ArticlePDF Available

Maximizing the use of aquifer thermal energy storage systems in urban areas: effects on individual system primary energy use and overall GHG emissions

Authors:
  • Delft University of Technology & KWR Water research institute

Abstract

The primary energy use of ATES systems evaluated for high and low aquifer utilisation levels. • High aquifer utilisation levels reduce energy use of individual systems, as more wells can be placed. • The highest aquifer utilization level considered is 115% and resulted in 82% ATES adoption. • For aquifer utilization <80%, energy use of buildings is not affected by subsurface interactions. • For aquifer utilization >80%, interactions affect gas use +15% and electricity use +/-15%. A R T I C L E I N F O Keywords: Aquifer Thermal Energy Storage (ATES) Subsurface interaction between ATES systems Individual ATES system performance Optimal utilisation of subsurface space A B S T R A C T Low temperature (<25 • C) Aquifer Thermal Energy Storage (ATES) systems have a worldwide potential to provide low-carbon space heating and cooling for buildings by using heat pumps combined with the seasonal subsurface storage and recovery of heated and cooled groundwater. ATES systems increasingly utilize aquifer space, decreasing the overall primary energy use for heating and cooling for an urban area. However, subsurface interaction may negatively affect the energy performance of individual buildings with existing ATES systems. In this study, it is investigated how aquifer utilization levels, obtained by varying well placement policies, affect subsurface interaction between ATES systems and how this in turn affects individual primary energy use. To this end, a building climate installation model is developed and integrated with a MODFLOW-MT3DMS thermal groundwater model. For the spatial distribution and thermal requirements of 26 unique buildings as present in the city centre of Utrecht, the placement of ATES wells is varied using an agent-based modelling approach applying dense and spacious placement restrictions. Within these simulations ATES adoption order and well placement location is randomized. Well placement density is varied for 9 scenarios by changing the distance between wells of the same and the opposite type. The results of this study show that the applied dense well placement policies lead to a 30% increase of ATES adoption and hence overall GHG emission reduction improved with maximum 60% compared to conventional heating and cooling. The primary energy use of individual ATES systems is affected at varying well placement policies by two mechanisms. Firstly, at denser well placement, ATES systems are able to place more wells, which increases the capacity of their ATES system, thereby decreasing their electricity and gas use. Secondly, aquifer utilization increases with denser well placement policies and thus interaction between individual ATES increases. At subsurface utilization up to 80%, individual primary energy use does not change significantly due to subsurface interaction. At aquifer utilization level > 80%, both negative and positive interaction is observed. Negative interaction between wells of the opposite type leads to an increase of gas or electricity use up to 15% compared to spacious well placement. On the other side, buildings may experience a maximum decrease of 15% electricity use at dense well placement due to positive interaction between wells of the same type. Local conditions like building location, plot size, distance to other buildings and 2 heating/cooling demand determine the specific effect per building. The optimal well placement policy result from the aquifer utilisation levels discussed above. Maximum GHG emission reduction while maintaining individual ATES system performance, is achieved with well distances of 0.5-1 times the yearly average thermal radius for wells of the same type (cold-cold and warm-warm). Opposite well types (cold-warm) should be placed apart ~2 times the thermal radius to prevent negative subsurface interaction.
Applied Energy 311 (2022) 118587
Available online 4 February 2022
0306-2619/© 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Maximizing the use of aquifer thermal energy storage systems in urban
areas: effects on individual system primary energy use and overall
GHG emissions
Stijn Beernink
a
,
b
, Martin Bloemendal
a
,
b
,
*
, Rob Kleinlugtenbelt
c
, Niels Hartog
a
a
KWR Water Research Institute, Nieuwegein, the Netherlands
b
Delft University of Technology, the Netherlands
c
IF Technology, the Netherlands
HIGHLIGHTS
The primary energy use of ATES systems evaluated for high and low aquifer utilisation levels.
High aquifer utilisation levels reduce energy use of individual systems, as more wells can be placed.
The highest aquifer utilization level considered is 115% and resulted in 82% ATES adoption.
For aquifer utilization <80%, energy use of buildings is not affected by subsurface interactions.
For aquifer utilization >80%, interactions affect gas use +15% and electricity use +/-15%.
ARTICLE INFO
Keywords:
Aquifer Thermal Energy Storage (ATES)
Subsurface interaction between ATES systems
Individual ATES system performance
Optimal utilisation of subsurface space
ABSTRACT
Low temperature (<25 C) Aquifer Thermal Energy Storage (ATES) systems have a world-wide potential to
provide low-carbon space heating and cooling for buildings by using heat pumps combined with the seasonal
subsurface storage and recovery of heated and cooled groundwater. ATES systems increasingly utilize aquifer
space, decreasing the overall primary energy use for heating and cooling for an urban area. However, subsurface
interaction may negatively affect the energy performance of individual buildings with existing ATES systems. In
this study, it is investigated how aquifer utilization levels, obtained by varying well placement policies, affect
subsurface interaction between ATES systems and how this in turn affects individual primary energy use. To this
end, a building climate installation model is developed and integrated with a MODFLOW-MT3DMS thermal
groundwater model. For the spatial distribution and thermal requirements of 26 unique buildings as present in
the city centre of Utrecht, the placement of ATES wells is varied using an agent-based modelling approach
applying dense and spacious placement restrictions. Within these simulations ATES adoption order and well
placement location is randomized. Well placement density is varied for 9 scenarios by changing the distance
between wells of the same and the opposite type. The results of this study show that the applied dense well
placement policies lead to a 30% increase of ATES adoption and hence overall GHG emission reduction improved
with maximum 60% compared to conventional heating and cooling. The primary energy use of individual ATES
systems is affected at varying well placement policies by two mechanisms. Firstly, at denser well placement,
ATES systems are able to place more wells, which increases the capacity of their ATES system, thereby decreasing
their electricity and gas use. Secondly, aquifer utilization increases with denser well placement policies and thus
interaction between individual ATES increases. At subsurface utilization up to 80%, individual primary energy
use does not change signicantly due to subsurface interaction. At aquifer utilization level >80%, both negative
and positive interaction is observed. Negative interaction between wells of the opposite type leads to an increase
of gas or electricity use up to 15% compared to spacious well placement. On the other side, buildings may
experience a maximum decrease of 15% electricity use at dense well placement due to positive interaction be-
tween wells of the same type. Local conditions like building location, plot size, distance to other buildings and
Abbreviations: ATES, Aquifer Thermal Energy Storage; BCI, Building Climate Installation; COP, Coefcient of Performance; GHG, Greenhouse gas.
* Corresponding author.
E-mail addresses: j.m.bloemendal@tudelft.nl, martin.bloemendal@kwrwater.nl (M. Bloemendal).
Contents lists available at ScienceDirect
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
https://doi.org/10.1016/j.apenergy.2022.118587
Received 14 July 2021; Received in revised form 4 December 2021; Accepted 21 January 2022
Applied Energy 311 (2022) 118587
2
heating/cooling demand determine the specic effect per building. The optimal well placement policy result
from the aquifer utilisation levels discussed above. Maximum GHG emission reduction while maintaining indi-
vidual ATES system performance, is achieved with well distances of 0.51 times the yearly average thermal
radius for wells of the same type (cold-cold and warm-warm). Opposite well types (cold-warm) should be placed
apart ~2 times the thermal radius to prevent negative subsurface interaction.
1. Introduction
Heating and cooling of buildings contributes to about 25% of the
total worldwide energy end-use [1], hence constituting an important
source of greenhouse gas (GHG) emissions. Typical for moderate cli-
mates, the demand for space heating and cooling alternates seasonally.
Using aquifer thermal energy storage (ATES, Fig. 1) systems for har-
vesting and storing cooling potential during winter and heating poten-
tial during summer, results in fewer emissions for heating and cooling.
Low-temperature (LT) ATES systems are therefore increasingly used to
reduce primary energy consumption and associated GHG emissions [2].
Worldwide potential for ATES systems exists across Europe, Asia and
North-America [23]. Life cycle assessment studies indicate that emis-
sions associated to installation activities and component have a
negligible contribution to the overall GHG emission of ATES systems
[45]. Therefore, focussing on operational performance is key in
assessing ATES systems.
ATES systems use tube wells
1
to inject and extract groundwater from
aquifers in the subsurface. Each system consists of at least two tube wells
coupled to a heat exchanger to provide heating/cooling to the building.
Around each ATES well the system takes up space in the subsurface
where the heat/cold is stored. Interaction between wells of the same
type leads to positive thermal performance, while interaction between
opposite well types negatively affects thermal recovery efciency of
these systems e.g. Bloemendal et al. [6,38] and Fig. 2. The increasing
Nomenclature
A
A
Surface area of ATES area under consideration [m
2
]
B Energy balance [-]
c
aq
Volumetric heat capacity of saturated porous medium; 2.8
×10
6
[J/m
3
/K]
c
w
Volumetric heat capacity of water; 4.2 ×10
6
[J/m
3
/K]
COP Coefcient Of Performance: ratio between provided
thermal energy and consumed electricity [-]
D Multiplier for thermal radius, to obtain distance (x)
between wells [-]
Δp Hydraulic resistance or required pressure increase [kg/m/
s
2
]
ε
Specic heating/cooling demand [J/m
2
/y]
E Energy [J]
f correction factor [-]
F
s
Allocated aquifer space fraction for ATES [m
3
/m
3
]
g Gravitational acceleration; 9.81 [m/s
2
]
GHG Greenhouse gas emissions [kg CO
2
]
h counter for number of buildings with conventional system
[-]
i counter for number of buildings with ATES system [-]
j counter for number of well [-]
k number of ATES systems [-]
l number of conventional systems [-]
L Well screen length [m]
L
aq
Aquifer thickness [m]
m number of wells [-]
n Porosity [-]
η
th
Recovery efciency [-]
η
p
Pump efciency; 0.25 [-]
P Thermal or electrical power [J/s]
ρ
Water density; 1,000 [kg/m
3
]
Q Hourly pumping rate of ATES wells [m
3
/hr]
r adoption ratio [-]
R
th
Thermal radius [m]
t time [s hr d]
T Temperature [C]
V
tot
Yearly storage volume of groundwater [m
3
/y]
x Distance between wells [m]
Sub-scripts
aq Aquifer
b Associated to building
boiler Associated to boiler energy use/supply
c Associated to cooling and/or cold well
cap capacity
cp Associated to pumps circulating medium inside the
building climate installation
cond Associated to heat pump condenser
dc Associated to dry-cooler energy use/supply
demand Associated to energy demand
dense Dense well placement setting
e electricity
em emission
est estimated
evap Associated to heat pump evaporator
g gas
h Associated to heating
hp Associated to heatpump
inf Associated to inltration in warm/cold well
loss Associated to heat/temperature losses
max maximum
min minimum
o opposite type of wells
passive Demand associated with the passive cooling mode
pl Associated to partial load
r reference
req required
s same type of wells
spacious Spacious well placement setting
storage Associated to heating/cooling diverted to the ATES
t time
tot total
th thermal
w Associated to warm well
well Associated to the well ow or temperature
y year
1
please note that environmental hazards related to drilling of ATES wells
may differ depending on local geology, and should always be properly
addressed
S. Beernink et al.
Applied Energy 311 (2022) 118587
3
implementation of ATES systems in urban areas has resulted in the
(potential) interaction between multiple systems, leading to questions
on the impact of their individual performance, optimal use of the sub-
surface and overall GHG emissions [78]. When the wells of neigh-
bouring ATES systems are forced apart to prevent any thermal
interaction, this may result in a sub-optimal use of subsurface space for
ATES systems and hence lower overall GHG emission reductions
[6,911]. Under the current legislation in the Netherlands and many
other countries, it is not allowed to affect existing ATES systems [12],
and 3 times the radius of thermal inuence around an ATES well is often
used as a distance between wells [7]. The current placement policy
prevents a spatially dense positioning of ATES wells. Although Bloe-
mendal et al. [6] and Sommer et al. [13] showed that a higher spatial
adoption rate of buildings with ATES, leads to an overall reduction of
GHG emissions, and smaller distances are also applied in practice, it is
still unclear how severe the poorer performance of ATES systems is due
to interaction between ATES wells and how this leads to higher primary
energy use for buildings that are affected by new neighbouring ATES
systems. Furthermore, it is not yet clear how changes in individual ATES
system GHG emission relate to the overall decrease in GHG emissions for
the entire region.
Therefore, the goal of this study is to assess how the different well
placement scenarios lead to subsurface interaction between ATES wells
and how these interactions affect the individual primary energy use and
GHG emissions of each specic building as well as overall GHG emis-
sions. The insights from this study aim to foster practical planning and
design rules suitable to ensure optimal ATES utilisation with improved
GHG emissions reduction while preventing negative impact on the pri-
mary energy use of individual buildings with ATES, which is crucial for
the societal and governmental support of high density placement
policies.
1.1. General approach
A simulation study is carried out to evaluate the primary energy use
of buildings with ATES systems under spacious and dense ATES well
placement policies. To identify optimal well placement rules while
evaluating individual performance of ATES systems, 3 aspects are
important to take into account:
1. The building characteristics (thermal requirements, building plot
space), because they determine the number of wells needed and the
available plot space for well placement.
2. The performance of the ATES wells and their interaction between
neighbouring wells.
3. The primary energy use of the building facilities following from the
ATES well performance.
The calculated GHG emissions for individual ATES users and the
total emissions for all buildings in a case study area are evaluated using
an analysis of the energy use of the building climate installation for
which a detailed building climate installation (BCI) model is developed
and validated. This BCI model simulates the energy ows of the build-
ings during the simulation period and tracks the primary energy use of
the buildings in the study area. The subsurface interactions between the
ATES wells in the model are simulated with an existing groundwater
model code [6,38], which is fully integrated with the developed BCI
model. Placement of ATES wells is done with an agent based model, to
assess various placement policies while taking into account spatially
constrains in the well placement procedure such as e.g. buildings,
infrastructure and surface water [14]. For the spatial distribution and
thermal requirements of 26 unique buildings as present in the city centre
of Utrecht, the ATES system density is varied using an agent-based
approach applying various placement restrictions, within which ATES
adoption order and well placement was randomized. Well placement
density is varied for 9 scenarios by changing the distance between wells
of the same and the opposite type taking into account the thermal
storage volumes that would be required for the adoption of ATES by an
individual building.
2. Methods
2.1. Well placement based on thermal radius
During inltration in an ATES well, a cylindrical shaped volume of
warm/cold groundwater is formed around the well (Fig. 3). This is called
the thermal radius (R
th
) and is calculated based on the total seasonal
storage volume of a well:
Fig. 1. Basic working principle of a low-temperature (LT) ATES-doublet. Right: In winter, buildings are heated with a heat pump (HP) which extracts heat that was
previously stored in the warm well. This creates cooling capacity which is stored in another well in the subsurface. This cooling capacity is used in summer (left) to
cool the building, by storing the excess heat in the warm well. From: Bloemendal et al. [6].
S. Beernink et al.
Applied Energy 311 (2022) 118587
4
Rth =
cwV
caq
π
L
(1)
With c
w
and c
aq
being the volumetric heat capacity (J/K/m
3
) of water
and the saturated aquifer, V the total injected volume (m
3
) and L the
aquifer thickness (m). In a given area, the aquifer thickness and volu-
metric heat capacity are often assumed to be constant. The building
heating/cooling demand and associated inltration/abstraction volume
therefore determines the variation in size of the thermal radius and the
number of used wells. Distance between wells of the opposite type (x
o
)
and same well types (x
s
) is calculated as a function of the thermal radii of
the two wells [6]:
xo=DoRth
xs=DsRth
(2)
Well placement policies follow distance rules which depend on these
thermal radii: wells of the same type (D
s
) can be placed closer to each
other, while wells of the opposite type (D
o
) should be placed further
apart, Fig. 3. When wells do not have the same thermal radii, the dis-
tance between wells is calculated based on the average of the two
thermal radii. A combination of values for D
o
and D
s
is dened as a
placement policy or scenario. For convenience and simplication rea-
sons the abbreviations x
s
=2*R
th
‘D
s
=2and similarly x
o
=3*R
th
‘D
o
=3are used in this study. This approach allows for representative
translation of the results of this study to other locations with a wide
range in aquifer thickness available for ATES.
2.2. Modelling approach
The coupled simulation framework is programmed in python code,
which is a commonly used scripting language in science and engineering
[15]. This framework uses two external simulation codes: MODFLOW/
MT3DMS for groundwater temperature simulations section, section
2.2.3, [1617] and agent based model NetLogo for the well placement,
section 2.2.2 [18]. Additionally, a Building Climate Installation (BCI)
Fig. 2. Schematic top view of how two wells of the same and the opposite type interact at various distances from each other. Distances between wells: top: 200 m,
middle: 50 m. lower: 5 m. The bottom graph shows the left well (warm well in both cases) temperature development during extraction for the 6 cases, the presented
well temperature plots are at the end of extraction from the warm well.
S. Beernink et al.
Applied Energy 311 (2022) 118587
5
model is developed to A) calculate energy demands for each building in
the simulation and to B) subsequently determine primary energy use of
each component of the BCI (section 2.2.1). During simulation, the BCI is
coupled to the groundwater model. The simulation procedure is sche-
matically depicted in Fig. 4 and consists of two main steps:
1. Initialisation: Like is done in practice, a reference climate year is
used to calculate the energy demand and sizes of the ATES system and
BCI components for each building. The NetLogo model is used to
initialize the well locations depending on the predened well placement
policy. If there is no space to place a minimum of 2 wells for the ATES
system, a conventional BCI is assigned to those buildings for that specic
simulation.
2. Simulation: The building-climate model and MOFLOW model are
run subsequently for each hourly time-step of the simulation period. The
building-climate model serves as an input for the MODFLOW model and
vice versa.
2.2.1. Building climate installation
The main functionality of the BCI model is to correctly represent the
interaction between the ATES system and the building climate instal-
lation, as well as conventional climate installation for heating and
cooling. Therefore, two BCI models are created, one for buildings with
an ATES system and one for conventional buildings, Fig. 4. Conventional
heating and cooling systems consists of a gas boiler for heating and a
chiller, which uses electrical energy, for cooling. The ATES system BCI
model is described in the remainder of this section. For simplication
reasons, the ATES system BCI is described for one building with one
doublet (1 warm and 1 cold well), please note that the model infra-
structure can handle any type and number of buildings with any number
of ATES well doublets.
The ATES system BCI model is a conceptual model with components
and operation modes according to Dutch design standards for ATES
systems, Fig. 5 [1920]. The main components of the ATES system BCI
are the heat pump, which directly interacts with the ATES wells, the dry
cooler to exchange heat or cooling capacity with the outside air and a
peak boiler, Fig. 5. Each building has these same components and modes
of operation, due to differences in size and function of the building the
energy demand varies and with that also the size of components.
2.2.2. Basic description of ATES climate installation
An ATES system consists of a cold and a warm well connected to a
heat pump (HP) via a chilled water net and a heat exchanger. Tube wells
produce and inject groundwater. Submersible pumps use electricity to
pump the groundwater from one well to another via a heat exchanger
which allows for heat exchange with the BCI. In summer (cooling de-
mand), cold groundwater is used directly for cooling, Fig. 5-A. In winter
(heating demand) the heat pump condenser provides heating at the
required temperature level, while at the evaporator side of the heat
pump heat is taken from the warm well, cooling down the groundwater,
hence cooling capacity is stored in the cold well, Fig. 5-B. To operate, the
heat pump uses electricity. In summer, the heat pump is mostly not used,
because free cooling is used directly from the cold well. But, when the
cooling capacity of the cold well is not sufcient, the HP can be used to
Fig. 3. Schematic overview of distance rules between wells of the same (D
s
)
and the opposite (D
o
) type based on the thermal radius (R
th
) [32].
Fig. 4. Schematic representation of the simulation framework of the different components of the used modelling system.
S. Beernink et al.
Applied Energy 311 (2022) 118587
6
provide extra cooling capacity to the building. Cooling capacity from the
evaporator is provided to the building, the heat produced at the
condenser side is then stored in the warm well or dumped to the outside
air via the dry cooler. In case of simultaneous heating and cooling de-
mand (not shown in Fig. 5), heat produced at the condenser and cooling
capacity from the evaporator are both supplied to the building. The ratio
between heating and cooling demand will determine whether the ATES
supplies additional heat to the heat pump or direct cooling to the
building. When heating and cooling demand of the building is not in
balance over multiple years a dry cooler
2
is used to either store extra
heat in summer or extra cooling capacity in winter, to meet the next
Fig. 5. Climate installation schematics, depend-
ing on heating/cooling demand conditions
different modes of operations may exists. The
condenser (Con) and evaporator (Eva) together
represent the heat pump. Two out of ve possible
modes are illustrated. A: cooling mode: cooling is
directly delivered from the cold well, the buffer
tank is used to prevent frequent switching of the
submersible well pump. When cooling capacity
from the cold well is insufcient, evaporator of
the heat pump delivers cooling, depending on the
ATES energy balance condition the heat produced
at the condensor is either dispersed into atmo-
sphere via the dry cooler, or stored in warm well.
B: heating mode: warm groundwater is used by
the heat pump evaporator, condenser provides
heat to the distributor demanding heat. When
warm well/ heat pump cannot deliver the
required heat capacity, the peak boiler add extra
heat.
2
Instead of a dry cooler also other techniques can be used, asphalt or solar
collectors, or surface water. Their energy use varies little and depends on local
conditions. Therefore only the dry cooler is evaluated in this study.
S. Beernink et al.
Applied Energy 311 (2022) 118587
7
seasonsenergy demand. Dry cooler operation requires electricity use of
circulation pumps and ATES well pump. The heat pump is installed at
around 3550% of maximum required heating capacity and then pro-
vides ~80% of the total heating demand (Fig. 6). The remaining ~20%
is provided with a boiler which uses natural gas. This is done in practice
to save costs because a monovalent ATES systems requires a signicantly
larger heat pump and double the ATES wells capacity, which is more
expensive.
2.2.3. Temporal discretization and simulation horizon
The model simulates the BCI parameters at an hourly time steps,
which is the minimum needed time step to account for diurnal variations
in energy demand. This therefore also requires hourly temperature data
input.
The total simulation horizon is set to 5 years. Although this is shorter
than the expected life span of ATES systems, it is sufciently long to
distinguish between performance under varying well placement policies
[2122].
2.2.4. Initialisation: Energy demand of buildings
The energy demand of each building is identied using the gross
surface area of the building and the estimated specic yearly heating
and cooling demand from a Dutch database (RVO). The total demand is
then subdivided to hourly rates using both the outside air temperature of
a reference year and the characteristics of the specic building (e.g.
opening hours). The detailed calculation sub-steps are presented in ap-
pendix I. The result of this step is that each building in the simulation has
an expected hourly heating and cooling demand which varies among
each building because building functions, opening hours and size vary.
Based on the maximum hourly values for heating and cooling demand
the required size of the heat pump is determined.
2.2.5. Initialisation: Sizing of components
2.2.5.1. Heat pump. Depending on the type of ATES system BCI:
monovalent or bivalent, the size of the heat pump is determined. In case
of a bivalent system, the HP is designed to provide 35% of the maximum
required heating capacity, Fig. 6. In such conditions the HP can still
provide the vast majority of the heat, because the maximum capacity is
only needed during a limited amount of time during the year, Fig. 6.
When choosing a monovalent system, the heat pump needs to cover
100% of the required capacity. Thus, the HP properties follow from the
maximum required heating capacity found for each building in the
reference year calculation and the setting for the required heat pump
capacity and Coefcient of Performance (COP), Table 1.
2.2.5.2. Number of wells. The required number of wells (m
needed
) for
each building follow from the maximum ow needed to provide the
required cooling capacity and heating capacity (P
h/c_i_max
) from the
wells, please note that the maximum capacity from the warm well de-
pends on size and COP of the heat pump. The temperature difference
between the wells (ΔT) and the volumetric heat capacity of water (c
w
)
determines the required ow capacity from the wells (Q
req
). It is
assumed that the aquifer of choice in the study area can host wells with a
capacity of Q
max
=100 m
3
/hr, the number of wells follows from dividing
the required capacity by the maximum ow rate per well. If the required
number of wells (m
needed
) is actually available depends on the NetLogo
well placement following the determined well placement policy.
mneeded =Qreq
Qmax
=Ph/c i max
ΔTcwQmax
(3)
N.B. In any case no maximum is set to the chiller (when no ATES
wells can be placed) boiler and dry cooler size. For the interaction and
energy use of these components it doesnt matter what capacity/size
they have.
2.2.6. Simulation: BCI operation dynamics
The climate installation model consists of a series of calculations to
determine the functioning and energy use of the different components of
the installation. The building energy uxes described below are calcu-
lated and saved for each hourly time step (t) for each individual building
(k). Each variable is not explained individually in the text, the nomen-
clature gives the description of the used variables, constants and sub-
scripts. The scheme in Fig. 7 is the blueprint for the calculations
explained in the 10 steps described below.
1. Maximum passive cooling
To determine the amount of passive cooling that can be provided by
the system under a given cooling demand, the maximum passive cooling
needs to be calculated. To calculate this, the expected inltration tem-
perature in the warm well due to passive cooling (T
inf_w_est
) is estimated:
Tinf w est =Tc return Thex loss (4)
Fig. 6. Heat pump design: monovalent versus bivalent.
Adapted from: [36]
Table 1
Parameters settings used for the BCI model.
BCI parameter Value Description
T
c_return_min
[C]
13 The minimum return temperature from the building
after passive cooling (cold well >warm well).
T
c_return_max
[C]
16 The maximum return temperature from the building
after passive cooling (cold well >warm well).
T
w_supply
[C] 45 Supply temperature to the building for heating
T
w_return
[C] 35 Return temperature from the building after heating
Q
well_max
[m
3
/
h]
100 Maximum well pumping capacity
T
inf_w_max
[C] 25 Maximum inltration temperature during cooling
(cold well >warm well)
C
1
[-] 0.001604 Heat pump correction parameters, used to calculated
the heat pump correction factor that is used to
calculate the electricity use of the heat pump under
varying partial loads.
C
2
[-] 0.739065
C
3
[-] 9.624631
T
evap_c
[C] 8 Evaporator exit temperature during heating
F
pl_min
[-] 0.25 The minimum partial load of the heat pump
T
loss
[C] 1 The temperature loss in the heat exchanger.
COP
dc
[-] 25 The COP of the dry cooler
COP
C
[-] 3 The COP of the chiller that is used by the BCI to cool
building without an ATES system.
COP
HP
[-] 5 The COP of the HP at full load
COP
circulation
[-]
75 The COP for circulation pumps
COP
well
[-] 40 The COP of the well pump during heating/cooling
COP
boiler
[-] 0.825 Energy efciency of the gas boiler, based on upper
caloric value Dutch gas mix
1
1
For clarity/uniformity reasons in this study COP is used as a symbol instead
of the usual
η
or EER for boilers
S. Beernink et al.
Applied Energy 311 (2022) 118587
8
Where T
c_return
is the return temperature of the cooling grid inside the
building and T
hex_loss
is the temperature loss over the heat exchanger.
T
c_return
varies between a minimum temperature (T
c_return_min
) at low
cooling capacities and a maximum return temperature (T
c_return_max
) at
maximum cooling capacities. T
c_return
is determined by the size of the
cooling demand divided by the maximum cooling demand:):
Tc return =Tc return min +Ec hr(Tc return max Tc return min )
Ec h max
(5)
The maximum amount of passive cooling (P
ATES_c_passive_max
) is sub-
sequently calculated as:
PATES c passive max =QATES w max mcw
Δt(Tinf w est Twell c )(6)
With c
w
being the heat capacity of water, m the number of wells for
the building under consideration and time step size (Δt), in this simu-
lation one hour.
2. Maximum total heating & maximum condenser power
The maximal ATES total heating capacity (PATES w max) is calculated
based on the temperature difference between the warm well and the
inltration temperature in the cold well, the latter being a xed set point
of the heat pump.
PATES w max =QATES w maxmcw
Δt(Twell w Tinf c )(7)
During high cooling demand, passive cooling capacity can be insuf-
cient, in this case the heat pump can be used to provide extra cooling
demand. This results in an amount of condenser heat. The maximum
amount of condenser heat that can be stored in the wells is given by:
PATES cond max = (QATES c maxmcw
Δt(Tinf w max Twell c )) − PATES passive max
(8)
3. Calculation of partial load of the heat pump
The heat pump usage is a percentage of the maximum heat pump
capacity and can therefore never be higher than 100%. The heat pump
can be used for both cooling and heating. The partial load factor (f
hp_pl
)
of the heat pump is determined by the heating/cooling demand, size of
the heat pump and the temperature of the warm and the cold wells.
Considering that all heat demand is supplied by the heat pump and all
excess heat of the condenser is stored in the ATES system at peak
cooling, the partial load of the heat pump is determined by the minimum
and maximum partial load conditions of both the evaporator and
condenser. The maximum partial load factor of the condenser capacity is
calculated by:fpl cond max =(Ph+PATES cond max)
Pcond max (9)
In which P
h
follows from the hourly heating demand E
h_t
, divided by
one hour. The maximum partial load of the heat pump evaporator is
calculated using the cooling demand and the maximum thermal capacity
of the warm well (both are connected to the heat pump evaporator):
fpl evap max =(Pc+PATES w max)
Pevap max
(10)
The minimal partial load of the heat pump evaporator is calculated
by:
fpl evap min =(PcPATES passive max)
Pevap max
(11)
The partial load of the heat pump condenser based on the actual heat
demand is calculated by:
fpl cond demand =Ph
Pcond max
(12)
In practice, there will also be a minimal partial load at which the heat
pump can still operate. In this model, it is assumed that in such a case,
excess heat or cooling capacity is stored temporarily in a buffer tank in
the plant room. This buffer is not modelled, but partial loads between
0% and 100% are allowed instead.
4. Condenser / Evaporator power of the heat pump
Based on the calculated partial load of the heat pump, the actual
Fig. 7. Schematic of BCI component functioning during one hour in winter (heating demand).
S. Beernink et al.
Applied Energy 311 (2022) 118587
9
condenser and evaporator power can be calculated as:
Pcond =fhp plPcond max and Pevap =fhp plPevap max (13)
5. Heat pump operation
From the calculated heat pump operation it is now calculated how
much will be stored in the wells. When the heating demand is smaller
than the condenser heating capacity, the amount of stored heat is
calculated as the condenser heating capacity minus the heating demand.
Similarly, the amount of stored evaporator capacity is the difference
between the cooling demand and the available evaporator capacity.
During heating, the amount of heat that needs to be supplied by the
boiler is:
Pboiler =Ph− (Pcond Pcond storage)(14)
During cooling, the amount of passive cooling that is delivered is
determined by the heat pump is calculated as:
Pc passive=MIN (PATES_passive_max,(Pc-(Pevap - Pevap storage )(15)
6. Energy ows Heat Pump
Similarly the required heating/cooling capacity from the ATES wells
follow from the heat pump operation. The evaporator power therefore
denes the needed capacity from the warm wells (P
ATES_h
=P
evap
) and
the needed capacity from the cold wells is dened by the sum of passive
cooling capacity and the stored warm energy at the condenser. The re-
sidual heating/cooling directly utilized from the heat pump is calculated
as the evaporator capacity minus the heating capacity.
7. Volume ow from ATES wells (injection & extraction)
The amount of volume that is extracted from a type of well and is
injected into the other type of well depends on the total energy ow and
the ΔT between injection and extraction. In heating mode, the injection
temperature is known (set at certain temperature, e.g. 6 C). During
cooling, the injection temperature is not exactly known but is estimated
in step 1. The volume ow is subsequently calculated as for cooling:
Qc=ΔtPc passive
(Tc return Twell c Tloss)cw
(16)
And for heating as:
Qh=ΔtPATES h
(Twell w Tinf c )cw
(17)
8. Inltration temperature of warm and cold well
An important parameter for the interaction with the groundwater
model is the temperature of the inltrated thermal energy. For storage at
the cold well a xed temperature is set, which is similar like in practice;
the temperature of the evaporator of the heat pump, plus a temperature
loss that occurs in the heat exchanger
Tinf c =Tevap c +Tloss (18)
With (T
loss
is 1 C in this study). Heat injection is not as straightfor-
ward. During cooling different modes of operation may occur, which
together result in an injection temperature for the warm well. Low ca-
pacity passive cooling yields relatively low injection temperatures (e.g.
14 C) while high capacity passive cooling results in relatively high
temperatures (e.g. 17 C). In case of peak cooling demands, condenser
heat of the heat pump (functioning as chiller) will increase the injection
temperature even further:
Tinf w =Twell c +ΔtPATES c
Qccw
(19)
9. Energy balance of the wells
When the wells of the ATES system are not in balance the climate
installation model will respond to this and will force the ATES system to
restore the energy balance. This is done with the dry cooler and can only
be done if the outside air conditions are suitable, in winter it is only
possible to store extra cooling capacity and in summer heating capacity.
First the required heating and cooling capacity is delivered, the
remaining capacity of the wells can be used for energy balance correc-
tions. The energy balance (B) is assessed after a period of continuous
operation of 2 years. When the imbalance is >15%, the ATES system will
use its dry cooler capacity to store more heat or cooling capacity.
B=EhEc
Eh+Ec
(20)
B <0: cold well is growing, more heat needs to be charged to the
ATES wells.
B >0: warm well is growing, more cold water needs to be charged
into the wells.
Restoring the energy balance can be done when Q
h/c
< Q
max
. When
this is the case, the available amount of pumping (Q
max
Q
h/c
) will be
used to store extra heat or cooling capacity. This is done with the tem-
perature calculated in the previous step (T
inf_w
/ T
inf_c
).
Pdc =(Qmax Qh)cw⋅ΔT
Δt(21)
10. Determine Energy use HP, Dry cooler, boiler, circulation pumps
From the operation modes identied in the previous calculation step
the energy use of each component for each hourly time step can now be
calculated. For the dry cooler (E
dc
), circulation/well pumps (E
wp
, E
cp
)
and boiler (E
boiler
) this is straightforward:
Eboiler hr =Pboiler
COPboiler
⋅Δt(22)
Edc hr =Pdc
COPdc
⋅Δt(23)
Ewp hr =PATES c +PATES h
COPwell
⋅Δt(24)
Ecp hr =PATES c +PATES h
COPcirculation
⋅Δt(25)
In the calculation of energy use by the circulation pumps are
included the pumps for circulation in evaporator, condenser, boiler, dry
cooler, heat exchangers and circulation circuit to the building for both
heating and cooling mode.
The energy use of the heat pump depends on the quality of the source
and is not linear, at partial load the electricity use is lower than at full
capacity and the COP is therefore higher at that moment. To take this
into account the heat pump COP is corrected with a factor y
hp
(Fig. 8).
The electricity consumption of the heat pump is calculated using the
following equation:
Ehp hr =fhp Php max
COPhp
⋅Δt(26)
The electricity use of the heat pump is corrected with the factor f
hp.
This is calculated following the relationship shown in Fig. 8. This rela-
tionship is determined from heat pump data (Appendix II).
Please note:
- In this study, we use a constant COP for the well pump and the cir-
culation pump. In practice, the well and circulation pumps may have
S. Beernink et al.
Applied Energy 311 (2022) 118587
10
a limited variation in COP. Our results show that the energy use of
these components is relatively small. Hence applying a constant COP
for well and circulation pumps has limited impact on the results.
- Chiller COP, could be considerably lower under partial load condi-
tions, and thus affect results. Because the chiller is mainly used in the
conventional BCI, running it with a constant best case COP will un-
derestimate the electricity use of the chiller. Therefore, this is a
worst-case situation for comparing ATES to conventional. Adding
chiller COP variability would not seriously change the results and
conclusions of this study, it would make the results more in favour of
high density ATES.
- Boilers under partial load will have a lower efciency than boilers
running in full load. In this study, boilers run in partial load. For gas
boiler, lower but constant value for efciency are used due to partial
load.
2.2.7. Placement of wells with NetLogo
Agent based modelling is used to model the systematic outcomes that
emerge from the behaviour of individual actors (in this case building
owners) and is used for socio-environmental simulation of common pool
resource problems like groundwater for ATES utilisation in cities [8,14].
For this study NetLogo is used to initialize ATES wells with their
placement behaviour during initialisation in a pre-dened area. Each
agent is characterized by the size and function of the buildings in the
study area. The number of wells and the size of each ATES system follow
from the energy demand and size of components of the BCI. The
following placement procedure is implemented in NetLogo to represent
the stochastic nature of ATES adoption dynamics:
The study site is divided in 1x1m patches, each patch can be used by
buildings to place a well. Each building in the study site has been
appointed a plot in the NetLogo environment; the buildings can only
build wells on free patches within their plot and adjacent sidewalks.
The placement of new ATES systems is in random order. New ATES
system choose a random location for one of its wells from the
available patches in their plot, respecting the placement rules with
already existing ATES wells. The other well of this system is then
placed in the plot respecting the placement rules. The patches around
each well can no longer be used for placement of other wells,
depending on the size of the well and the placement rules, Fig. 9.
Each ATES system installs its wells using this procedure resulting in
success when there was available space, or in failure when the wells
couldnt nd a location in the grid while respecting the placement
rules.
During placement of new ATES systems, less and less space remains
available to place new wells, as a space around each well is required
around to prevent mutual interaction. If only a limited space was
available for well placement, this can results in a lower amount of
placed well doublets than was originally planned (calculated during
ATES initialisation).
Within the imposed spatial constraints, ATES systems continue to be
added until no more well locations can be found because the area is
lled with ATES wells and separation space between wells.
Due to the stochastic nature of the simulation in agent based
modelling, a single simulation realization is not representative. There-
fore, each scenario comprises of 24 complete model realizations; model
testing has showed that with 24 realizations per scenario the distribution
of the results was sufciently stable to conrm representative behaviour
Fig. 8. The heat pump correction factor for the range of possible partial loads of the heat pump, used to calculate the electricity use of the heat pump.
Fig. 9. Example output of the Agent Based model: placement of 39 well pairs
(78 total wells).
S. Beernink et al.
Applied Energy 311 (2022) 118587
11
suitable for analysis. Using this approach to assess different placement
policies for ATES wells allows to translate simulation results also to
other areas, because the stochastic uncertainties associated with
different city lay-out are taken into account when assessing the
ensemble results, rather than individual model realizations.
2.2.8. Geohydrological modelling with MODFLOW/MT3D-MS
The geohydrological model used for simulation of the subsurface
interactions between ATES systems is MODFLOW/MT3D-MS run
through the SEAWATv4 model. SEAWATv4 combines MODFLOW
(groundwater ow simulations by using a nite-difference method [23]
and MT3DS (Multi-Species modular 3D transport model [17]). Because
of the similarity between the equations for solute and heat transport,
MT3DMS can be used to model transport of heat, by treating heat as a
solute species [2425]. The MODFLOW/MT3DS model is used to
simulate subsurface ow with heat transport, to obtain the extraction
temperatures of the wells to be fed into the BCI model. Subsurface
conditions are considered homogenous. The FloPy model, is used to
build and run these models in python [26].
2.2.8.1. Spatial discretization. For this study the thermal distribution
and heat loss in the horizontal plane is of interest, while vertical loss is
not. Vertical distribution and losses to conning layers is also expected
to be relatively small compared to horizontal losses [2122]. Therefore,
using only 1 layer in the vertical direction with the height of the total
aquifer thickness (26 m) will not affect the interaction effects between
the wells of interest in this study while limiting the model complexity
and required computational resources [6].
To accurately calculate horizontal movement of groundwater and
thermal energy, the city centre of Utrecht was modelled with a 2.5x2.5
m grid in the area of the building locations. A zone of 100 m with the
same cell sizes is constructed around the well-area to minimize nu-
merical dispersion. Around this area the model extents for anoth-
er1000m with a logarithmically increasing cell size up to a maximum of
200 m at the grid boundaries. This results in a model grid with an extent
of 3500x3500m (Fig. 10). The resolution at the area of interest stays well
within the minimum required cell-size of 5x5m identied by Sommer
et al. [22] to adequately model the heat transport around ATES wells. A
description of the subsurface and geohydrological conditions is provided
in Appendix III.
2.2.8.2. Initial and boundary conditions. Model boundaries are set to
have xed heads and temperatures at the boundaries. Ambient tem-
peratures are set at 12 C, which is the assumed average ambient
groundwater temperature of the shallow subsurface. Initial and starting
heads are set to surface level of the model and are constant at model
boundaries.
2.2.8.3. Parameter settings. Aquifer properties are homogeneous
because the effect of heterogeneity on ATES well recovery efciency is
shown to be insignicant [2729] and may disturb the analysis of the
subsurface interactions. Inuence of buoyancy ow due to density dif-
ference that occur due to temperature differences are negligible for LT-
ATES and is therefore not taken into account [21,3031]. Because hy-
draulic conductivity has negligible effects on thermal losses under ho-
mogeneous conditions [21], the horizontal and vertical hydraulic
conductivity is set to a constant value of 30 m/d and 6 m/d respectively.
Both are common values for the aquifers found in the Netherlands
(anisotropy factor of 5). The other thermal and numerical parameters
follow literature values and are given in Table 2.
2.2.9. Combined model validation
The BCI model results are aggregated over the time step length at
which the groundwater model runs, which is 30 days. This results in a
net ow and weighted averaged inltration temperature, which func-
tions as input for the MODFLOW/MT3D-MS model which subsequently
simulates the subsurface changes for the next time step. Thereafter, well
temperatures are used as input for the climate installation model. It is
computationally expensive and time consuming to also run the
groundwater model at hourly time steps. This is not needed because the
groundwater systems reacts slower to dynamics of the climate installa-
tion. Test runs with time steps of the groundwater model of 5 days, a
week and a month were carried out, which gave no signicant difference
in outcomes, indicating that monthly time steps are sufciently small to
assess well performance under varying well placement policies and
capture seasonal storage cycle dynamics. Please note that MT3D-MS
Fig. 10. Left: top view plot of the horizontal discretization. Right: conceptual model of the 3D grid.
Table 2
MODFLOW simulation parameters [2425,27]
Parameter value unit
Porosity 0.3
Longitudinal dispersion 1 m
Transversal dispersion 0.1 m
Horizontal conductivity 30 m/d
Vertical conductivity 6 m/d
Bulk density 1889 kg/m
3
Bulk thermal diffusivity 0.16 m
2
/day
Specic heat capacity solids 750 J/kg C
Specic heat capacity water 4183 j/kg C
Thermal conductivity solid 3 W/m C
Thermal conductivity water 0.61 W/m C
Thermal conductivity of aquifer 2.28 W/m C
Effective molecular diffusion 110
10
m
2
/day
S. Beernink et al.
Applied Energy 311 (2022) 118587
12
automatically takes smaller (internal) time steps if necessary to meet
convergence criteria (Courant number <0.8). The extraction tempera-
tures from the wells are due to the monthly time steps of the ground-
water model also constant during each month in the BCI model.
Because ATES systems can have multiple doublets, the average
extraction temperature (T
extracted
) of the wells is used as input of the BCI
model. The average temperature of the extracted water is calculated for
each ATES system at each groundwater model time step using the
extraction volume (V) of each well (j) belonging to ATES system (i),
following:
Textracted i =
m
j=0
Textracted jVextracted j
m
j=0Vextracted j
(27)
The groundwater model set-up is the same as the work carried out by
[6]. The results of both models showed a good match, which indicates
that the model captures the relevant processes and produces realistic
results. This allows for comparison of different scenarios with the goal to
assess the impact of subsurface interactions on the primary energy use of
individual ATES systems. Due to the lack of historical eld data, it was
not possible calibrate the model to real data. However, the BCI model
results were benchmarked to energy demand monitoring data of two
existing buildings, and showed a good resemblance, also the energy use
of the various components showed a reasonable match, given many
uncertainties and gaps in respective data sets.
2.3. Assessment framework
2.3.1. Primary energy use & GHG emission
2.3.1.1. ATES system. The analysis is carried out for the entire simu-
lation period (t
0
t). The total electricity used by the heat pump (E
hp
),
dry coolers (E
dc
), circulation pumps (E
cp
) and well pumps (E
wp
) is
calculated for each building (i) with an ATES system by:
Ee ATES =Ehp +Edc +Ecp +Ewp =t
t0
(Php +Pdc +Pcp +Pwp)dt (28)
Similarly the gas use of each ATES system is calculated via:
Eg ATES =Eboiler =t
t0
Pboilerdt (29)
The total GHG emission is retrieved by calculating the CO
2
emissions
of the considered (k) ATES systems:
GHGATES =
k
i=1
(Ei
eATESfem e +Ei
gATESfem g )(30)
in which f
em_e
and f
em_g
are the emissions factors for gas and elec-
tricity, and k the number of active ATES systems.
2.3.1.2. Conventional boiler and chiller. Buildings without ATES have a
conventional boiler and compression chiller in this model. The total
electricity used by the chiller (E
c
) and circulation pumps (E
cp
) is calcu-
lated for each building without an ATES system by:
Ee conv =Ec+Ecp =t
t0
(Pc+Pcp)dt (31)
Similarly the gas use of each building without an ATES system is
calculated via:
Eg conv =Eboiler =t
t0
Pboilerdt (32)
The total GHG emission is retrieved by calculating the CO
2
emissions
of the considered buildings:
GHGconv =
l
n=1Echiller,hfem e +Eboiler,hfem g (33)
in which f
em_e
and f
em_g
are the emissions factors for gas and elec-
tricity, and l the number of active conventional systems.
2.3.1.3. Relative change in GHG emissions and primary energy use. The
change in primary energy use and GHG emissions as a result of
increasing density of ATES wells is compared to the amount of energy
use or GHG emissions that would be emitted if the building used con-
ventional heating and cooling system:
ΔGHGi=GHGi
GHGi conv
ΔGHG =1
k
k
i=1
GHGi
GHGi conv
(34)
This relative change is calculated for each building or for the entire
case study area.
2.3.1.4. Emission factors for gas and electricity. For this analysis the
Dutch emission factors for 2019 are used (Table 3). The CO
2
emissions
per unit electricity use are subjected to change due to the mix of power
source contributing to the electricity grid (e.g. coal, solar PV, windmills,
nuclear). It is expected that emission factors for electricity of the Dutch
grid are decreasing up to almost 50% by 2030, due to an increase of
renewable electricity sources, Table 3.
2.3.2. Recovery efciency of ATES wells
The energy recovery efciency (
η
) of the wells of each ATES system
over the simulation period is calculated by dividing the extracted
amount of thermal energy by the inltrated amount of thermal energy:
η
(t0t) = Eout
Ein
=t
t0(Tout Tamb)cwQout dt
t
t0(Tin Tamb)cwQin dt =ΔToutVout
ΔTinVin
(35)
The thermal recovery efciency of all ATES systems in the simulation
(
η
tot
) is the average of the individual efciencies weighted by the indi-
vidual total storage volume of the wells:
η
tot =k
i=1
η
iVi
k
i=1Vi
(36)
The recovery efciencies discussed in this paper are the average ef-
ciencies, averaged for all warm and cold wells of the specic ATES
system.
2.3.3. Adoption ratio
As a result of varying well placement policies, the number of build-
ings with an ATES system and the amount of wells that are being used for
ATES systems varies across simulations. This is shown with the adoption
ratio of the buildings in the case study area (r
ATES
), and the adoption
ratio of the number of wells (r
w
) compared to the needed number of
wells for optimal ATES use:
rATES =k
k+l
rwell =mplaced
mneeded
(37)
Table 3
Parameters to calculate CO
2
emissions from gas and electricity use.
Name Value Description
E
g
35.17 Upper caloric value natural gas in MJ/m
3
f
em_g
1.77 The amount of CO
2
-emission (kg) per m
3
gas.
f
em_e
2019: 0.342030: 0.18 The amount of CO
2
-emission (kg) per KWh
e
[37].
S. Beernink et al.
Applied Energy 311 (2022) 118587
13
2.3.4. Allocated fraction of subsurface space
For various case study areas, the demand for ATES, the settings at
street level and the aquifer thickness will differ. The stress on subsurface
space will therefore also be different. To allow for proper comparison
between simulations of this study, as well as translation of the results
provided in this paper, a spatial parameter is calculated. The allocated
fraction of subsurface space quanties the density of the ATES scenario
and allows comparison between different model realisations. The allo-
cated fraction of subsurface space is dened as the yearly stored volume
of groundwater of all ATES wells, divided by the available aquifer space
in the district under consideration:
Fs=k
i=1Vi
LaqAA
(38)
With A
A
the ATES area [m
2
] and L
aq
the aquifer thickness [m]. The
ATES area (A
A
) is dened as the district area (Fig. 11).
2.4. Data and materials
2.4.1. Case study area
The developed methods are applied to the case study area Utrecht,
Fig. 11. The city centre is a densely populated/built-up area and con-
tains 26 medium to large (multi utility) buildings with a suitable size
and energy prole for ATES. The building characteristics are presented
in appendix III.
Because wells cannot be placed beneath buildings or on the train
track, available space is limited. In Fig. 11 the building plots are shown
for each building. The green area is available to place wells. Around each
building plot a buffer is created to correct for the expected extra area
that can be used by wells when wells are placed close to the edge of
building plots, e.g. in sidewalks. This area is therefore also used to
calculate the total available area for ATES (A
a
), which is 854,452 m
2
for
the case study area Utrecht.
2.4.2. Climate data and reference year
The used climate data to drive the simulations is obtained from the
Royal Netherlands Meteorological Institute (KNMI) in the Netherlands.
Daily temperature data of the years 20102015 of measuring station ‘de
Biltare used, Appendix III. For the initialisation of the climate model
(Section 2.2.1) a reference year is used, Appendix III. The reference year
is an articial dataset consisting of separate temperature data fragments
and designed as the ‘normaltemperature year in the Netherlands and is
used in practice to identify the yearly average energy demand and peak
load of buildings.
2.5. Well placement policy scenarios
In this study, the ‘lowest densityplacement policy is based on cur-
rent Dutch design standard (D
o
=3, D
s
=2) and higher density place-
ment policies follow from Martin Bloemendal et al. [6]. Therefore, D
o
is
varied as 3, 2.5 or 2, and D
s
as 2, 1 or 0.5. Previous research showed that
decreasing D
o
below 2 leads to a large negative effect on the recovery
efciency, opposite well distance is therefore not tested lower than D
o
=
2 [6,32]. This results in a total of 9 different well placement policy
combinations.
3. Results
3.1. ATES adoption and recovery efciency
The simulation results illustrate that with applying denser ATES well
placement policies (Fig. 12) the number of ATES systems, the individual
storage volume and the overall storage volume increases. Consequently,
wells of the same type form larger clustered warm and cold volumes,
that increasingly adjoin with other clustered storage volumes at their
borders.
The densest well placement policy (D
s
=0.5 & D
o
=2) results in more
than double the amount of ATES wells compared to the lowest density
placement policy (D
s
=2, D
o
=3). The stored volume has a strong
positive correlation with the allocated fraction of subsurface space (F
s
),
Table 4. The allocated fraction of subsurface space for the most dense
well placement policy is more than twice as large than the most spacious
well placement policy.
Decreasing the same well distance (D
s
), compared to decreasing the
opposite well distance (D
o
), has a different effect on well placement and
the ability of specic buildings to place wells (Table 4). A smaller D
o
has
a strong effect on the number of ATES systems that are able to place a
minimum of one well doublet, while decreasing D
s
has a stronger effect
on the number of wells that can be placed for each ATES system. This
difference is visible between the D
s
=2, D
o
=2 and the D
s
=1, D
o
=3
policy in Table 4. The latter having more total wells on the one hand
while these wells are coupled to a smaller amount of buildings. It is
therefore benecial to apply small separation distances between same
type of wells in areas with a relatively large share of large buildings in
need of many doublets.
Fig. 13 shows the ATES adoption ratio (r
ATES
), the well adoption ratio
(r
wells
) and the recovery efciency (
η
th
) for the buildings for 3 well
placement policies. This shows that the difference between the two
dense well placement policy is relatively small. The highest recovery
efciencies are observed for the most spacious well placement policy
and lowest recovery efciencies are observed for the dense well place-
ment policy. The recovery efciencies of the buildings, as well as to what
extent they are affected by the placement policies varies. These varia-
tions are linked to the differences in building size and the associated
total storage volume, the proximity of other ATES building plots and the
size of the plot, representing how easy it is for buildings to identify
suitable well locations under various placement policies. Most buildings
do not have a balanced heating and cooling demand (on average
50100% difference, see table in Appendix III). This has an effect on the
heating and cooling loads from the ATES wells, and impacts the energy
performance of the cold and warm wells. As most buildings have a
higher heating demand, also recovery efciencies of warm wells are
usually higher. However, the results show that this effect is limited
(<15%, Appendix IV), which is partly because of the automatic energy
balance control of the ATES systems (explained in bullet point 9 in
subsection 2.2.1). Buildings 2, 8, 10 and 15 have the lowest recovery
Fig. 11. Case study area Utrecht. The different colours depict the building plot
areas and the building itself. The total available building plot space is
854,452 m
2
.
S. Beernink et al.
Applied Energy 311 (2022) 118587
14
efciencies (in the dense and spacious policy), they are relatively small
buildings and located in the centre of the case study area, making them
prone to interaction with other buildings, Fig. 11. The large spread in
recovery efciencies across all the model realisations for these buildings
presented in Appendix IV illustrates this. While other buildings are
limited affected by changes in well placement policies, because they
have a large plot to place wells (building 7), are very large (14) or at the
boundary of case study area (24). Again, this is illustrated by the limited
spread in the efciencies across all model realisation which are pre-
sented in Appendix IV.
Please note that some buildings are missing and some buildings have
no result for the low density policy. This is a result of the stochastic well
placement approach in the agent based model. Each model realisation is
different, also when applying the same well placement policy. This
random well placement led to<5 model realisations in which these
buildings have an active ATES system for the low density placement
policy. 5 realisations or less is considered to be insufcient for repre-
sentative analysis of the energy use. It concerns buildings 4, 19, 21, 23 in
all policies, and buildings 2, 22 and 25 for the most spacious policies.
These buildings are therefore excluded from the analysis. The reason for
limited ATES adoption for these building is the small available building
plot area, which makes it impossible to place an ATES doublet while
maintaining the well placement constraints. Even when these buildings
are among the rst adaptors, the opposite well distance constraint often
appeared to be the limiting factor to place an ATES doublet, for both the
high as well as low density scenarios. In practice such constraints would
be relaxed or well locations would searched somewhat further from the
building.
Buildings 0, 1, 11 and 14 have a large building plot and adopt ATES
in each model realisation, which means they always have at least one
doublet. This does not mean that these buildings are always able to place
the designed maximum amount of wells required to fully meet the
building demand for all model realisations. For building 0 and 1 this
only occurs in about 25% of the realisations.
3.2. Overall GHG emissions
With an increase of the subsurface space use, the average recovery
efciency of all ATES wells decreases, Fig. 14A. This indicates that
negative interaction between wells of the opposite type have a stronger
effect on the recovery efciency than the positive interaction between
wells of the same type at dense well placement policies (high allocated
subsurface fraction, F
s
). The decline in recovery efciency shows a non-
linear trend with increasing density, indicating that further increase of
ATES density may result in negative recovery efciencies, i.e. short-
circuit ow occurs. Despite the lower average recovery efciency the
Fig. 12. The effect of different well placement policies based on the relative thermal radius distances between same (D
s
) and opposite (D
o
) thermal well types on
temperature distribution in the storage aquifer in the end of the winter of the 5th year of the simulations for 1 random model realisation for 3 different placement
policies. From A-C density well placement density increases with decreasing D
s
from 2 to 0.5 times and D
o
from 3 to 2 times the thermal radius.
Table 4
Relative amount of buildings with ATES systems (compared to total number of buildings in case study area), placed wells (compared to total well demand), and
allocated fraction of subsurface space F
s
for the range of policies.
S. Beernink et al.
Applied Energy 311 (2022) 118587
15
total GHG emissions diminish with increasing ATES density, indicating
that the increase of buildings with ATES contributes to overall GHG
emission reduction, similar to observations by Martin Bloemendal et al.
[6].
The ATES adoption ratio is strongly linked to the allocated fraction of
subsurface space use, Fig. 14B. The load duration curve (Fig. 6) shows
that the rst added ATES wells to a building have the largest impact on
emission reduction. The simulation results conrm this expected
Fig. 13. A: Average ATES adoption ratio (r
ATES
), B: Average well adoption ratio (r
wells
) and recovery efciency (
η
th
) of the individual buildings for the most dense (D
s
=0.5, D
o
=2), dense (D
s
=1, D
o
=2) and spacious (D
s
=2, D
o
=3) well placement policies.
Fig. 14. A: The average recovery efciency for all wells in different well placement policies and the relative GHG emissions compared to if all buildings used
conventional heating and cooling system. B: the ATES adoption ratio, i.e. the percentage of buildings with an ATES system at difference well placement policies. And
the average well adoption ratio of all buildings, i.e. the percentage of realised wells at different well placement policies compared to designed amount of wells. Please
note that at high density, space in between warm and cold wells can be used twice during a year, at the end of winter it is occupied by the cold well, while at the end
of summer it is occupied by the warm well. Hence the allocated surface fraction can be >1.
S. Beernink et al.
Applied Energy 311 (2022) 118587
16
behaviour: the largest difference in energy savings are initiated at high
adoption ratios, i.e. when more buildings have the opportunity to use
ATES. This is also illustrated by the contribution of GHG emission by gas
and electricity use for the ATES and the conventional situation, pre-
sented in Fig. 15. With an increase in number of ATES systems, less
conventional systems are operational in the case study area. This results
in a mild increase of GHG due to gas and electricity use by ATES
buildings (Fig. 15A). While GHG emissions due to electricity and gas use
by conventional systems strongly decreases (Fig. 15B). Most buildings
do not have their desired amount of wells in the different placement
policies, the fact that they have wells already has a large effect on overall
emissions.
Fig. 15 also shows that the ratio of GHG emission by electricity/gas
for the area as a whole shifts with spacious to dense well placement
policy from approximately 50/50 to 70/30. Because electricity use is
dominant at dense well placement, future lower emission factors of
electricity will even further decrease total CO
2
emissions. In 2030, the
GHG emission rate is expected to decrease almost 50% compared to
2019 (Table 3). At high well placement densities this will then result in
an even further reduction from 40% emissions in 2019 to only 25%
emission in 2030, compared to current emissions by conventional
systems.
3.3. The energy performance of individual buildings
The decrease of the overall GHG emissions at high ATES well density
is caused by the increase of buildings with ATES and is not related to the
energy performance change of individual ATES systems due to a denser
well placement policy. To what extent the increased interaction between
wells (resulting in the lower recovery efciencies) at denser well
placement policy, affects the primary energy use of individual ATES
systems is assessed by evaluating the energy use of each individual
building with ATES. The change in gas use, electricity use and GHG
emission of each building is calculated relative to the situation with
conventional heating and cooling, for a dense and spacious well place-
ment policy (Fig. 16A-B-C). This shows that the gas and electricity use of
each building does not change considerably under dense and spacious
well placement, Fig. 16 A and B. But because the gas use of a building
with ATES is much lower compared to conventional heating and cooling
(only 1020%), the difference in gas use of each building between the
dense and spacious well placement policy is still quite large for some
buildings, Fig. 16D.
From these results the following observations stand out:
- In general, the average gas use change varies when comparing the
dense and spacious well placement policies, while electricity use is
less dependent on well placement policy. This is explained by the
increase of well doublets per system, which increases the heat pump
capacity, and results in less gas is use by the peak boiler, rather than
mutual interactions. But there are cases where gas use increases,
which is a result of too few warm wells being placed.
- The differences in electricity use is consistently a little bit lower in
the dense placement policy. This decrease is caused because more
wells can be placed and more cooling is obtained from the ATES
rather than the cooling machine. The differences are much smaller
than for gas use changes and result in a <5% change, predominantly
lower in the dense setting, but in one building there is also a larger
electricity use.
- The differences in GHG emissions are dominated by the differences
in gas use.
The results from Fig. 16 are not conclusive on the effect of mutual
interaction on the performance of the ATES system, because the addition
of wells mask the impact of interactions on the performance. To
discriminate between the implications of the increase of wells and the
effect of subsurface interaction, the gas and electricity use of all sce-
narios for which a building obtains their maximum desired number of
wells is presented in Fig. 17. Here, the primary energy use variation is
solely inuenced by subsurface interaction. The energy use change at
increasing aquifer utilization levels is shown compared to the energy use
at the lowest aquifer utilization level for which each building has ob-
tained their maximum number of wells. Negative impact at aquifer
utilization levels up to 80% is limited to max 5% for both gas and
electricity use. At higher levels, some building observe a gas use or
electricity use increase of maximally 15%. However, this is only limited
to a few buildings. The maximum and minimum change increases with
higher well placement density. The average change of all buildings
however is limited to <5% at the highest densities for both electricity
Fig. 15. Share of total GHG emissions due to buildings with an ATES systems (left) and building with conventional buildings (right). With higher allocated fraction of
subsurface space (F
s
) the share of emitted GHG emissions increases for ATES and decreases for conventional buildings.
S. Beernink et al.
Applied Energy 311 (2022) 118587
17
and gas use (solid line, Fig. 17).
Fig. 17 shows that the effect of subsurface interaction at increasing
aquifer utilization varies strongly for each individual building, and is
thus affected by local conditions. Therefore, 5 buildings that had their
maximum required wells in most of the model realisations across all
placement policies are showed in Fig. 18 to assess what conditions cause
this variability. Please note that:
Buildings 6 and 7 and their neighbouring buildings have much more
space available to identify well locations, see Fig. 11.
Buildings 6, 12 and 17 have the same function and specic energy
use and therefore respond similarly, especially in the electricity use.
They have a strongly imbalanced heating and cooling demand,
heating demand is larger, see Appendix III.
Building 7 is much larger than the other buildings and has a larger
cooling demand than buildings 6, 12 and 17, see Appendix III.
Building 24 has a balanced heating and cooling demand (For details,
see Appendix III)
Fig. 18 shows the gas use, electricity use and GHG emissions as a
function of the allocated subsurface fraction for these 5 buildings. For all
5 buildings it is concluded that both gas and electricity use is not much
affected by subsurface interaction with allocated subsurface fraction up
to 0.8. At higher density placement policies, the smaller buildings may
suffer from a considerable amount of gas use increase, max 9%. For all 5
buildings, the electricity use varies max. 5% for F
s
<0.8, and for the
smaller buildings, the electricity use does not vary >2% up to 0.9 allo-
cated aquifer fraction. Building 7 has, a much lower electricity use
compared to the other buildings (around 1.1 vs. 1.35), due to a relatively
large cooling demand. But as a result it is also more sensitive for mutual
interactions at higher allocated subsurface fractions. Building 24 with
the balanced heating and cooling demand doesnt seem to suffer from
the subsurface interaction. This may be caused by the fact that 1) it is on
the boundary of the study site and 2) due to itshigh cooling demand,
warm wells are always fully charged and therefore gas use is not much
affected. Please note that for the buildings with an imbalanced heating
and cooling demand the correction for the imbalance usually lags behind
by a year, which is also the case in practice (see step 9 in the operation
dynamics description in section 2.2.1). Building 7 (in the middle of the
study area), which also has a higher cooling demand compared to the
other 3 buildings with a stronger imbalance towards larger heating de-
mand, also shows less sensitivity for gas use due to interactions.
These results show that ATES adoption can be accommodated in
areas up to an aquifer allocated subsurface fraction of 80%, without
Fig. 16. Gas (A) and Electricity (B) use and GHG emissions (C) for the buildings with an ATES system relative to conventional systems, compared between spacious
(D
s
=2, D
o
=3) and dense (D
s
=1,D
o
=2) well placement scenario. The error bars represents 2 standard deviations. D: the average relative change in electricity and
gas use and GHG emissions for each building between high and low density placement scenarios. Please note that the buildings 2, 22 and 25 do not have ATES wells
in the spacious well placement scenario.
S. Beernink et al.
Applied Energy 311 (2022) 118587
18
considerable negative impact on the performance of individual ATES
systems. These results are in correspondence with the earlier work of
Martin Bloemendal et al. [6] and allow for higher aquifer utilisation
than current practice for ATES planning does (i.e. 2540%, Martin
Bloemendal et al. [6]; Martin Bloemendal et al. [7]. Additionally, the
results indicate that buildings with a balanced heating and cooling de-
mand suffer less from mutual interactions.
4. Discussion
4.1. Permit vs. real storage volume: which thermal radius to use for well
placement?
In the simulations in this study, the thermal radius used for well
placement is representative for the actual average thermal radius that
develops for the ATES systems in the simulation. The maximal yearly
thermal radius varies slightly between consecutive years, due to varia-
tions in outside air temperature, like they would also in practice.
However, the thermal radius used for planning in practice, is based on
the maximal storage volume requested in the permit. This is the allowed
storage volume in case of an extremely warm summer, or extremely cold
winter. As a design value, the permitted capacity is often chosen to be
1.5 times the expected yearly capacity, while in practice monitoring
data shows that the permitted total volume is >2 times larger than the
actual average storage volume [3334]. Hence, applying the permit
volume to calculate the maximum thermal radii for well placement will
overestimate the thermal radii, and thus result in underutilisation of
aquifer space. Therefore, it is recommended to use the thermal radius
resulting from the average expected storage volume for the well place-
ment (or adapt the placement rules to account for the safety factor in
permit capacity). This can be done by requesting both the estimated
average and maximum storage volume for the new ATES system, when
applying for a permit. The permit application of course has to be
checked to prevent that the maximum yearly storage volume is sub-
mitted as the average storage volume, as this is preferential for the ATES
owner.
Fig. 17. Gas use, electricity use and GHG emissions change for individual ATES systems in different model realisations due to increasing levels of aquifer utilization
(F
s
) compared to the lowest aquifer utilization model run that resulted in the maximum number of desired wells for each individual ATES system / building. The solid
line indicates the average change of the ATES systems at given F
s
.
Fig. 18. Relative change in gas (A) and electricity (B) use compared to conventional systems for 5 buildings which have the maximum number of ATES wells for all
model realisations. The differences in energy use are therefore only caused by mutual interactions.
S. Beernink et al.
Applied Energy 311 (2022) 118587
19
In the Netherlands, local authorities also apply the use it, or lose it
rule. When permit capacity is not used for several years, the allowed
storage volume in their permit is reduced. This creates space for new
ATES initiatives. However, wells of the existing ATES systems will not be
relocated when this happens. Therefore, it is key for local authorities to
prevent the use of overestimated storage volumes in the ATES permit, to
enable correct well placement and optimize utilization of subsurface
space.
Given the large uncertainties in expected energy use of new buildings
and tendency for individual ATES owners to request for larger storage
capacity than they expect to use, it is logical to tend towards more dense
well placement distances: <0.5 for D
s
and <2 for D
o
.
Another aspect that may be of inuence on the yearly storage vol-
ume, is the temperature of the warm and cold well. If for future BCI
systems, larger temperature differences between the warm and cold well
can be obtained, the storage volume reduces, which may be an efcient
way to improve accommodation of ATES system. However, with smaller
storage volumes energy losses increase [21]. And possibly, also the
sensitivity of lower recovery efciency on total primary energy use
could increase.
4.2. Difference between gross and net storage volume
The total (gross) storage volume is used for the thermal radius used
for well placement. But, during fall and spring ATES systems change
ow direction regularly to supply both heating and cooling [35]. These
pumping operations contribute to the yearly (gross) storage volume, but
the (net) storage volume in one well is smaller and results in a smaller
thermal radius, than expected. Also during these seasons, thermal radius
is not at its largest, hence, these short-term injections and extraction do
not result in mutual interaction. This explains why the monthly time
steps for the groundwater model works well with the hourly BCI time
steps for the simulations in this study (see section 2.2.3). In this study,
the average difference between gross and net storage volume was about
15%, while in practice this may go up to almost 30% [35]. In the sim-
ulations this effect was taken into account as a result of the longer
groundwater model time step. Therefore, the distances identied are
still applicable for moderate climatic zones such as in the Netherlands.
In climatic zones with a stronger pronounced heating and cooling sea-
son, resulting in a small difference between gross and net storage vol-
ume, ATES systems performance may react stronger to dense well
placement setting and well distances may need to be somewhat larger.
At the same time, in climatic zones that have a heating and cooling
season that is even less pronounced compared to the Dutch situation, the
well distances may be reduced. This may also be the case for the
Netherlands, as the simulated difference between gross and net storage
volume is lower than in practice, according to the observations in M
Bloemendal et al. [35].
4.3. Placement rules in practice
4.3.1. Placement rules applied to individual ATES building plots
In the simulations, same placement rules apply to all wells, also for
the wells of one individual ATES system. However, in practice ATES
designers are more exible with placement of the wells of one individual
system. The results of the most dense placement policies are represen-
tative for these conditions: in practice wells of the same type are often
placed at distance 0.5 R
th
, or even shorter. The opposite well types
distance is usually not smaller than 2 R
th
. This means that in practice
wells of one individual ATES system are currently in practice already
placed using the dense placement policy. When there are few buildings
with many doublets, a spacious well placement policy would in practice
look like the dense well placement policy as it is simulated in this study.
However, in the study site there are many buildings (26) of which 9 and
10 have respectively 1 or 2 doublets, 7 buildings require more doublets.
Hence, the results are representative for areas with many buildings with
a limited number of required doublets (1 or 2). For areas with fewer, but
larger buildings with many doublet, the spacious well placement may
differ less from the dense placement policy than is identied in this
study.
For the dense well placement scenario, it is reasonable to expect that
this artefact results in smaller thermal inuence outside the buildings
plot for a given amount of wells, or similar thermal inuence for more
wells placed. Which in both cases is positive for the overall performance.
The results of this study are worst-casein that sense.
Furthermore, it would be benecial from a practical point of view, if
there is no need to make a distinction between how wells are placed
relative to each other, within an ATES system, or between different
ATES systems. Therefore, it is logical to stay on the small side for the
distance multipliers: <0.5 for D
s
and ~ 2 for D
o
.
4.3.2. Wells in public space
When it is not possible to nd an optimal well location, or a well
location at all, ATES wells are often placed in the public space around
the associated building. In the Netherlands, the extent to which mu-
nicipalities facilitate this, differs. In this study, the building plot
boundaries are set in such a way that adjacent sidewalks, squares and
parks are included, to allow for well locations in public space. Some of
the buildings in the study area benet greatly from this, while others
dont. It is not analysed to what extend well locations indeed end up in
public space. However, the large differences in ATES adoption ratio and
the resulting GHG emissions reductions at denser well placement pol-
icies indicate a great benet of allowing ATES system designers to place
wells in public space when individual building plot space is limiting.
4.4. Comparison to other ATES design and sustainable energy systems
As is common, the use of a conventional boiler and compression
chiller was used in this study as a reference for conventional building
climate control. But they cannot be evaluated as a sustainable alterna-
tive for ATES. This study would therefore benet from a follow-up
analysis to also asses alternative sustainable heating/cooling systems.
Furthermore, all ATES systems in this study have the exact same BCI
design and control, following the basis of the Dutch standard [19]. In
practice however, this standard is always tuned to specic building re-
quirements, which may (considerably) affect behaviour and perfor-
mance of the BCI system. For example, the ATES system may be designed
to exclude the use of gas (monovalent ATES system), meaning that the
BCI is equipped with a heat pump at the required maximum heating
capacity and no peak boiler. Dense and spacious placement policies are
used to simulate the use of monovalent ATES systems in this case study
area, see Fig. 6. As a result the well placement dynamics change: due to
the increased peak demand, the total capacity (m
3
/h) of the ATES wells
needs to be increased. Consequently, the number of wells required for a
given ATES system approximately doubles. While at the same time, the
required storage volume increases with only about 20%. As a result,
more wells with relatively small thermal radii need to placed. The total
combined thermal radius of the wells only increases with 20%. Table 5
conrms this change in well placement behaviour, as many more wells
are placed under the same placement policies. Due to the smaller ther-
mal radii it is easier to nd well locations, also at dense placement
policies. As a results the well adoption ratio is not much affected, despite
the increased number of wells. For example: a required 250 000 m
3
storage volume in a 25 m thick aquifer for a bivalent ATES system has a
thermal radius of 70, requiring <35 m distance to same type of wells.
When two of these wells are cluster the total thermal radius is 100 m. If
this is now a monovalent system, each well needs 250 000 ×1.2 / 2 =
150 000 m
3
, resulting in an individual thermal radius of 55 m and a
combined thermal radius (now of 4 wells) of 105 m.
Fig. 19 shows the recovery efciency, GHG emissions and energy use
for the bivalent and monovalent scenarios. Please note that there is still
gas use in the monovalent results because ATES systems which do not
S. Beernink et al.
Applied Energy 311 (2022) 118587
20
receive the desired number of wells, or no wells at all, still need a(n
additional) source of heat to meet their heating demand. This is again
done with gas red boilers. For the same placement rules, more wells
can be placed, which results in a higher allocated fraction of subsurface
space, F
s
. Compared to the bivalent ATES concept, the average recovery
efciency is affected less for similar subsurface space use, which is due
to the smaller radii of the stored thermal volumes which makes their
placement easier and thus results in less inuence by other wells. The
recovery efciency has the same trend towards higher allocated sub-
surface fractions (same slope). The GHG emissions show a different
trend, because monovalent systems in principle do not use gas, the GHG
emissions mainly depend on electricity, and are as a result less sensitive.
The GHG emission reductions are smaller compared to bivalent systems,
this is because there is little gas use in the monovalent cases and due to
the large emission factors in the current electricity grid mix in the
Netherlands, which is likely to decrease. Whereas the electricity use does
Table 5
Simulation results meta data: increase of active ATES systems, number of wells and aquifer utilization for monovalent ATES compared to the bivalent simulation results
for dense (D
s
=1,D
o
=2) and spacious (D
s
=2,D
o
=3) well placement policies.
Active ATES systems [-] Number of wells [-] Well adoption ratio [-] Allocated aquifer fraction (F
s
) [-]
dense spacious dense spacious dense spacious dense spacious
Bivalent 20.8 13.0 93.9 41.1 0.70 0.31 1.02 0.52
Monovalent 23.5 20.3 194.5 82.8 0.72 0.30 1.54 0.93
Fig. 19. Monovalent and bivalent simulation results. A: recovery efciency and GHG emissions reduction. B: electricity and gas use.
Fig. 20. Relative change in gas (A) and electricity (B) use compared to conventional systems for building 6 with the maximum number of ATES wells for all
monovalent and bivalent model realisations. The differences in energy use are therefore only caused by mutual interactions.
S. Beernink et al.
Applied Energy 311 (2022) 118587
21
not change much at higher densities in the bivalent case, in the mono-
valent scenarios this increases signicant, hence the less sensitive trend
in the GHG emissions.
Therefore, also for the monovalent ATES design decreasing the well
distance also helps for better utilisation of subsurface space and overall
reduction of GHG emissions. Again also the gas and electricity use of
individual buildings are analysed for the model realisations where they
have the maximum desired wells installed, unfortunately only one of the
previous 5 now has model realisations with all required wells in the low
density scenario. Fig. 20 shows the gas and electricity use as a function
of the allocated subsurface fraction for building 6. Despite that only one
building has its maximum desired wells it is still possible to derive how
interaction due to increased subsurface space use affects the perfor-
mance of individual monovalent ATES systems. Please note that the gas
use is included because the model automatically provides heat with gas
when the ATES system is not possible to provide the required heating.
But in the situation that all desired well are installed, the building would
not have a gas boiler. Therefore, the gas use in Fig. 20 represents to what
extent the ATES system is not able to provide the required heating due to
subsurface interactions. For the low density scenario there is no gas use
so the ATES system can always provide heating demand. At the dense
well placement policy interactions result in poorer performance,
resulting in some cases a couple of percent of the heating demand not
able to be fullled, however, most of the cases<1%. Also the electricity
use is affected by a couple of percent. Because it is only one building it is
hard to draw strong conclusions. Nevertheless, the results are similar
with the bivalent systems for building 6. There is a small impact due to
mutual interaction, only where bivalent systems sometimes also have a
positive effect of mutual interaction on total energy use, the monovalent
systems only have higher energy use due to mutual interactions.
5. Conclusions
The results of this study show that the change in primary energy use
for individual ATES systems under denser well placement policies re-
sults in two main aspects:
1) Buildings can place more wells, therefore this leads to increased heat
pump electricity use and less boiler gas use. Even when buildings
only have a part of their required ATES well capacity installed, their
GHG emissions reduce considerably. The largest effect on GHG
emissions reductions is observed for the rst added ATES well
doublet (Fig. 16), due to the shape of the load duration curve (Fig. 6).
2) ATES wells have lower recovery efciencies. This, however, has
negligible effect on the combined electricity consumption of the heat
pump and well pumps and peak boiler gas use at allocated subsurface
fraction <0.8. At higher density there may be a considerable nega-
tive effect on peak boiler gas use consumption for relative small
buildings as well as a >5% change in electricity use (increase or
decrease) for buildings with a large cooling demand.
Furthermore, the overall results for this study conrm earlier work
that denser placement policies lead to an overall decrease of GHG
emissions due to larger adoption rates of ATES technology.
5.1. Implications for practice:
By increasing well placement density, rst adopters who have
installed the designed/desired amount of wells may suffer some per-
centage gas use increase due to lower ATES performance caused by
increased subsurface interaction. However, the rst ATES well doublet
effectresults in much larger gas use reduction for late adopters than it
results in a gas use increase for rst adopters. Therefore, it is important
to either limit the number of wells for rst adopters, or better: ensure in
their permit conditions, that they allow future neighbours to affect their
well efciency in a controlled way. For rst adopters it is more benecial
to have their desired number of wells, which might perform a little less
once the area is lled up with ATES systems, rather than having a limited
amount of wells. For late adopters a more dense well placement setting is
always benecial. Even if this enables them to place only one ATES
doublet of many required, a considerable emission reductions can be
achieved.
The optimal well placement policy is: 0.5 for D
s
and 2 for D
o
. And, if
possible, it is more preferable to use smaller multipliers than larger,
especially for the same well types. The storage volume used to determine
the thermal radius for well placement should be the average yearly
storage volume, not the maximum. Local authorities should carefully
verify permit storage volumes used for planning of ATES wells, to pre-
vent unnecessary claims on subsurface space due to ATES permits with
too large maximum and/or average yearly storage volume.
5.2. Further research:
The monovalent ATES concept leads to more exible well placement
conditions, therefore a higher allocated fraction of subsurface space and
increase of GHG emissions reduction. The performance of individual
monovalent ATES systems is a bit more sensitive for mutual interaction,
than bivalent ATES systems. But the effect of mutual interaction on
different ATES systems design (e.g. monovalent) should be further
investigated as the results in this study are not conclusive.
The results indicate that ATES systems with a balanced heating and
cooling demand are less sensitive for mutual interactions. Which implies
the need for stronger anticipation on imbalances, e.g. using predictive
control to restore the energy balance in time, rather than lag behind a
year like is common in practice. To what extend this would indeed
reduce sensitivity, and which methods can be used, is still to be
determined.
CRediT authorship contribution statement
Stijn Beernink: Methodology, Software, Validation, Investigation,
Visualization, Writing original draft, Writing review & editing.
Martin Bloemendal: Conceptualization, Methodology, Software,
Investigation, Supervision, Funding acquisition, Writing original draft,
Writing review & editing. Rob Kleinlugtenbelt: Methodology, Soft-
ware, Writing review & editing. Niels Hartog: Writing review &
editing.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Acknowledgements
This research was funded by TKI (Topconsortia voor Kennis en
Innovatie), BodemenergieNL (kennisplatform bodemenergie), KIBO
(Kennis en Innovatieprogramma Bodem en Ondergrond), Province and
Municipality of Utrecht and NWO (grant number 408-13-030). The
authors thank dr. Marc Jaxa-Rozen for his help with setting up the agent
based model and the TKI project partners for their input on the simu-
lation results: Johan Valstar of Deltares, Marian van Asten and Marlous
van der Meer of the Province of Utrecht, Harry Boerma of the Munici-
pality of Utrecht and Jan Frank Mars of the Ministry of Infrastructure
and Environment. Furthermore we thank the editor and 4 anonymous
reviewers for their valuable suggestions for improvements.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.apenergy.2022.118587.
S. Beernink et al.
Applied Energy 311 (2022) 118587
22
References
[1] IEA, IRENA, & REN21. (2020). Renewable Energy Policies in a Time of Transition -
Heating and Cooling (ISBN 978-92-9260-289-5). Retrieved from https://www.iea.
org/.
[2] Fleuchaus P, Godschalk B, Stober I, Blum P. Worldwide application of aquifer
thermal energy storage A review. Renew Sustain Energy Rev 2018;94:86176.
https://doi.org/10.1016/j.rser.2018.06.057.
[3] Bloemendal M, Olsthoorn T, van de Ven F. Combining climatic and geo-
hydrological preconditions as a method to determine world potential for aquifer
thermal energy storage. Sci Total Environ 2015;538:62133. https://doi.org/
10.1016/j.scitotenv.2015.07.084.
[4] Stemmle R, Blum P, Schüppler S, Fleuchaus P, Limoges M, Bayer P, et al.
Environmental impacts of aquifer thermal energy storage (ATES). Renew Sustain
Energy Rev 2021;151:111560. https://doi.org/10.1016/j.rser.2021.111560.
[5] Tomasetta C, van Ree CCDF, Grifoen J. Life Cycle Analysis of Underground
Thermal Energy Storage 2015;Vol. 2:12137.
[6] Bloemendal M, Jaxa-Rozen M, Olsthoorn T. Methods for planning of ATES systems.
Appl Energy 2018;216:53457. https://doi.org/10.1016/j.apenergy.2018.02.068.
[7] Bloemendal M, Olsthoorn T, Boons F. How to achieve optimal and sustainable use
of the subsurface for Aquifer Thermal Energy Storage. Energy policy 2014;66:
10414. https://doi.org/10.1016/j.enpol.2013.11.034.
[8] Hardin G. tragedy of the commons. Science 1968;162:12438. https://doi.org/
10.1126/science.162.3859.1243.
[9] Bloemendal M, Jaxa-Rozen M, Rostampour V. Use it or lose it, adaptive ATES
planning. Paper presented at the 12th IEA Heat Pump conference Rotterdam. 2017.
[10] Jaxa-Rozen M, Kwakkel JH, Bloemendal M. The Adoption and Diffusion of
Common-Pool Resource-Dependent Technologies: The Case of Aquifer Thermal
Energy Storage Systems. Portland: Paper presented at the PICMET; 2015.
[11] Rostampour V, Jaxa-Rozen M, Bloemendal M, Kwakkel J, Keviczky T. Aquifer
Thermal Energy Storage (ATES) smart grids: Large-scale seasonal energy storage as
a distributed energy management solution. Appl Energy 2019;242:62439. https://
doi.org/10.1016/j.apenergy.2019.03.110.
[12] H¨
ahnlein S, Bayer P, Ferguson G, Blum P. Sustainability and policy for the thermal
use of shallow geothermal energy. Energy policy 2013;59:91425. https://doi.org/
10.1016/j.enpol.2013.04.040.
[13] Sommer W, Valstar J, Leusbrock I, Grotenhuis T, Rijnaarts H. Optimization and
spatial pattern of large-scale aquifer thermal energy storage. Appl Energy 2015;
137:32237. https://doi.org/10.1016/j.apenergy.2014.10.019.
[14] Jaxa-Rozen M, Kwakkel JH, Bloemendal M. A coupled simulation architecture for
agent-based/geohydrological modelling with NetLogo and MODFLOW. Environ
Modell Software 2019;115:1937. https://doi.org/10.1016/j.envsoft.2019.01.020.
[15] Perez F, Granger BE, Hunter JD. Python: An Ecosystem for Scientic Computing.
Comput Sci Eng 2011;13(2):1321. https://doi.org/10.1109/MCSE.2010.119.
[16] Harbaugh, A. W., Banta, E. R., Hill, M. C., & McDonald, M. G. (2000). Modow-
2000, the U.S. Geological survey modular ground-water modeluser guide to
modularization concepts and the ground-water ow process Virginia: US
Geological Survey. Retrieved from https://www.usgs.gov/mission-areas/water-
resources/science/modow-and-related-programs.
[17] Zheng, C., & Wang, P. (1999). MT3DMS v5.3. A modular three-dimensional
multispecies transport model for simulation of advection, dispersion and chemical
reactions of contaminants in groundwater systems. Documentation and Users
Guide. Retrieved from https://hdl.handle.net/11681/4734.
[18] Wilensky, U. (1999). NetLogo. Evanston, IL.: Center for Connected Learning and
Computer-Based Modeling, Northwestern University, . Retrieved from http://ccl.
northwestern.edu/netlogo/.
[19] ISSO. ISSO-publicatie 39 Energiecentrale met warmte en koude opslag (WKO).
Rotterdam: ISSO; 2017.
[20] KvINL. (2017). BRL 6000-21/00 - Ontwerpen en installeren van energiecentrales
van bodemenergiesystemen en het beheren van bodemenergiesystemen. In.
Zoetermeer: KvINL.
[21] Bloemendal M, Hartog N. Analysis of the impact of storage conditions on the
thermal recovery efciency of low-temperature ATES systems. Geothermics 2018;
17:30619. https://doi.org/10.1016/j.geothermics.2017.10.009.
[22] Sommer WT, Doornenbal PJ, Drijver BC, van Gaans PFM, Leusbrock I,
Grotenhuis JTC, et al. Thermal performance and heat transport in aquifer thermal
energy storage. Hydrogeol J 2014;22(1):26379. https://doi.org/10.1007/s10040-
013-1066-0.
[23] McDonald MG, Harbaugh AW. A modular three-dimensional nite-difference ground-
water ow model (06A1). Retrieved from 1988. http://pubs.er.usgs.gov/p
ublication/twri06A1.
[24] Hecht-Mendez J, Molina-Giraldo N, Blum P, Bayer P. Evaluating MT3DMS for Heat
Transport Simulation of Closed Geothermal Systems. Ground Water 2010;48(5):
74156. https://doi.org/10.1111/j.1745-6584.2010.00678.x.
[25] Langevin, C. D., Thorne, D. T., Dausman, A. M., Sukop, M. C., & Guo, W. (2008).
SEAWAT Version 4: A computer program for simulation of multi-Species Solute
and heat transport. Retrieved from Reston, Virginia: https://www.usgs.gov/
mission-areas/water-resources/science/modow-and-related-programs.
[26] Bakker M, Post V, Langevin CD, Hughes JD, White JT, Starn JJ, et al. Scripting
MODFLOW Model Development Using Python and FloPy. Groundwater 2016;54
(5):7339. https://doi.org/10.1111/gwat.12413.
[27] Calj´
e R. Future use of aquifer thermal energy storage inbelow the historic centre of
Amsterdam. (Msc). Delft: Delft University of Technology; 2010.
[28] Possemiers M, Huysmans M, Batelaan O. Application of multiple-point geostatistics
to simulate the effect of small-scale aquifer heterogeneity on the efciency of
aquifer thermal energy storage. Hydrogeol J 2015;23(5):97181. https://doi.org/
10.1007/s10040-015-1244-3.
[29] Xynogalou M. Determination of optimal separation well distance for Single
Borehole ATES systems in the Netherlands, implementing an axisymmetric
numerical model. (MSc). Delft: Delft University of Technology; 2015.
[30] Anderson MP. Heat as a ground water tracer. Ground Water 2005;43(6):95168.
https://doi.org/10.1111/j.1745-6584.2005.00052.x.
[31] Doughty C, Hellstrom G, Tsang CF. A dimensionless approach to the Thermal
behaviour of an Aquifer Thermal Energy Storage System. Water Resour Res 1982;
18(3):57187. https://doi.org/10.1029/WR018i003p00571.
[32] Duijff, R. (2019). Interaction between multiple ATES systems. Retrieved from Delft:
https://repository.tudelft.nl/islandora/object/uuid%3Acf4bcee3-b775-4676-
b683-0e38fabb70ee.
[33] Beernink, S., Hartog, N., Bloemendal, M., & Meer, M. v. d. (2019). ATES systems
performance in practice: analysis of operational data from ATES systems in the
province of Utrecht, The Netherlands. Paper presented at the European Geothermal
Congress, The Hague, Netherlands.
[34] Willemsen, N. (2016). Rapportage bodemenergiesystemen in Nederland. Retrieved
from Arnhem.
[35] Bloemendal, M., Beernink, S., Bel, N., Hockin, A. E., & Schout, G. (2020). Transitie
open bodemenergiesysteem Koppert-Cress naar verhoogde opslagtemperatuur. Evaluatie
van energiebesparingen en grondwatereffecten. Retrieved from Nieuwegein:
https://library.kwrwater.nl/publication/61755396/.
[36] Buday, T. (2014). Reduction of environmental impacts of heat pump usage with
special regard on systems with borehole heat exchangers. Landscape &
Environment, 8(2). doi:https://ojs.lib.unideb.hu/landsenv/article/view/2310.
[37] Schoots, K., Hekkenberg, M., & Hammingh, P. (2017). Nationale Energieverkenning
2017. Retrieved from Petten.
[38] Duijff Rogier, Bloemendal Martin, Bakker Mark. Interaction Effects Between
Aquifer Thermal Energy Storage Systems. Groundwater 2021. https://doi.org/
10.1111/gwat.13163. In this issue.
S. Beernink et al.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Aquifer thermal energy storage (ATES) is an energy efficient technique to provide heating and cooling to buildings by storage of warm and cold water in aquifers. In regions with large demand for ATES, ATES adoption has lead to congestion problems in aquifers. The recovery of thermal energy stored in aquifers can be increased by reducing the distance between wells of the same temperature while safeguarding individual system performance. Although this approach is implemented in practice, the understanding of how this affects both the recovery efficiency and the neeeded pumping energy is lacking. In this research the effect of well placement on the performance of individual systems is quantified, and guidelines for planning and design are developed. Results show an increase in thermal recovery efficiency of individual systems when the thermal zones of wells of the same temperature are combined, which is explained by reduced surface area of the thermal zone over which losses occur. The highest increase of the thermal recovery efficiency is found for systems with a small storage volume and long well screens. The relative increase of the thermal recovery efficiency is 12% for average-sized systems with a storage volume of 250,000 m3 /year, and 25% for small systems (50,000 m3 /year). The optimal distance between wells of the same temperature is 0.5 times the thermal radius, following the trade-off between an increase of the thermal recovery efficiency and the increase in pumping energy. The distance between wells of opposite temperature must be larger than 3 times the thermal radius to avoid negative interaction.
Conference Paper
Full-text available
Energy consumption for space heating and cooling of buildings can be decreased by 40-80% by use of Aquifer Thermal Energy Storage (ATES). ATES is a proven technique, however, it is not known how efficient currently operating systems are recovering stored energy from the subsurface and how this can be determined with available data. Recent research suggests that storage conditions have a large influence on the recovery (e.g. shape and size of stored volume). In addition, literature and previous research show that other aspects of ATES system are often unfavorable (e.g. subsurface energy imbalance, small ∆T). Therefore, the main goal of this research is to define a framework to determine overall performance of ATES systems by analysis of monitoring data from operational ATES systems. The province of Utrecht was selected for this. Monthly operational data of 57 ATES systems (40% of the ATES systems in the province) was provided by the authorities and pre-processed accordingly. Results showed that recovery efficiency is positively correlated to system size (stored volume) and that ambient groundwater temperature is site-specific and should be determined for each ATES system individually. Ambient groundwater temperature can vary more than 4 °C and are spatially correlated. Next to this, a large part of the analyzed systems are not equally storing heat and cold in the subsurface. More than 80% of the studied ATES systems have an subsurface heat imbalance larger than 10%. Altogether, results indicate that a big part of the ATES systems in the province of Utrecht can substantially improve their ATES system (management) to increase long-term energy savings. This research provides an useful assessment framework to determine if an ATES system is performing correctly and what aspect of the specific ATES system needs most improvement.
Article
Full-text available
A benchmark of existing ATES plans is carried out and the ATES planning method is improved. • The effectiveness of design parameters and assessment criteria for ATES planning are identified and quantified. • Aquifer usage density thresholds beyond which planning is necessary are determined. A R T I C L E I N F O Keywords: Aquifer Thermal Energy Storage (ATES) ATES planning Optimal use of subsurface space A B S T R A C T Aquifer Thermal Energy Storage (ATES) systems contribute to reducing fossil energy consumption by providing sustainable space heating and cooling for buildings by seasonal storage of heat. ATES is important for the energy transition in many urban areas in North America, Europe and Asia. Despite the modest current ATES adoption level of about 0.2% of all buildings in the Netherlands, ATES subsurface space use has already grown to congestion levels in many Dutch urban areas. This problem is to a large extent caused by the current planning and permitting approach, which uses too spacious safety margins between wells and a 2D rather than 3D perspective. The current methods for permitting and planning of ATES do not lead to optimal use of available subsurface space, and, therefore, prevent realization of the expected contribution of the reduction of greenhouse gas (GHG) emissions by ATES. Optimal use of subsurface space in dense urban settings can be achieved with a coordinated approach towards the planning and operation of ATES systems, so-called ATES planning. This research identifies and elaborates crucial practical steps to achieve optimal use of subsurface space that are currently missing in the planning method. Analysis from existing ATES plans and exploratory modeling, coupling agent-based and groundwater models were used to demonstrate that minimizing GHG emissions requires progressively stricter regulation with intensifying demand for ATES. The simulations also quantified both the thresholds beyond which such stricter rules are needed as well as the effectiveness of different planning strategies, which can now effectively be used for ATES planning in practice. The results provide scientific insight in how technical choices in ATES well design, location and operation affect optimal use of subsurface space, and what trade-offs exist between the energy efficiency of individual systems and the combined reduction of the GHG emissions from a plan area. The presented ATES planning method following from the obtained insights now fosters practical planning and design rules suitable to ensure optimal and sustainable use of subsurface space-that is, maximizing GHG emission reductions by accommodating as many ATES systems as possible in the available aquifer, while maintaining a high efficiency for the individual ATES systems.
Technical Report
Full-text available
The SEAWAT program is a coupled version of MODFLOW and MT3DMS designed to simulate three dimensional, variable-density, saturated ground-water flow. Flexible equations were added to the program to allow fluid density to be calculated as a function of one or more MT3DMS species. Fluid density may also be calculated as a function of fluid pressure. The effect of fluid viscosity variations on ground-water flow was included as an option. Fluid viscosity can be calculated as a function of one or more MT3DMS species, and the program includes additional functions for representing the dependence on temperature. Although MT3DMS and SEAWAT are not explicitly designed to simulate heat transport, temperature can be simulated as one of the species by entering appropriate transport coefficients. For example, the process of heat conduction is mathematically analogous to Fickian diffusion. Heat conduction can be represented in SEAWAT by assigning a thermal diffusivity for the temperature species (instead of a molecular diffusion coefficient for a solute species). Heat exchange with the solid matrix can be treated in a similar manner by using the mathematically equivalent process of solute sorption. By combining flexible equations for fluid density and viscosity with multi-species transport, SEAWAT Version 4 represents variable-density ground-water flow coupled with multi-species solute and heat transport. SEAWAT Version 4 is based on MODFLOW-2000 and MT3DMS and retains all of the functionality of SEAWAT-2000.SEAWAT Version 4 also supports new simulation options for coupling flow and transport, and for representing constant-head boundaries. In previous versions of SEAWAT, the flow equation was solved for every transport timestep, regardless of whether or not there was a large change in fluid density. A new option was implemented in SEAWAT Version 4 that allows users to control how often the flow field is updated. New options were also implemented for representing constant-head boundaries with the Time-Variant Constant-Head (CHD) Package. These options allow for increased flexibility when using CHD flow boundaries with the zero-dispersive flux solute boundaries implemented by MT3DMS at constant-head cells. This report contains revised input instructions for the MT3DMS Dispersion (DSP) Package, Variable-Density Flow (VDF) Package, Viscosity (VSC) Package, and CHD Package. The report concludes with seven cases of an example problem designed to highlight many of the new features.
Article
Full-text available
Aquifer thermal energy storage (ATES) is a technology with worldwide potential to provide sustainable space heating and cooling using groundwater stored at different temperatures. The thermal recovery efficiency is one of the main parameters that determines the overall energy savings of ATES systems and is affected by storage specifics and site-specific hydrogeological conditions. Although beneficial for the optimization of ATES design, thus far a systematic analysis of how different principal factors affect thermal recovery efficiency is lacking. Therefore, analytical approaches were developed, extended and tested numerically to evaluate how the loss of stored thermal energy by conduction, dispersion and displacement by ambient groundwater flow affect thermal recovery efficiency under different storage conditions. The practical framework provided in this study is valid for the wide range of practical conditions as derived from 331 low-temperature (< 25 °C) ATES systems in practice. Results show that thermal energy losses from the stored volume by conduction across the boundaries of the stored volume dominate those by dispersion for all practical storage conditions evaluated. In addition to conduction, the displacement of stored thermal volumes by ambient groundwater flow is also an important process controlling the thermal recovery efficiencies of ATES systems. An analytical expression was derived to describe the thermal recovery efficiency as a function of the ratio of the thermal radius of the stored volume over ambient groundwater flow velocity (Rth/u). For the heat losses by conduction, simulation results showed that the thermal recovery efficiency decreases linearly with increasing surface area over volume ratios for the stored volume (A/ V), as was confirmed by the derivation of A/V-ratios for previous ATES studies. In the presence of ambient groundwater flow, the simulations showed that for Rth/u <1 year, displacement losses dominated conduction losses. Finally, for the optimization of overall thermal recovery efficiency as affected by these two main processes, the optimal design value for the ratio of well screen length over thermal radius (L/Rth) was shown to decrease with increasing ambient flow velocities while the sensitivity for this value increased. While in the absence of ambient flow a relatively broad optimum exists around an L/Rth-ratio of 0.5–3, at 40 m/year of ambient groundwater flow the optimal L/Rth-value ranges from 0.25 to 0.75. With the insights from this study, the consideration of storage volumes, the selection of suitable aquifer sections and well screen lengths can be supported in the optimization of ATES systems world-wide.
Article
Aquifer Thermal Energy Storage (ATES) is an open-loop geothermal system allowing long-term storage of thermal energy in groundwater. It is a promising technology for environmentally friendly energy generation that can reduce greenhouse gas (GHG) emissions. In the literature, there are few studies on the greenhouse gas emissions caused by ATES systems over their entire life cycle. Thus, this study presents a novel life cycle assessment (LCA) regression model that can be used for a wide range of ATES configurations due to its parametric structure. This model is a fast alternative to conventional time-consuming LCAs. Combined with a Monte Carlo simulation, it enables the analysis of the environmental impacts of a large variety of hypothetical ATES systems and therefore the evaluation of the technology as a whole. Compared to conventional heating systems based on heating oil and natural gas, the median value of the Monte Carlo simulation results in GHG savings of up to 74%. In comparison to cooling techniques using today's electricity mix, ATES can save up to about 59% of GHG emissions, while also being economically competitive. When considering a projected electricity mix for the year 2050, the GHG emission savings resulting from a second LCA regression model are as high as 97%. The findings of our sensitivity analysis show which ATES design parameters should be optimized when planning new systems. In particular, the most important design parameters operating time cooling and coefficient of performance (COP) of the heat pump should be carefully considered.
Article
Aquifer Thermal Energy Storage (ATES) is a building technology used to seasonally store thermal energy in the subsurface, which can reduce the energy use of larger buildings by more than half. The spatial layout of ATES systems is a key aspect for the technology, as thermal interactions between neighboring systems can degrade system performance. In light of this issue, current planning policies for ATES aim to avoid thermal interactions; however, under such policies, some urban areas already lack space for the further development of ATES, limiting achievable energy savings. We show how information exchange between ATES systems can support the dynamic management of thermal interactions, so that a significantly denser layout can be applied to increase energy savings in a given area without affecting system performance. To illustrate this approach, we simulate a distributed control framework across a range of scenarios for spatial planning and ATES operation in the city center of Utrecht, in The Netherlands. The results indicate that the dynamic management of thermal interactions can improve specific greenhouse gas savings by up to 40% per unit of allocated subsurface volume, for an equivalent level of ATES economic performance. However, taking advantage of this approach will require revised spatial planning policies to allow a denser development of ATES in urban areas.
Article
The modelling of social-ecological systems can provide useful insights into the interaction of social and environmental processes. However, quantitative social-ecological models should acknowledge the complexity and uncertainty of both underlying subsystems. For example, the agent-based models which are increasingly popular for groundwater studies can be made more realistic by incorporating geohydrological processes. Conversely, groundwater models can benefit from an agent-based depiction of the decision-making and feedbacks which drive groundwater exploitation. From this perspective, this work introduces a Python-based software architecture which couples the NetLogo agent-based platform with the MODFLOW/SEAWAT geohydrological modelling environment. This approach enables users to design agent-based models in NetLogo's user-friendly platform, while benefiting from the full capabilities of MODFLOW/SEAWAT. This workflow is illustrated for a simplified application of Aquifer Thermal Energy Storage (ATES).
Article
To meet the global climate change mitigation targets, more attention has to be paid to the decarbonization of the heating and cooling sector. Aquifer Thermal Energy Storage (ATES) is considered to bridge the gap between periods of highest energy demand and highest energy supply. The objective of this study therefore is to review the global application status of ATES underpinned by operational statistics from existing projects. ATES is particularly suited to provide heating and cooling for large-scale applications such as public and commercial buildings, district heating, or industrial purposes. Compared to conventional technologies, ATES systems achieve energy savings between 40% and 70% and CO2 savings of up to several thousand tons per year. Capital costs decline with increasing installed capacity, averaging 0.2 Mio. € for small systems and 2 Mio. € for large applications. The typical payback time is 2–10 years. Worldwide, there are currently more than 2800 ATES systems in operation, abstracting more than 2.5 TWh of heating and cooling per year. 99% are low-temperature systems (LT-ATES) with storage temperatures of < 25 °C. 85% of all systems are located in the Netherlands, and a further 10% are found in Sweden, Denmark, and Belgium. However, there is an increasing interest in ATES technology in several countries such as Great Britain, Germany, Japan, Turkey, and China. The great discrepancy in global ATES development is attributed to several market barriers that impede market penetration. Such barriers are of socio-economic and legislative nature.