Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Applied Energy 355 (2024) 122261

Contents lists available at ScienceDirect

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

Cost-efficient cooling of buildings by means of geothermal borefields with


active and passive cooling
Matthijs Coninx a ,∗, Jarne De Nies a , Louis Hermans a,c , Wouter Peere a , Wim Boydens b,d ,
Lieve Helsen a,c
a
University of Leuven (KU Leuven), Celestijnenlaan 300, box 2421, Leuven, Belgium
b boydens engineering (part of Sweco), Noordkustlaan 10, Dilbeek, Belgium
c EnergyVille, Thor Park, Waterschei, Belgium
d Department of Architecture and Urban Planning, University of Ghent, Ghent, Belgium

ARTICLE INFO ABSTRACT

Keywords: Due to the increasing demand for space heating and cooling and emphasis on sustainable solutions, ground
Ground source heat pump source heat exchangers become more attractive. However, ground source heat pumps using direct cooling face
Sizing adoption challenges due to their high upfront investment costs. These costs can be reduced by limiting the
Economic optimisation
required borefield size through combining active and passive cooling regimes with ground source heat pumps.
Borefield design
The absence of a widely-available sizing methodology for combined cooling regimes results in suboptimal
Open source
Geothermal cooling
borefield design. Hence, the full economic potential of combining these systems is not utilised. Therefore, this
study introduces an innovative approach for cost-efficient borefield sizing that integrates active and passive
cooling during the design phase, aiming for minimal TCO. The methodology is based on a doubly iterative
sizing algorithm using an hourly time resolution. Novel to this domain, it relies on Bayesian optimisation
to determine the most cost-efficient borefield size. It shows that load profiles constrained by maximum
temperature limits in passive cooling configurations or those characterised by significant cooling demand peaks
exhibit substantial potential benefits from a combined active and passive cooling approach. Real-world case
studies further validate our approach, demonstrating achievable total cost of ownership (TCO) savings of 35%
and 33% when optimising the integration of active cooling, compared to traditional passive cooling methods.
To facilitate further research and industry applications, the algorithm has been integrated into the open-source
software GHEtool.

1. Introduction Fig. 1 displays the three classical modes of operation of geothermal


borefields in moderate climates [5]. The first mode is the combination
The building sector is responsible for 25% of global CO2 -eq emis- of the borefield with a ground source heat pump (GSHP) to heat
sions, making this a key sector to decarbonise to deliver the pledges the building. For the cooling of buildings, two different regimes are
of the Paris agreement [1,2]. Decreasing emissions related to heating
possible: active and passive cooling. Passive cooling, also referred to
and cooling of buildings are the low hanging fruit in emission abate-
as direct ground cooling, uses the borefield as a direct heat exchanger,
ment, given the technological developments in this area in the past
decade. The current trends in the building sector, with emphasis on where heat from the building is dissipated into the borefield and no
the expected increase in space cooling demand due to higher global additional GSHP is required. This requires a temperature difference
temperatures and an increase in building floor area, confirm the need between the borefield fluid and the building for sufficient heat transfer,
for sustainable cooling of buildings [3]. whereby the fluid temperature is limited by a maximum temperature.
Aforementioned trends demand for low cost and sustainable climate Active cooling on the other hand allows higher temperatures in the
control techniques for buildings, in which geothermal borefields have borefield due to the use of a GSHP. Here, a reversible GSHP can be
the potential to play a significant role. These systems, possibly com- used to cover both the heating and cooling needs of the building.
bined with other technologies, allow to heat and cool buildings in an
Geothermal borefields in general have experienced difficulties com-
efficient and sustainable way, with a low temperature heating (LTH)
peting with conventional heating and cooling technologies due to
and high temperature cooling (HTC) solution [4].

∗ Correspondence to: Columbuslei 21, 2900 Schoten, Belgium.


E-mail address: matthijsconinx@gmail.com (M. Coninx).

https://doi.org/10.1016/j.apenergy.2023.122261
Received 29 May 2023; Received in revised form 30 October 2023; Accepted 2 November 2023
Available online 15 November 2023
0306-2619/© 2023 Elsevier Ltd. All rights reserved.
M. Coninx et al. Applied Energy 355 (2024) 122261

Nomenclature

C𝑎𝑏𝑎𝑡𝑒𝑚𝑒𝑛𝑡 Abatement cost


C𝑒𝑙𝑒𝑐 Electricity price
C𝑖𝑛𝑣 Borefield investment cost
C𝑜𝑝 Borefield operating cost
COP Coefficient of performance
DR𝑛𝑜𝑚 Nominal discount rate
DR𝑟𝑒𝑎𝑙 Real discount rate
EER Energy Efficiency Ratio
g(t) G-function evaluated at t
IC Investment cost
k𝑠 Ground thermal conductivity
Fig. 1. Modes of operation of geothermal borefields: heating, passive cooling, active
L Borefield size, corresponding to total bore- cooling. A GSHP is required for both heating and active cooling operations, while
field length passive cooling solely requires a heat exchanger [5].
L𝑏ℎ Borefield depth, the depth of one borehole
L𝑚𝑎𝑥 Largest relevant borefield size correspond-
ing to passive cooling only and investment costs associated with the cooling system.1 The need for
L𝑚𝑖𝑛 Smallest possible borefield size correspond- a TCO approach in borefield sizing is already highlighted in the work
ing to maximal combined cooling of Robert and Goslin [9]. Here, the authors make abstraction of the
OC Operating cost optimisation of the ratio between active and passive cooling, resulting
OCc Cooling operating cost in an unsuitable approach for optimally sizing a combined system.
OCh Heating operating cost Therefore, the aim of this work is twofold: explore the potential
p𝑒𝑙𝑒𝑐 Electricity price of the use of active cooling in combination with passive cooling in
geothermal borefields in economic and ecologic terms and develop a
Q𝐻 Thermal energy injected into the heat sink
global borefield sizing optimisation as a part of the open source bore-
Q𝐿 Thermal energy extracted from the heat
field sizing algorithm GHEtool [10]. Both challenges are summarised in
source
the main research question of this work: What is the most economical
r Real discount rate
way to cool buildings by means of geothermal borefields with active
R𝑖 Average yearly HCIP increase and passive cooling?
R∗𝑏 Equivalent thermal borehole resistance Direct ground cooling systems, equivalent to passive cooling with
R𝑔 Ground thermal resistance borefields, have the most potential in colder climates (e.g. Sweden)
T𝑏 Borehole wall temperature with low underground temperatures [11]. In warmer climates, very
T𝑓 𝑝 Average fluid temperature in the borefield large systems for passive cooling would be required to meet cooling de-
due to the peak load mand, offsetting potential environmental and economical benefits [12].
T𝑓 Average fluid temperature in the borefield Arghand concludes in his follow-up work that for larger imbalances,
T𝑔 Undisturbed ground temperature passive cooling combined with district heating is economically justified
T𝑙𝑖𝑚 Limiting temperature in the sizing algo- compared to a combination of active and passive cooling [13]. In the
rithm latter case, active cooling is merely used to cool peak demand, with
T𝑚𝑎𝑥𝐴 Maximum temperature limit for active a non-optimised borefield size (pre-defined cooling load for passive
cooling cooling). This confirms the gap in the literature for accurately sized
borefields for combinations of active and passive cooling.
T𝑚𝑎𝑥𝑃 Maximum temperature for which passive
cooling is possible The sizing of borefields is a well studied topic in the geothermal
sciences, with the development of the sizing algorithms of Monzo [14]
T𝑚𝑖𝑛 Minimum temperature limit
and Bernier [15], depending on whether the temperature of the bore-
TC Total cost
field is limited in the first or last year of operation. Both methodologies
TCO Total cost fo ownership
determine the borefield depth by modelling the ground response of the
borefield to the relevant thermal pulses, representing the cooling and
heating loads. This ground response leads to the temperature profile of
the borefield during its expected lifetime, which is not allowed to ex-
higher associated costs [6,7]. The large investment cost is also consid-
ceed its predetermined limits. Furthermore, Peere et al. [10] combined
ered as the key issue for borefield solutions in the work of Florides [8]. these methodologies in a hybrid sizing algorithm. These methods [14–
Although this is valid for both active and passive configurations, they 16] are designed to size borefields using passive cooling only and are
are both characterised by important differences in their respective cost inadequate to take into account active cooling combined with passive
structure, which is thoroughly discussed in Section 2.3. cooling.
However, there is untapped potential to minimise the costs of The work of Peere [17] acknowledges that the cost of geothermal
these geothermal borefields: the combination of active cooling (as- borefields can be reduced by using hybrid technologies, i.e. combining
borefields with other technologies like solar collectors or dry cool-
sociated with higher operating costs) and passive cooling (associated
ers. Olabi et al. [18] describe the benefits of integrating geothermal
with higher investment costs) can yield cost savings compared to the
borefields with technologies such as solar systems, boilers, biomass
use of only one technology. This is possible by using passive cooling
when possible, and shift to active cooling when required. To reap the
potential benefits, an accurately sized borefield is required based on a 1
TCO (total cost of ownership) and TC (total cost) is used interchangeably
Total Cost of Ownership (TCO) optimisation to balance the operational through this work.

2
M. Coninx et al. Applied Energy 355 (2024) 122261

installations, chillers and cooling towers to support high load applica- borefield fluid must respect (𝑇𝑙𝑖𝑚 ), the necessary borefield size can be
tions. Given the computational complexity of the proposed combined found at any time step.
cooling method, an in-depth quantitative comparison for this technol- 1
ogy is lacking. More recently, Gondal demonstrated geothermal heat 𝑇𝑏 (𝑡𝑛 ) = 𝑇𝑔 +
2𝜋𝑘𝑠 𝐿
exchangers in combination with thermally active building elements can ⎡ 𝑔(𝑡𝑛 ) − 𝑔(𝑡𝑛−1 ) ⎤
significantly reduce building energy use [19]. Finally, the work of [20] [ ] ⎢𝑔(𝑡𝑛−1 ) − 𝑔(𝑡𝑛−2 )⎥
also concludes that hybrid GSHP systems are in most circumstances ⋅ 𝑄̇ 1 𝑄̇ 2 … 𝑄̇ 𝑛 ⋅ ⎢ ⎥ (1)
⎢ ⋮ ⎥
more cost-effective than regular GSHP systems. ⎢ ⎥
⎣ 𝑔(𝑡1 ) ⎦
To formulate a comprehensive answer on the main research ques-
tion, the work follows a three step approach. Firstly, a suitable sizing
̇ 𝑛 )𝑅∗
𝑄(𝑡
methodology for borefields is determined. Secondly, this sizing method- 𝑇𝑓 (𝑡𝑛 ) = 𝑇𝑏 (𝑡𝑛 ) + 𝑏
(2)
ology is extended to include the combination of passive and active 𝐿
cooling. Finally, this extended sizing methodology is used to solve the To accurately determine the increase in operating cost as a result
global economic optimisation problem to minimise the TCO of the of the use of active cooling, a monthly time resolution is insufficient.
borefield, which directly answers the main research question. Therefore, an hourly temporal resolution is used in this work, as a
This methodology is thoroughly explained in Section 2. Then, a the- temperature profile with a sufficient temporal resolution is required.
oretical assessment of this newly developed methodology is presented Furthermore, this sizing methodology evaluates the necessary borefield
in Section 3. Finally, the theoretical assessment is linked to real-life size every time step and saves the largest as final borefield size. This
case studies in Section 4 to validate the findings and showcase later allows, contrary to the method of Monzo and Bernier [14,15] to size a
use-cases of the implemented work in GHEtool [10]. borefield over a period with varying yearly thermal load profiles.
The novelty of this work is twofold. Firstly, the methodology in- The computational cost of evaluating Eq. (1) increases 𝑂(𝑛2 ) with
novatively integrates active and passive cooling in a borefield sizing the amount of evaluations. Considering the hourly time resolution, this
algorithm. Secondly, the combination of the sizing process with an leads to unacceptable computational times. Multiple researchers [24–
economic optimisation using Bayesian optimisation, a novel technique 27] propose various aggregation schemes to reduce the computational
in this domain, resulting in a comprehensive tool for cost-efficient cost but are, per definition, characterised by a loss in accuracy. There-
geothermal borefield sizing. fore this work proposes an alternative method that solves the matrix
product in Eq. (1) as a convolution. It uses the discrete convolution of
𝑄[𝑖] and 𝑔 ′ [𝑖] where 𝑄[𝑖] represents the hourly load and 𝑔 ′ [𝑖] represents
2. Methods
the discrete function of g-value differences: 𝑔 ′ [𝑖] = 𝑔(𝑡𝑖+1 ) − 𝑔(𝑡𝑖 ). The
SciPy package in Python is used to perform the computation, effectively
This paper is build around the following research question: What
reducing computational time by 98%. For a more extensive description
is the most economical way to cool buildings by means of geothermal
on the use of the convolution operator, the reader is referred to [28].
borefields with active and passive cooling? The answer proposed in this
This proposed methodology has been validated in Dymola using the
work follows a three step approach. Firstly, a suitable sizing method-
IDEAS Modelica library and is implemented in GHEtool [10].
ology for borefields is determined (Section 2.1). Secondly, this sizing
methodology is extended to include the combination of passive and
2.2. Iterative sizing approach for combined cooling
active cooling (Section 2.2). Finally, the main research question is ad-
dressed by solving a global economic optimisation problem to minimise
When cooling actively, the upper temperature limit for the borefield
the total cost of ownership (TCO) of the borefield (Section 2.3).
fluid temperature is raised and additional heat is injected to the bore-
field due the operation of the reversed heat pump in case of cooling.
2.1. Borefield sizing Therefore, the traditional sizing approach as presented in Section 2.1
is inadequate and a modified approach is necessary.
The necessary borefield size is such that the fluid circulating inside To accurately size the borefield and model the temperature profile
the borefield does not exceed its predefined temperature limits as a over the lifetime of the borefield, a doubly iterative algorithm is
result of an applied thermal load. Therefore, the ground response to presented following Fig. 2. An outer loop sizes the borefield with a mod-
an applied thermal load is modelled using the g-function proposed ified load profile and an inner loop adjusts the load profile to account
by Eskilson [21]. This function returns a normalised borehole wall for the temperature-dependent COP of the heat pump and the change in
temperature response factor in function of the duration 𝑡𝑞 of a constant thermal load on the borefield due to a changing share of active cooling.
thermal load 𝑞 per unit length for a set of given ground parameters and As the aim of this algorithm is to determine the minimum necessary
a given borefield geometry [22]. borefield size, the temperature of the fluid in the borefield will reach
For a non-constant thermal load profile, the borehole wall tempera- the temperature limit. In this work, the upper temperature limit for
ture 𝑇𝑏 at any time 𝑡𝑛 can be found according to Eq. (1) (based on [23]) passive cooling is set to 16 ◦ C whereas the upper temperature limit for
where 𝑄̇ 𝑖 represents the resulting thermal load (result of subtracting active cooling is set to 25 ◦ C, inspired by local regulations [29]. The
the heat extracted from the borefield from the heat injected into the lower temperature limit is set to 0 ◦ C in both cases. In this approach, a
borefield) on the borefield in a given time step 𝑡𝑖 and 𝐿 the total temperature dependent COP (or EER) is integrated for the heat pump
borefield size or total GHE length.2 The borefield fluid temperature 𝑇𝑓 based on a Galletti heat pump (type: WRE092HSG0) [30]. The impact
can be derived from the borehole wall temperature using the equiva- of these temperature limits is discussed in Section 3.2.
lent thermal borehole resistance 𝑅∗𝑏 expressed in mK W
as described by The first step (see Fig. 2) in the methodology is the initial borefield
Claesson and Javed [24] and shown in Eq. (2) (based on [23]) where sizing with the given hourly load profile as elaborated in Section 2.1.
̇ 𝑛 ) represents the resulting thermal load at time step 𝑡𝑛 . Combining
𝑄(𝑡 In this process an initial temperature profile is constructed.
Eqs. (1) and (2) and substituting 𝑇𝑓 for the temperature limit that the Next, the initial load profile is modified to take into account the
supplementary heat that is dissipated into the ground by operating
the heat pump (active cooling) and is represented by the right hand
2
Borefield size or total GHE length (𝐿) and borehole depth (𝐿𝑏ℎ ) are often side of Fig. 2. Based on the initial temperature profile, the values of
used interchangeably as they are linked to one another by the fixed borefield the COP, and depending on the fluid temperature also the EER, are
geometry. computed every time step and the load is modified accordingly to

3
M. Coninx et al. Applied Energy 355 (2024) 122261

Fig. 2. Schematic representation of the inner (right) and outer loop (left) of the doubly iterative borefield sizing methodology with combined cooling. ‘Adjust load profile for
variable COP’ in the outer loop represents the inner loop.

represent the actual load experienced by the borefield.3 As the modified represent the operating cost for cooling and heating respectively, with
load profile could result in a change in temperature profile and thus the 𝑝𝑒𝑙𝑒𝑐 the electricity price. Finally, the TCO is computed by Eq. (6) where
share of active cooling, this cycle must be continually executed until 𝐶𝑖𝑛𝑣,𝑏𝑜𝑟𝑒𝑓 𝑖𝑒𝑙𝑑 and 𝐶𝑖𝑛𝑣,𝐺𝑆𝐻𝑃 represent the investment cost of the borefield
convergence. This is the inner loop of the doubly iterative approach and heat pump respectively (GSHP). The lifetime of the borefield and
for sizing the borefield with combined cooling. GSHP are defined as 𝑡𝑏𝑜𝑟𝑒𝑓 𝑖𝑒𝑙𝑑 and 𝑡𝐺𝑆𝐻𝑃 , with 𝑟 representing the dis-
After the load is modified to account for the temperature-dependent count rate.5 The lifetime of the borefield is considered as the investment
efficiency of the heat pump and the increased (or decreased) share horizon.
of active cooling, the borefield is sized again using the methodology 1
𝑃ℎ𝑒𝑎𝑡,𝑒𝑙 = ⋅ 𝑄ℎ𝑒𝑎𝑡,𝑙𝑜𝑎𝑑 (3)
explained in Section 2.1. 𝐶𝑂𝑃
Lastly, the algorithm checks whether the most recently computed 1
𝑃𝑐𝑜𝑜𝑙,𝑒𝑙 = ⋅ 𝑄𝑐𝑜𝑜𝑙,𝑙𝑜𝑎𝑑 (4)
borefield size converges with the previously computed borefield size 𝐸𝐸𝑅
(initial sizing in the first pass). As long as convergence is not reached, 𝐶𝑜𝑝,𝑐𝑜𝑜𝑙 + 𝐶𝑜𝑝,ℎ𝑒𝑎𝑡 = 𝑝𝑒𝑙𝑒𝑐 ⋅ (𝑃𝑐𝑜𝑜𝑙,𝑒𝑙 + 𝑃ℎ𝑒𝑎𝑡,𝑒𝑙 ) (5)
the algorithm starts the outer loop of the sizing approach again where
the most recent load profile is considered to be the initial load profile. 𝑇 𝐶𝑂 = 𝐶𝑖𝑛𝑣,𝑏𝑜𝑟𝑒𝑓 𝑖𝑒𝑙𝑑 +
In case the borefield size converges, the algorithm returns the most 𝑡𝑏𝑜𝑟𝑒𝑓 𝑖𝑒𝑙𝑑
⌈ 𝑡 ⌉ 𝑡𝑏𝑜𝑟𝑒𝑓 𝑖𝑒𝑙𝑑
recent temperature profile and size as outputs. ∑
𝐺𝑆𝐻𝑃 𝐶𝑖𝑛𝑣,𝐺𝑆𝐻𝑃 ∑ 𝐶𝑜𝑝,𝑐𝑜𝑜𝑙 + 𝐶𝑜𝑝,ℎ𝑒𝑎𝑡
+ (6)
𝑖=1 (1 + 𝑟)(𝑖−1)⋅𝑡𝐺𝑆𝐻𝑃 𝑖=1
(1 + 𝑟)𝑖
2.3. Economic optimisation
The upfront investment cost of the borefield relates to the total
Section 2.2 describes the approach to determine the minimum borefield size 𝐿. To match the heating and cooling demand of the
borefield size for a given thermal load profile, ensuring a minimal building, the borefield size is determined by the yearly imbalance
investment cost. As this work aims to identify the most cost-efficient (the difference in total cooling and heating load of the borefield) and
solution, it is the total cost of ownership (TCO), consisting of the determined by the building’s peak demand. The variable or operational
investment cost and the operating cost, that must be minimised. When costs of the borefield are linked to the base load, covering the electricity
starting from the minimal borefield size using active and passive cool- cost required to meet the cooling demand. Therefore, passive cooling
ing,4 an increase in borefield size reduces the share of active cooling (as only has an investment cost component driven by peak demand, while
shown in Fig. 3). If the operational cost associated with the reduced active cooling also has operational costs associated with the electricity
active cooling is larger than the investment cost associated with the use of the heat pump, driven by base load [31].
enlarged borefield, a more cost efficient solution has been found. The investment cost of the GSHP can be omitted for the purposes
The investment cost is considered to be a linear relationship with of this work, i.e. determining the potential of active cooling compared
the borefield size. The electrical energy required for heating (𝑃ℎ𝑒𝑎𝑡,𝑒𝑙 ) to passive cooling. This is because of the assumption that a GSHP
and cooling (𝑃𝑐𝑜𝑜𝑙,𝑒𝑙 ) is computed by Eqs. (3) and (4) respectively. The is required to meet the heating demand and no additional cost is
total operating cost is computed by Eq. (5) where 𝐶𝑜𝑝,𝑐𝑜𝑜𝑙 and 𝐶𝑜𝑝,ℎ𝑒𝑎𝑡 required for a reversible GSHP. This assumption is not valid for very
large differences in maximal heating and cooling load, given that this
investment cost scales with maximal load of the GSHP.
3
When the borefield fluid temperature remains below 16 ◦ C, the efficiency The optimisation of the TCO of the borefield aims to determine
of cooling is assumed to be 100% to model the passive cooling operation, whether combining active and passive cooling results in lower costs
corresponding to a perfect heat exchanger.
4
This corresponds to the case where the temperature profile of the borefield
fluid temperature reaches the minimal or maximal temperature limit (0 ◦ C or 5
The ceiling function is used to include additional investments in the GSHP
25 ◦ C). before the end of lifetime of the borefield is achieved.

4
M. Coninx et al. Applied Energy 355 (2024) 122261

function. Furthermore, Latin Hypercube Sampling (LHS) is used to


generate initial points.

2.4. CO2 emissions and abatement cost

Combining active and passive cooling has the potential to reduce


the TCO by decreasing the investment cost of the borefield for a
limited increase in operating cost due to the heat pump used for active
cooling. This electricity use leads to CO2 emissions. Using 127 g CO2 -
eq/kWh as the average carbon intensity of electricity production in
Belgium according to [35], the CO2 emissions related to active cooling
in the optimal solution can be computed easily. In other words, the
economical benefit of combined active and passive cooling comes with
the cost of additional GHG emissions.
The abatement cost (as defined in Eq. (7) where 𝛥𝑇 𝐶𝑂 and 𝛥𝐸
represent the change in TCO and the change in emissions respectively)
allows for easy comparison of effectiveness of different emission reduc-
tion efforts and is widely used by climate change policy makers [36].
Fig. 3. Graphical representation of hypothetical temperature profiles with high Applied to the topic of this work, it represents the additional cost of
potential and low potential for active cooling [28]. reducing CO2 -eq emissions by one unit when moving from combined
active and passive cooling to passive cooling only. A reduction effort
characterised by a lower abatement cost is seen as a more efficient
compared to a solution solely using passive cooling. If so, the bore- allocation of financial resources.
field is sized for a minimal TCO. Therefore, this work identifies two 𝛥𝑇 𝐶𝑂
𝐶𝑎𝑏𝑎𝑡𝑒𝑚𝑒𝑛𝑡 = (7)
extreme borefield sizes. The smallest possible borefield size 𝐿𝑚𝑖𝑛 is a 𝛥𝐸
borefield with maximal combined active and passive cooling, in which This abatement cost can be compared to the price of one European
the temperature limit is reached (25 ◦ C), for a given set of n boreholes. Union emissions Allowance (EUA) under the European Union Emission
The peak demand causes this limit, as this is the limiting factor for the Trading System (EU ETS) which equals roughly 80 EUR/ton CO2 -
borefield size. The largest interesting borefield size 𝐿𝑚𝑎𝑥 is a borefield eq [37] (January 2023). The abatement cost will be further used
using passive cooling only, with no associated variable costs. Also in in Section 4 to justify the activation of active cooling compared to
this case, the temperature limit is reached during peak demand (16 ◦ C), solutions using only passive cooling.
for a given set of n boreholes. As this temperature limit is significantly
lower than in the case with active cooling, a larger borefield is required 3. Theoretical assessment
for the same heating and cooling loads compared to configurations with
active cooling. This is referred to as the maximal interesting borefield This section assesses the performance of the methodology proposed
size as further increasing the size is possible, but does not benefit the in Section 2 by using synthetic load profiles for each of the different
system in any way. borefield quadrants as introduced by Peere et al. [38] (Section 3.1).
However, the borefield size corresponding to the lowest TCO can Afterwards, the insights generated by this theoretical assessment are
also be located in between those extrema. This behaviour can be discussed in Section 3.2.
explained by understanding the peak and base load for heating and
cooling and is graphically shown in Fig. 3. The peak load drives the 3.1. Quadrants and synthetic load profiles
investment costs as the temperature limits will be reached during peak
demand. The base load, on the other hand, drives the operation costs Using the borefield fluid temperature profiles, thermal load profiles
in case active cooling is required. By having a slightly larger borefield are assigned to one of four borefield quadrants based on the imbalance
than the minimal size 𝐿𝑚𝑖𝑛 , a larger portion of the base load can be and the limiting year of operation.6 Fig. 4 shows four load profiles, each
cooled in a passive way for only a limited increase in investment costs. assigned to a different quadrant as introduced in [38]. For example, Q4
To determine the optimal borefield size that leads to a minimal represents a heating dominated thermal load (the temperature profile
TCO over the full lifetime of the borefield for a given load profile, decreases over the simulation period as more heat is extracted from
an economic optimisation consisting of two steps is proposed. First, the borefield than heat that is injected into it) and temperature limited
the minimal borefield size (referred to as size 𝐿𝑚𝑖𝑛 ) and the maximal in the last year of operation. This categorisation is further used to
interesting borefield size (referred to as size 𝐿𝑚𝑎𝑥 ) are determined. This fully understand the potential of using active cooling in geothermal
is done by sizing the borefield for combined active and passive cooling borefields.
as described in Section 2.2 and by sizing the borefield for passive Using synthetic load profiles, the potential for active cooling in all
cooling only, respectively. four borefield quadrants can be analysed. This work proposes using
Next, the TCO function is optimised towards its minimum. However, two types of yearly synthetic load profiles that can be tweaked using
the optimisation of the TCO in function of the borefield size is a non- parameters 𝑀, 𝑁 and 𝑓0 to fit any quadrant. The first load profile
convex optimisation problem without guarantee to reach the global (referred to as profile A), presented by Eqs. (8) and (9) shows seasonal
minimum. Furthermore, evaluating the TCO function for any borefield variations but does not include periodic daily variations. The second
size requires sizing and constructing a full temperature profile with (referred to as profile B) also shows seasonal variations and includes
an hourly time resolution over the full simulation period which is a periodic daily variations on top, presented by Eqs. (10) to (11). In this
computationally expensive task (see Section 2.2). Therefore, Bayesian way, also the effect of highly varying thermal loads on the potential
optimisation is used to find the global minimum of the TCO func-
tion within acceptable confidence levels [32,33]. This is implemented
using the gp-minimise function which is part of the Scikit package 6
Note that the categorisation into the four quadrants is done considering
in Python [34]. The Bayesian optimisation problem is solved using only passive cooling (𝑇𝑙𝑖𝑚 equals 16 ◦ C) and the yearly thermal load profile is
Gaussian processes with the Expected Improvement (EI) acquisition considered to be repeated over the 20 years simulation period.

5
M. Coninx et al. Applied Energy 355 (2024) 122261

Fig. 4. Spatial representation of the four borefield quadrants as introduced n [38]. Q1 and Q4 represent heating dominated load profiles, Q2 and Q3 represent cooling dominated
load profiles. The red and blue curves represent the temperature profile as a result of the heating load and the cooling load respectively. The temperature profiles are shown over
a simulation period of 20 years.

of active cooling can be determined. Note that in Eqs. (8)–(11), 𝑠𝑖𝑛+


represents the positive values of the sine function, all negative values
are set to zero.
( )
𝑄ℎ𝑒𝑎𝑡 = 𝑀 ⋅ 𝑠𝑖𝑛+ 2𝜋𝑓0 (𝑡 − 𝑡0 ) (8)
( )
𝑄𝑐𝑜𝑜𝑙 = 𝑁 ⋅ 𝑠𝑖𝑛+ 2𝜋𝑓0 (𝑡 − 𝑡0 ) (9)

( )
2𝜋
𝑄ℎ𝑒𝑎𝑡 = 𝑀 ⋅ 𝑠𝑖𝑛+ (𝑡 − 6ℎ) ⋅ 𝑠𝑖𝑛+ (2𝜋𝑓0 (𝑡 − 𝑡0 )) (10)
24
( )
2𝜋
𝑄𝑐𝑜𝑜𝑙 = 𝑁 ⋅ 𝑠𝑖𝑛+ (𝑡 − 6ℎ) ⋅ 𝑠𝑖𝑛+ (2𝜋𝑓0 (𝑡 − 𝑡0 )) (11)
24
Fig. 5. Costs curves as a function of borefield depth 𝐿𝑏ℎ , for 8 × 8 rectangular borehole
The proposed borefield sizing methodology is applied to both syn- configuration. Cost curves represented as Total Cost (TC), investment cost (IC), total
thetic load profiles, each tweaked to all four quadrants. This results operation cost (OC), consisting of operation cost for cooling (OCc) and for heating
(OCh). Optimum at 67.02 m depth for fixed 8 × 8 borefield configuration.
in eight theoretical case studies (not all shown here), that determine
the optimal borefield size to minimise the TCO in each case and that
collectively present two important insights.
Evaluating the borefield size for the load profile corresponding covered by active cooling, marginally increasing the operating cost and
to Eqs. (8) and (9) in quadrant 1, results in Fig. 5. This figure presents significantly decreasing the investment cost.
the TCO and its components over the lifetime of the borefield for A comprehensive sensitivity analysis is conducted in [28]. This
different borefield sizes (the result of the Bayesian optimisation). The shows that undisturbed ground temperature 𝑇𝑔 , ground conductivity
rightmost point on the figure represents 𝐿𝑚𝑎𝑥 and the leftmost point 𝑘𝑠 and the maximal temperature for passive cooling 𝑇𝑚𝑎𝑥𝑃 have a
large impact on the TCO of the borefield. Only the latter is a design
represents 𝐿𝑚𝑖𝑛 . The TC curve, representing the TCO of the borefield,
parameter, as 𝑇𝑔 and 𝑘𝑠 are fixed geological parameters. 𝑇𝑚𝑎𝑥𝑃 is
shows an optimum in between the extrema 𝐿𝑚𝑖𝑛 and 𝐿𝑚𝑎𝑥 , displaying
directly influenced by the emission system of the building, making a
the concepts introduced in Section 2.3 (Fig. 3). The utilisation of
case for high-temperature cooling emissions systems, as described in
Bayesian optimisation in the economic optimisation problem results in
Section 1. The second conclusion from the sensitivity analysis is that
the concentration of function evaluations near the optimal solution, as
the electricity price, discount rate and expected lifetime of the borefield
is also shown in Fig. 5.
heavily influence the final result. Therefore, the potential of each use
case is also linked to the economic climate.
3.2. Insights
4. Results and discussion
The first insight gained is that borefields to which load profiles
corresponding to Q2 and Q4 are applied, do not benefit from includ- The proposed methodology has been used to determine the borefield
ing active cooling. The reason is that the minimum temperature in size that minimises the TCO over its lifetime in two case studies using
these cases is limiting. As including active cooling does not alter the the thermal load profile of the building complex ‘The Loop’ in Ghent,
minimum temperature limit, active cooling does not reduce borefield Belgium and the thermal load profile of a local Belgian high school [28,
size. 30]. The assumptions for this case study are based on the Belgian
Borefield installations to which load profiles corresponding to Q1 economical and environmental context and presented in Section 4.1.
and Q3 are applied, do benefit strongly from including active cooling. Then, the results are presented and discussed in Sections 4.2 and 4.3.
The second insight shows that strongly varying load profiles (such The results of these case studies are linked to the insights of the
as profile B) benefit more from active cooling than smoothly varying theoretical assessment in Section 4.4. Finally, the limitations of this
load profiles (such as profile A) as short peaks in cooling load can be work and potential next steps are discussed in Section 4.5.

6
M. Coninx et al. Applied Energy 355 (2024) 122261

Table 1
Numerical values of the parameters in Sections 4.2 and 4.3. The borefield geometry of
the borefield in Sections 4.2 and 4.3 are 20 × 20 and 4 × 4 boreholes respectively.
Variable Value and unit
Lifetime of the borefield 50 years
Temperature dependent COPa −0.3916𝑇𝑓 𝑙𝑢𝑖𝑑 + 17.901
Temperature dependent EERa 0.122𝑇𝑓 𝑙𝑢𝑖𝑑 + 4.365
Fluid 𝑇𝑚𝑎𝑥 for passive cooling 16 ◦ C
Fluid 𝑇𝑚𝑎𝑥 for active cooling 25 ◦ C
Fluid 𝑇𝑚𝑖𝑛 0 ◦C
Electricity price 0.2159 EUR/kWh
Borefield investment cost 35 EUR/m
Real discount rate −0.11%
CO2 -eq intensity electricity 127 g/kWh
Conductivity of the soil 2.1 W/mK
Undisturbed ground temperature 11 ◦ C
Equivalent borehole resistance 0.12 mK/W
Borefield configuration 20 × 20/4 × 4 boreholes
Borehole spacing 6 m
a 𝑇𝑓 𝑙𝑢𝑖𝑑 is the fluid temperature expressed in ◦ C.

4.1. Parameter values

The parameter values used in the methodology described in Sec-


tion 2.1 are presented in Table 1. The lifetime of the borefield is
Fig. 6. Heating and cooling load in presented case studies. Both load profiles
assumed to be 50 years [29]. The electricity price is based on the characterised by high cooling peaks, resulting in additional opportunity for combined
average price in Belgium for non-household consumers whose annual heating and cooling.
consumption is larger than 20 MWh and less than 50 MWh [39]. The
real discount rate is derived from Fishers formula 𝐷𝑅𝑟𝑒𝑎𝑙 = 𝐷𝑅𝑛𝑜𝑚 − 𝑅𝑖
where 𝐷𝑅𝑛𝑜𝑚 is assumed to be 5% and 𝑅𝑖 is the average yearly HICP
(Harmonised Index of Consumer Prices) increase between 2012 and
2021 equal to 5.11% [40]. The investment cost, the undisturbed ground
temperature and the thermal load profiles used in the case studies are
taken from personal correspondence with boydens engineering (part
of Sweco) [30]. Finally, the physical parameters (soil conductivity and
equivalent borehole resistance) are chosen based on [29].

4.2. Case study 1: The loop

Fig. 7. Costs curves as a function of borefield depth 𝐿𝑏ℎ , for 20 × 20 rectangular


Fig. 6(a) shows the yearly thermal load profile of the ‘The Loop’ borehole configuration subjected to the load profile described in Fig. 6(a). Cost
building complex. This is a cooling dominated load and it is limited in curves represented as Total Cost (TC), investment cost (IC), total operation cost (OC),
consisting of operation cost for cooling (OCc) and for heating (OCh). Smallest possible
the last year of operation, and therefore assigned to Q3 according to
borefield size has lowest total cost of ownership by using combined active and passive
Fig. 4. Furthermore, Fig. 6(a) illustrates the strongly varying nature of cooling.
the load profile. Following the insights from the analysis using synthetic
profiles in Section 3.2, a strong benefit from active cooling is expected.
The optimal borefield depth for a configuration with 400 bore- 4.3. Case study 2: Gymnasium
holes determined with the proposed methodology coincides with 𝐿𝑏ℎ,𝑚𝑖𝑛
equalling 39.10 m with a total cost of 1360 kEUR. Compared to the case Fig. 6(b) shows the yearly thermal load profile of the local Belgian
where the borefield is sized traditionally using passive cooling only, this gymnasium. Similar to ‘The Loop’ it is a strongly varying load profile
represents a 34.9% cost saving. This significant cost saving is achieved but in this case it is heating dominated. Due to the large peaks in
by a significant decrease in investment cost for only a small increase in cooling demand (as can be seen in Fig. 6(b)), its temperature profile
cooling operational costs. This effect is enforced by the highly varying is limited in the first year of operation and therefore assigned to Q1.
load profile, as depicted in Fig. 6(a). The cost savings are visualised by The cost components of this case study are not shown in this paper
the TC-curves in Fig. 7. as they are, disregarding the described differences, similar to Fig. 7.
Again, based on the insights provided in Section 3.2, a great potential
Moving from passive cooling only to the optimal situation (where
for active cooling is expected.
part of the load is covered by active cooling) saves 727 kEUR, but The optimal borefield depth for a configuration of 16 boreholes is
leads to the emission of an additional 77.3 ton CO2 -eq. Using Eq. (7), determined using the proposed methodology and is equal to 49.93 m
the abatement cost of saving 1 ton of CO2 by moving towards passive which again coincides with the minimal borefield depth 𝐿𝑏ℎ,𝑚𝑖𝑛 . Moving
cooling equals 9.40 kEUR. This is more than 100 times the current EU from passive cooling only (𝐿𝑏ℎ,𝑚𝑎𝑥 > 125 m) to the optimal combination
ETS allowance price, proving that the use of passive cooling only is of active and passive cooling reduces the TCO by 35.9 kEUR or 33%.
in this case not a cost-efficient emission reduction strategy [37]. This The cost saving leads to a CO2 -eq emission increase of 2.67 ton,
justifies the activation of active cooling regarding CO2 -eq emissions. corresponding to an abatement cost of 13.4 kEUR/ton CO2 -eq when

7
M. Coninx et al. Applied Energy 355 (2024) 122261

opting for passive cooling only. Again this is orders of magnitude higher of 10 kEUR/ton CO2 , demonstrating the usability of the developed
than the current EU ETS price and therefore justifies the activation of methodology in practice. Furthermore the algorithm is integrated in the
active cooling regarding emissions. open-source software GHEtool [10] to facilitate further research and
the application to industry.
4.4. Interpretation of case study
CRediT authorship contribution statement
Both case studies confirm the conclusions from the theoretical as-
sessment, described in Section 3.2. The case studies correspond to Q3 Matthijs Coninx: Writing – review & editing, Writing – origi-
and Q1 respectively, and confirm the large potential of combined active nal draft, Visualization, Validation, Software, Project administration,
and passive cooling in these quadrants. Both profiles are also charac- Methodology, Investigation, Formal analysis, Data curation, Conceptu-
terised by large cooling peaks, validating the second conclusion that alization. Jarne De Nies: Writing – review & editing, Writing – orig-
load profiles with large variations are more interesting for including inal draft, Visualization, Validation, Software, Project administration,
active cooling. Methodology, Investigation, Formal analysis, Data curation, Concep-
tualization. Louis Hermans: Writing – review & editing, Supervision,
4.5. Limitations and further work Resources, Project administration. Wouter Peere: Writing – review &
editing, Supervision, Resources, Project administration. Wim Boydens:
Firstly, this methodology is based upon the assumption of a constant Writing – review & editing, Supervision, Resources, Project administra-
undisturbed ground temperature 𝑇𝑔 irrespective of the depth, making tion. Lieve Helsen: Writing – review & editing, Supervision, Resources,
abstraction of the seasonal effects present in the first 5 m of the Project administration.
borehole. In reality a less deep borefield experiences a lower average
ground temperature compared to a deeper borefield which increases Declaration of competing interest
the EER in cooling operation. Therefore, by reducing the borefield
size, the operating costs for cooling would increase less than what is The authors declare that they have no known competing finan-
shown by the proposed methodology. Also, optimising the borefield cial interests or personal relationships that could have appeared to
configuration, i.e. the number of boreholes, will impact the TCO of the influence the work reported in this paper.
borefield.
Secondly, Fig. 4 clearly shows that the yearly imbalance has a Data availability
strong influence on the necessary borefield size. Regenerative heating
or cooling aims to influence the evolution of the ground tempera- Data will be made available on request.
ture by injecting/extracting heat into/from the borefield using cheap
resources. For example, injecting cheaply available heat (e.g. solar References
thermal energy) into the borefield when a Q4 load profile is applied,
will increase the average ground temperature leading to a smaller [1] International Energy Agency. Greenhouse gas emissions from energy: Emissions
necessary borefield size which might reduce the TCO. by sector. 2019, https://www.iea.org/reports/greenhouse-gas-emissions-from-
Thirdly, intelligently switching between active and passive cooling energy-overview/emissions-by-sector, Accessed on January 10, 2023.
[2] Broer R, Simjanovic Jelena, Toth Z. Implementing the Paris agreement and
using additional borefield control systems that take into account the reducing greenhouse gas emissions throughout the life cycle of buildings:
variability of energy prices, could lead to further cost reductions. European public policies, tools and market initiatives.
Finally, the present study evaluates the potential benefits of combin- [3] International Energy Agency. Is cooling the future of heating?. 2021, https:
ing active and passive cooling in geothermal borefields, based on estab- //www.iea.org/commentaries/is-cooling-the-future-of-heating, Accessed on De-
cember 6, 2022.
lished theoretical principles. Investigating the practical implementation
[4] Behzadi Amirmohammad, Holmberg Sture, Duwig Christophe, Haghighat Fari-
is therefore an important next step. borz, Ooka Ryozo, Sadrizadeh Sasan. Smart design and control of thermal energy
These topics call for additional research that might provide comple- storage in low-temperature heating and high-temperature cooling systems: A
mentary insights to those proposed in this work. comprehensive review. Renew Sustain Energy Rev 2022;166:112625.
[5] Verhelst Clara. Model predictive control of ground coupled heat pump
systems in office buildings (Ph.D. thesis), Leuven: K.U.Leuven. Faculteit
5. Conclusion
Ingenieurswetenschappen; 2012, Diss. doct..
[6] Liu Xiaobing, Polsky Yarom, Qian Defeng, Mcdonald Joshua. An analysis on
Borefields combined with heat pumps are generally praised as an cost reduction potential of vertical bore ground heat exchangers used for
efficient solution to serve the heating and cooling demand in large ground source heat pump systems. 2019, https://pangea.stanford.edu/ERE/pdf/
buildings. Unfortunately, wide-scale borefield deployment is held back IGAstandard/SGW/2019/Liu1.pdf, Accessed on February 10, 2023.
[7] Vasilyev GP, Gornov VF, Kolesova MV, Dmitriev AN, Silaeva VG. Efficiency of
by a large investment cost. This study shows that including active passive utilization of ground "cold" in adaptive geothermal heat pump heating
cooling operation to the standard passive cooling operation in the and cooling systems (AGHCS). In: MATEC web of conferences, vol. 77. Les Ulis:
design phase of the borefield can significantly reduce the total cost EDP Sciences; 2016, p. 6008.
of ownership whilst keeping CO2 -eq emission abatement costs orders [8] Florides Georgios A, Aresti Lazaros, Messaritis Vassilios, Christodoulides Paul.
Performance investigation of a Ground Source Heat Pump system for space
of magnitudes higher than the current EU ETS prices. This justifies
heating and cooling of a typical house in moderate climates. 2019.
economically and ecologically the use of combined active and passive [9] Robert Felix, Gosselin Louis. New methodology to design ground coupled
cooling. This cost reduction is, however, strongly dependent on the heat pump systems based on total cost minimization. Appl Therm Eng
thermal load profile of the application envisaged. 2014;62(2):481–91.
The newly developed borefield sizing algorithm and the economic [10] Peere Wouter, Blanke Tobias. GHEtool: An open-source tool for borefield sizing
in Python. J Open Source Softw 2022;7(76):4406.
optimisation approach using Bayesian optimisation are used together
[11] Dabaieh Marwa, Serageldin Ahmed A. Earth air heat exchanger, Trombe wall and
with theoretical load profiles to gain insights in the potential of using green wall for passive heating and cooling in premium passive refugee house in
active cooling. This work confirms that borefields with a maximum Sweden. Energy Convers Manage 2020;209:112555.
temperature limitation show large potential to benefit from a cost [12] Arghand Taha. Direct-ground cooling systems for office buildings: Design
decrease by combining active and passive cooling. Also, more varying and control considerations (Ph.D. thesis), Chalmers University of Technology,
Sweden; 2019.
load profiles are advantageous for combined active and passive cooling. [13] Arghand Taha, Javed Saqib, Dalenbäck Jan-Olof. Combining direct ground
These insights are confirmed using real-life case studies that show cost- cooling with ground-source heat pumps and district heating: Borehole sizing and
reduction potentials of up to 35% with abatement costs in the order land area requirements. Geothermics 2022;106:102565.

8
M. Coninx et al. Applied Energy 355 (2024) 122261

[14] Monzo Patricia, Bernier Michel, Acuña Jose, Mogensen Palne. A monthly based [28] Coninx Matthijs, De Nies Jarne. Cost-efficient cooling of buildings by means of
bore field sizing methodology with applications to optimum borehole spacing. borefields with active and passive cooling (Thesis), KU Leuven, Belgium; 2022.
ASHRAE Trans 2016;122(1):111–26. [29] Francois L, P. Van Den Bossche, Van Lysebetten G. Ondiepe geother-
[15] Ahmadfard Mohammadamin, Bernier Michel. Modifications to ASHRAE’s siz- mie. Ontwerp en uitvoering van bodemenergiesystemen met U-vormige
ing method for vertical ground heat exchangers. Sci Technol Built Environ bodemwarmtewisselaars. 2016, p. 40.
2018;24(7):803–17. [30] Data received from personal correspondence with boydens engineering (part of
[16] Bernier Michel A. Closed-loop ground-coupled heat pump systems. ASHRAE J Sweco Belgium).
2006;48(9):12–24. [31] Lu Qi, Narsilio Guillermo A, Aditya Gregorius Riyan, Johnston Ian W. Economic
[17] Peere Wouter. Methode voor economische optimalisatie van geothermische analysis of vertical ground source heat pump systems in Melbourne. Energy
verwarmings- en koelsystemen (Master thesis), Leuven: KU Leuven. Faculteit (Oxford) 2017;125:107–17.
Ingenieurswetenschappen; 2020, Diss. Master.. [32] Mockus Jonas. Bayesian approach to global optimization : Theory and
[18] Olabi Abdul Ghani, Mahmoud Montaser, Soudan Bassel, Wilberforce Tabbi, Ra- applications. Dordrecht: Kluwer; 1989.
madan Mohamad. Geothermal based hybrid energy systems, toward eco-friendly [33] Breque Charles. The intuitions behind Bayesian optimization with Gaussian
energy approaches. Renew Energy 2020;147:2003–12. processes. 2019, Accessed on May 4, 2022.
[19] Gondal Irfan Ahmad. Prospects of shallow geothermal systems in HVAC for [34] Scikit-Optimize. Skopt.gp-minimize. 2023, https://scikit-optimize.github.io/
NZEB. Energy Built Environ 2021;2(4):425–35. stable/modules/generated/skopt.gp_minimize.html, Accessed on January 12,
[20] Aditya G Riyan, Mikhaylova Olga, Narsilio Guillermo A, Johnston Ian W. 2023.
Comparative costs of ground source heat pump systems against other forms of [35] Nowtricity. Belgium: Energy and emissions profile. 2021, https://www.
heating and cooling for different climatic conditions. Sustain Energy Technol nowtricity.com/country/belgium/#:~:text=In%202021%20the%20average%
Assess 2020;42:100824. 20emissions,energy%20being%20Nuclear%20(56.7%25). Accessed on February
[21] Eskilson Per. Thermal analysis of heat extraction boreholes (Ph.D. thesis), 12, 2022.
University of Lund, Sweden; 1978. [36] Huang Shihping Kevin, Kuo Lopin, Chou Kuei-Lan. The applicability of
[22] Loveridge Fleur, Powrie William. Temperature response functions (G-functions) marginal abatement cost approach: A comprehensive review. J Clean Prod
for single pile heat exchangers. Energy (Oxford) 2013;57:554–64. 2016;127:59–71.
[23] Bernier Michel. Borefield sizing: Theory and applications. 2015, Slideshow. [37] Trading Economics. Carbon commodity price - trading economics. 2023, https:
[24] Claesson Johan, Javed Saqib. Explicit multipole formulas for calculating thermal //tradingeconomics.com/commodity/carbon, Accessed on March 6, 2023.
resistance of single U-tube ground heat exchangers. Energies 2018;11(1):214. [38] Peere Wouter, Picard Damien, Figueroa Iago Cupeiro, Boydens Wim,
[25] Bernier Michel A, Pinel Patrice, Labib Richard, Paillot Raphael. A multiple load Helsen Lieve. Validated combined first and last year borefield sizing method-
aggregation algorithm for annual hourly simulations of GCHP systems. HVAC&R ology.. In: Proceedings of building simulation 2021: 17th conference of IBPSA.
Res 2004;10(4):471–87. Building simulation, vol. 17, Bruges, Belgium: IBPSA; 2021.
[26] Lui Xiaobing. Development and experimental validation of simulation of hydronic [39] Eurostat. Energy consumption per capita. 2022, https://ec.europa.eu/eurostat/
snow melting systems for bridges (Ph.D. thesis), Tongji; 2005. databrowser/view/NRG_PC_205__custom_2736982/default/table?lang=en,
[27] Marcotte Denis, Pasquier Philippe. Fast fluid and ground temperature com- Accessed on May 12, 2022.
putation for geothermal ground-loop heat exchanger systems. Geothermics [40] Eurostat. Consumer price index - annual inflation rates. 2022, https:
2008;37(6):651–65, International. //ec.europa.eu/eurostat/databrowser/view/PRC_HICP_AIND__custom_2797462/
default/table?lang=en, Accessed on June 2, 2022.

You might also like