranging from 273.15 to 518.3 K and pressures ranging from 1 to 1570 bar. Four regression algorithms including regression tree (RT), boosted regression tree (BRT), least square support vector machine (LSSVM) and Gaussian process regression (GPR) were utilized for predicting methane solubility in pure water and mixed aquatic systems containing Na+, K+, Ca2+, Mg2+, Cl− and SO4−2. The experimental data collected from the literature were used to implement the models. We used Grid search (GS), Random search (RS) and Bayesian optimization (BO) for tuning hyper-parameters of the ML models. Moreover, the predicted values of methane solubility were compared against Spivey et al. (2004) and Duan and Mao (2006) equations of state. The results show that the BRT-BO model is the most rigorous model for the prediction task. The coefficient of determination (R2) between experimental and predicted values is 0.99 and the mean squared error (MSE) is 1.19 × 10−7. The performance of the BRT-BO model is satisfactory, showing an acceptable agreement with experimental data. The comparison results demonstrated the superior performance of the BRT-BO model for predicting methane solubility in aquatic systems over a span of temperature, pressure and ionic strength that occurs in deep marine environments.
Article
Full-text available
Oil and gas well leakage is of public concern primarily due to the perceived risks of aquifer contamination and greenhouse gas (GHG) emissions. This study examined well leakage data from the British Columbia Oil and Gas Commission (BC OGC) to identify leakage pathways and initially quantify incident rates of leakage and GHG emissions from leaking wells. Three types of leakage are distinguished: “surface casing vent flow” (SCVF), “outside the surface casing leakage” (OSCL), and “cap leakage” (CL). In British Columbia (BC), the majority of reported incidents involve SCVF of gases, which does not pose a risk of aquifer contamination but does contribute to GHG emissions. Reported liquid leakage of brines and hydrocarbons is rarer. OSCL and CL of gas are more serious problems due to the risk of long-term leakage from abandoned wells; some were reported to be leaking gas several decades after they were permanently abandoned. According to the requirements of provincial regulation, 21,525 have been tested for leakage. In total, 2,329 wells in BC have had reported leakage during the lifetime of the well. This represents 10.8% of all wells in the assumed test population. However, it seems likely that wells drilled and/or abandoned before 2010 have unreported leakage. In BC, the total GHG emission from gas SCVF is estimated to reach about 75,000 t/y based on the existing inventory calculation; however, this number is likely higher due to underreporting.
Article
Full-text available
The potential environmental impacts on subsurface water resources induced by unconventional gas production are still under debate. Solving the controversy regarding the potential adverse effects of gas leakages on groundwater resources is therefore crucial. In this work, an interesting real‐world case is presented in order to give further insight into methane multiphase and transport behavior in the shallow subsurface, often disregarded compared to the behavior in the deep subsurface. Multiphase flow and solute transport simulations were performed to assess the vulnerability of an existing shallow unconfined aquifer with respect to a hypothetical methane leakage resulting from a well integrity failure of a former deep geothermal well. The analysis showed that migration of gaseous methane through the aquifer under examination can be extremely fast (of the order of a few minutes), occurring predominantly vertically upwards, close to the well. By contrast, dissolved methane migration is largely affected by the groundwater flow field and occurs over larger time scales (of the order of months/years), covering a greater distance from the well. Overall, the real concern for this site in case of gas leakages is the risk of explosion in the close vicinity of the well. Predicted maximum gaseous fluxes (0.89‐22.60 m3/d) are comparable to those reported for leaking wells, and maximum dissolved methane concentrations may overcome risk mitigation thresholds (7‐10 mg/L) in a few years. Therefore, surface and subsurface monitoring before decommissioning is strongly advised to ensure the safety of the site.
Article
Full-text available
Methane leakage caused by well integrity failure was assessed at 28 abandoned gas wells and 1 oil well in the Netherlands, which have been plugged, cut and buried to below the ground surface (≥3 mbgl). At each location, methane concentrations were thoroughly scanned at the surface. A static chamber setup was used to measure methane flow rates from the surface as well as from 1 m deep holes drilled using a hand auger. An anomalously high flow rate from 1 m depth combined with isotopic confirmation of a thermogenic origin revealed ongoing leakage at 1 of the 29 wells (3.4%), that had gone undetected by surficial measurements. Gas fluxes at the other sites were due to shallow production of biogenic methane. Detailed investigation at the leaking well (MON-02), consisting of 28 flux measurements conducted in a 2 × 2 m grid from holes drilled to 1 and 2 m depth, showed that flux magnitude was spatially heterogeneous and consistently larger at 2 m depth compared to 1 m. Isotopic evidence revealed oxidation accounted for roughly 25% of the decrease in flux towards the surface. The estimated total flux from the well (443 g CH4 hr−1) was calculated by extrapolation of the individual flow rate measurements at 2 m depth and should be considered an indicative value as the validity of the estimate using our approach requires confirmation by modelling and/or experimental studies. Together, our findings show that total methane emissions from leaking gas wells in the Netherlands are likely negligible compared to other sources of anthropogenic methane emissions (e.g. <1% of emissions from the Dutch energy sector). Furthermore, subsurface measurements greatly improve the likelihood of detecting leakage at buried abandoned wells and are therefore essential to accurately assess their greenhouse gas emissions and explosion hazards.
Article
Full-text available
Faulty oil and gas wellbores are a primary pathway of concern for gas migration from the deep subsurface into shallow freshwater aquifers. Leaked gases migrating vertically along wellbores either collect and build a pressure at the wellhead known as sustained casing pressure (SCP) or escape into the atmosphere as surface casing vent flow (SCVF). SCP and SCVF are valuable indicators of integrity loss that provide insight into the potential for groundwater contamination through gas migration. Previous models of SCP and SCVF have focused on offshore wells and have not considered the relationship between SCP/SCVF and gas migration away from onshore wells. We present the first modeling framework for SCP and SCVF that is applicable to onshore oil and gas wells constructed with outermost annuli that are hydraulically connected or “open” to the surrounding formation. Our results show that SCP behavior is unique in onshore wells with open annuli, because steady state SCP is not achieved in an open annulus unless gas escapes the wellbore via gas migration (Darcy Flow), a wellhead leak, or a controlled bleed-off test. We show that appropriately modeling gas leakage along the wellbore and SCP/SCVF helps to constrain methane leakage fluxes from faulty wells and could be integrated with subsurface flow and transport models of stray gas contamination. By relating gas fluxes to SCP/SCVF, our model can provide information used to support regulatory actions informed by SCP/SCVF, which are easy to measure and monitor.
Article
Subsurface leakage of natural gas from petroleum wells can impact freshwater aquifers. Accurate prediction of gas migration in the subsurface will depend on knowledge of permeability, porosity, and flow system conditions. A series of two-dimensional numerical multi-phase flow simulations (CFbio) were conducted to investigate the role of multi-phase parameters (relative permeability and air entry pressure), flow system conditions (intrinsic permeability, anisotropy, and groundwater velocity), and geometric properties (layer thickness and layer lateral continuity) on the flow of gas-phase methane emanating from two variable-rate point sources in an unconfined sandy aquifer. Numerical simulations showed that for a homogeneous, weakly anisotropic aquifer, gas migrates almost exclusively vertically due to buoyancy, before venting to the vadose zone and atmosphere. As vertical migration became restricted through increased anisotropy, inclusion of lower-permeable layers, or increased horizontal groundwater velocity, an increase in the lateral component of gas migration was observed. This led to the formation of a broader lateral migration of the gas-phase plume and establishment of variably distributed vertical preferential flow paths, ultimately resulting in increased gas retention in the aquifer with relatively less methane reaching the vadose zone or atmosphere. The inclusion of a thin layer with moderately lower permeability (1-2 orders of magnitude) and increased air entry pressure was used to depict a fine-grained sand lens within a uniform aquifer. This subtle feature led to the formation of thin gas pools extending up- and down-gradient beneath the lens, allowing methane to travel much farther and faster than by groundwater advection alone, which is consistent with field observations during the experiment. In all scenarios investigated gas-phase methane was shown to migrate predominantly vertically due to buoyancy, until the aquitard permeability was <30% of the aquifer permeability. Our modelling demonstrates that even subtle permeability contrasts, together with capillary pressure changes demarcating grain-scale bedding, will lead to extensive lateral free-phase gas migration, and development of a more extensive and complex zone of impacted aquifer than presupposed.
Article
Petroleum resource development is creating a global legacy of active and inactive onshore energy wells. Unfortunately, a portion of these wells will exhibit gas migration (GM), releasing fugitive gas (FG) into adjacent geologic formations and overlying soils. Once mobilized, FG may traverse the critical zone, impact groundwater, and emit to the atmosphere, contributing to greenhouse-gas emissions. Understanding of GM and FG has increased in recent years but significant gaps persist in knowledge of (1) the incidence and causes of GM, (2) subsurface baseline conditions in regions of development required to delineate GM and FG, and (3) the migration, impacts, and fate of FG. Here we provide an overview of these knowledge gaps as well as the occurrence of GM and FG as currently understood in British Columbia (BC), Canada, a petroleum-producing region hosting significant reserves. To address the identified knowledge gaps within BC, the Energy and Environment Research Initiative (EERI) at the University of British Columbia is implementing several field-focused research projects including: (1) statistical analyses of regulatory data to elucidate the incidence and causes of GM, (2) characterization of regional hydrogeology and shallow subsurface conditions in the Peace Region of the Montney resource play, and (3) investigation of the migration, impacts, and fate of FG in the shallow subsurface through controlled natural-gas release. Together, the EERI investigations will advance understanding of GM and FG, provide scientific data that can inform regulations, and aid development of effective monitoring and detection methodologies for BC and beyond
Article
Extensive development of shale gas has generated some concerns about environmental impacts such as the migration of natural gas into water resources. We studied high gas concentrations in waters at a site near Marcellus Shale gas wells to determine the geological explanations and geochemical implications. The local geology may explain why methane has discharged for 7 years into groundwater, a stream, and the atmosphere. Gas may migrate easily near the gas wells in this location where the Marcellus Shale dips significantly, is shallow (∼1 km), and is more fractured. Methane and ethane concentrations in local water wells increased after gas development compared with predrilling concentrations reported in the region. Noble gas and isotopic evidence are consistent with the upward migration of gas from the Marcellus Formation in a free-gas phase. This upflow results in microbially mediated oxidation near the surface. Iron concentrations also increased following the increase of natural gas concentrations in domestic water wells. After several months, both iron and SO 4²⁻ concentrations dropped. These observations are attributed to iron and SO 4²⁻ reduction associated with newly elevated concentrations of methane. These temporal trends, as well as data from other areas with reported leaks, document a way to distinguish newly migrated methane from preexisting sources of gas. This study thus documents both geologically risky areas and geochemical signatures of iron and SO 4²⁻ that could distinguish newly leaked methane from older methane sources in aquifers.
Article
Global methane (CH4) emissions are becoming increasingly important due to the contribution of CH4 to global warming. Leaking oil and gas wells can lead to subsurface CH4 gas migration (GM), which can cause both aquifer contamination and atmospheric emissions. Despite the need to identify and quantify GM at oil and gas well pads, effective and reliable monitoring techniques are lacking. In this field study, we used CH4 and carbon dioxide (CO2) efflux measurements together with soil gas stable carbon isotopic signatures to identify the occurrence and to characterize the spatio-temporal migration of fugitive gas across 17 selected well pads in Northeastern British Columbia, Canada. At 13 of these sites, operators had previously reported the occurrence of GM; however, subsequent inspections based on visual, olfactory or auditory evidence only identified GM at two of these sites. Using the soil gas efflux method, evidence for GM was found at 15 of the 17 well pads with CH4 and CO2 effluxes ranging from 0.017 to 180 μmol m⁻² s⁻¹(0.024 to 250 g CH4 m⁻² d⁻¹) and 0.50 to 32 μmol m⁻² s⁻¹ (1.9 to 122 g CO2 m⁻² d⁻¹), respectively. Stable carbon isotopic composition was assessed at 10 of the 17 well pads with 9 well pads showing evidence of GM. The isotopic values indicated that CH4 in soil gas was from the same origin as CH4 in the surface casing vent flow gas. There was no correlation between CH4 effluxes and the distance from the well head; an equal portion of elevated effluxes were detected >10 m from the well head as were detected <5 m from the well head. In addition, CH4 effluxes varied temporally with values changing by up to an order of magnitude over 2 h. Although the study was carried out in Northeastern British Columbia, the results are applicable on a global scale, suggesting that inspections mostly based on visual evidence (e.g. bubbling at the well head) are not reliable for the identification of GM and, that infrequent survey measurements at predefined locations close to the well head may overestimate, underestimate or even miss CH4 effluxes. Repetitive and relatively densely spaced gas efflux measurements using a dynamic closed chamber method proved to be a useful tool for detecting GM.