Applied Thermal Engineering 73 (2014) 689e696

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Geothermal helical heat exchangers: Comparison and use of

two-dimensional axisymmetric models
Xavier Moch a, b, c, *, Marc Palomares c, Fabrice Claudon d, Bernard Souyri a, b,
Benot Stutz a, b
a

Univ. Savoie, Lab LOCIE, Campus Scientique, Savoie Technolac, F-73376 Le Bourget-du-Lac Cedex, France
CNRS, Lab LOCIE, F-38000 Grenoble, France
RYB-Terra, 33 route de Grenoble, F-38590 Saint-Etienne de Saint-Geoirs, France
d

l'Energie Atomique et aux Energies Alternatives, Institut National de l'Energie Solaire (CEA/INES), F-73370 Le Bourget-du-Lac Cedex,
Commissariat a
France
b
c

h i g h l i g h t s

A thermal model is given for the subsoil.

2D axisymmetric models of helical geothermal heat exchangers are compared.

The necessity of taking into account the earth freezing is shown.

A comparison between the model and experimental results is done.

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 1 February 2014
Accepted 24 June 2014
Available online 24 July 2014

This study concerns near-surface geothermic heat pumps applied to the heating or the cooling of
buildings, and more especially the modeling of new helical heat exchangers buried in the subsoil between 1 and 4 m depth. Two 2D axisymmetric models are considered for the exchangers: a horizontal
rings model and an annular cylindrical conduit model. The models are described, and successfully
compared with literature results applied to ground thermal energy storage. The 2 models give comparable results. The simulations run signicantly faster when modeling the exchanger as an annular cylindrical conduit. The ability of the thermal model of the subsoil to simulate the earth freezing is
validated by comparison to experimental results. The inuence of the liquid fraction included into the
ground on the thermal response of the geothermic heat pumps is analyzed. It has a signicant impact on
the return temperature from the geothermal heat exchangers because of the phase change phenomena
and the increase of the conductivity due to the soil freezing.
2014 Elsevier Ltd. All rights reserved.

Keywords:
Helical heat exchanger
Geothermal spiral coil
2D axisymmetric models
Annular cylindrical conduit
Horizontal rings
Underground freezing

1. Introduction
Several geothermal heat exchangers can be used for lowenthalpy geothermal energy installations. The most common are
BHEs (borehole heat exchangers), pipes descending underground
to a depth of 100 m: their performance is quite steady over a year,
and they require little oor space [1]. The installation costs, in

* Corresponding author. Present address: S2E2, c/o STMicroelectronics, 10 rue

s de Milet, CS 97155, 37071 Tours Cedex 2, France.
Thale
E-mail addresses: xavier.moch@outlook.fr (X. Moch), marc.palomares@ryb.fr
(M. Palomares), fabrice.claudon@cea.fr (F. Claudon), bernard.souyri@univ-savoie.fr
(B. Souyri), benoit.stutz@univ-savoie.fr (B. Stutz).
http://dx.doi.org/10.1016/j.applthermaleng.2014.06.051
1359-4311/ 2014 Elsevier Ltd. All rights reserved.

particular the vertical boreelds, can slow their development.

Moreover, such systems can create a thermal unload of the underground after several years. The average temperature in the underground can decrease by 2 K over 10 years [2], even when there is
no interaction between exchangers.
An alternative method considered as vertical too are the pile
geothermal heat exchangers, as presented in Ref. [3]. They are easy
to include into the foundations of the buildings when piles are
needed for mechanical reasons, so that the installation costs are
reduced. Some countries like France are reticent about their use
because of uncertainties on the ageing of mechanical piles having a
thermal use, whereas they are quite widespread in other countries
like Swiss.

690

X. Moch et al. / Applied Thermal Engineering 73 (2014) 689e696

Nomenclature

Latin letters
a
thermal diffusivity (m2 s1)
cp
specic heat (J kg1 K1)
E
thermal effusivity (J K1 m2 s1/2)
Fo
Fourier number (e)
H
height of the exchanger (m)
h
convective heat transfer (W m2 K1)
L
specic latent heat (J kg1)
Nu
Nusselt number (e)
qv
volumetric ow rate (m3 s1)
R
radius of the exchanger (m)
Re
Reynolds number (e)
Rth
thermal resistance (K/W)
r
radius (m)
T
temperature (K)
t
date (s)
z
depth (m)
Greek letters
rst auxiliary coefcient (e)
second auxiliary coefcient (e)
3
emissivity of the ground (e)
h
ratio of water at liquid state (e)
k
volume fraction of water in the underground (e)
l
thermal conductivity (W m1 K1)
r
mass density
s
StefaneBoltzmann constant (W m2 K4)

a
b

Geothermal horizontal loops are buried pipes installed horizontally at a depth of around 1 m, covering a large area. In France,
the number of such installations has increased over the last few
years. The installation requires few specic tools so that it is not
expansive: but it needs a large available area (typically between
once and twice the heated area, [4]). These exchangers are sensitive
to the outside temperature, and poor sizing can impact vegetation.
Moreover, plants with deep roots have to be avoided since they may
damage the exchanger.
Another kind of exchanger exists whose shape is helical [5], as
shown in Fig. 1. They offer a compromise between the two previous geothermal exchangers. Several geometries are available:
cylindrical or conical; their heights vary between 2 and 6 m and
their diameters between 0.35 and 2 m. The upper part is
generally placed 1 m below the ground level. They do not require

heat ux (W)
heat ux density (W/m2)
annual pulse (rad/s)

Letters with special subscripts

htot
convective and radiative heat transfer (W m2 K1)
Ltot
helical length of the exchanger (m)
re
external radius of the pipe (m)
ri
internal radius of the pipe (m)
Tamp
yearly amplitude of the ground-level temperature (K)
Text
ambient air temperature (K)
Tmean
mean ground temperature (K)
tc
coldest day (s)
Subscripts
f
heat transfer uid
ow
ow
ice
solid water
in
inlet
liq.wat liquid water
low
lower part
mat
Dry matrix of the underground
mix
multi states
out
outlet
pipe
wall of the pipe
ring
ring (annular model)
Accentuation
e
Undisturbed by the exchanger

such a large area as horizontal loops and are claimed to be costeffective.

Various further uncommon exchangers have been reported in
the literature, such as the double coil [6] and exchangers with ns
that increase contact with the pipe [7].
BHEs and horizontal loops have been studied extensively,
experimentally as well as theoretically. There are numerous validated BHE models. Simple ones mainly use three geometries: an
innite linear source, a nite linear source, and an innite cylindrical source. The parameters which have the greatest inuence on
temperature recovery are ground thermal conductivity and heat
capacity of the underground [8]. Porosity plays a major role, since it
acts on both these parameters. The material for cementation is also
important. Atmospheric temperature, wind, and solar radiation
have a marginal inuence, because they act on the heat exchanges

Fig. 1. A house with four spiral heat exchangers (courtesy of RYB-Terra).

X. Moch et al. / Applied Thermal Engineering 73 (2014) 689e696

between the ground and the atmosphere. With a system-oriented

point of view Trillat-Berdal et al [2] and Trillat-Berdal [9] studied
the coupling of BHE with solar panels used to heat a house.
The simplest models of horizontal loops use a linear power source.
Philippe [10] inventoried the various models and reviewed their
principles. For instance, some models take into account the impact of
seasonal temperature variation or coupling with ground humidity:
the latter is generally assumed to be constant throughout the year. The
yearly thermal unload of the underground is not mentioned.
According to Ingersoll et al [11], these models can be used
when the Fourier number Fo at/R2 is greater than 20. In the case
of helical heat exchangers, the characteristic duration must be
greater than 2 months with classical values. Regarding the other
characteristic times, such as outdoor temperature variations or
changes in energy needs, it appears that this hypothesis is not
realistic and that other models have to be used for helical heat
exchangers.
An innite cylindrical model would be able to describe the
phenomena that occur over the short term much better. In this type
of model, the power is no longer extracted on the axis r 0; it is
distributed on a cylindrical area located on r R ([10]). The fact that
the cylinder is lled and does not exchange power with the r < R
part is a disadvantage.
The two main models for helical heat exchangers can be found
in Refs. [12,13,14]. The exchangers are modeled with several horizontal rings or an annular cylindrical conduit. The authors give
useful ways to model the spiral coil, but they do not detail how the
models work or what their advantages, disadvantages, and limits
are. The annular cylindrical conduit is said to be adapted to
describing a helical exchanger when the pitch is small enough
(4e12 cm for [12]; up to 30 cm for [14]).
The details of the models are not given. However, for both
models, it is generally not possible to preserve the volumes (soil in
the middle of the exchanger, heat transfer uid inside the
exchanger) and the contact area between the underground and the
exchanger simultaneously. The way the uid ows is not described
either. We detail these points in the following, and we run the
models to point out the differences in the results.
2. Thermal modeling of the underground
The study of geothermal exchangers involves a preliminary
underground model. In this domain, many authors consider the
underground as a semi-innite solid, spatially homogeneous, with
thermal transfers occurring only by conduction [12,14,15,16].
Its yearly temperature matches the heat equation, with an upper
boundary condition corresponding to a sinusoidal variation of the
~ t Tmean
upper temperature with a 1-year period: T0;
Tamp cosut  tc . The annual undisturbed temperature in the
underground is given by Ref. [12]:

r

 r
~ t Tmean  Tamp exp z u cos ut  tc z u
Tz;
2a
2a

691

Fig. 2. Temperature changes over 1 year.

Temperatures given by Eq. (1) are reproduced in Fig. 2:

In order to simulate working exchangers, the local underground
has to be modeled. This rst step consists in choosing a large
enough volume, so that thermal interactions are clearly described:
far from the exchanger, the temperatures must remain equal to the
undisturbed temperature, given by Eq. (1). Moreover, the boundary
conditions have to take into account the interactions with the
outside (upper surface) and with the far underground (lower
surface).
2.1. Notations
The objective of this study is to model the helical heat
exchanger. Therefore, we opted for a 2D axisymmetric model for
the underground: the axis r 0 corresponds to the axis of the
exchanger. As boundary conditions, we impose an adiabatic
boundary at r r, and heat ux on the upper (z 0) and lower
(z z) bounds. These heat uxes are specied in section 2.2.
r and z are, respectively, a radius and a depth where underground temperatures are not disturbed by the exchanger, that is
~ t for all (z,t) and Tr; z ; t Tz
~ ; t for all (r,t). As
Tr ; z; t Tz;
presented by Ref. [1], they need to be far enough from the
exchanger so that the temperatures such as the thermal ux remain
undisturbed by the geothermal exchanger. In presence of 2.4 mhigh and 1 m-diameter helical ground heat exchanger used to heat
buildings, temperature disturbances are lower than 0.1% at a distance of 7 m from the center of the exchanger ([18]).
2.2. Boundary conditions

(1)

The thermal phenomena next to the ground are not precisely

described. Indeed, the rst dozen of centimeters are sensitive to
daily temperature variations on the ground. However, these
changes are very quickly absorbed, so that this model presents a
good t for deep-enough values, i.e. around 40 cm in general.
Simulations are carried out for values corresponding to the
ry region (French Alps): Tmean 11.9  C and Tamp 9.1  C.
Chambe
The thermal parameters are compatible with many types of soils
(cf. [17]): the thermal conductivity is l 1.7 W m1 K1 and the
heat capacity is rcp 2.5  106 J m3 K1. Thus, the thermal
diffusivity is a l/rcp 6.8  107 m2 s1.

The air temperature near the ground level can be written as a

sum of sine functions. The linearity of the heat equation makes it
possible to consider these functions independently. Let us consider
one of these functions. The boundary condition stands for the atmosphere and is written as follows:

Text t Tmean  Tamp;ext cosut  tc

The temperature of the ground at z 0 can be related to the air
temperature thanks to a heat transfer coefcient htot detailed
below. With this assumption, the mathematical solution giving the
temperature of the underground is the sum of four sinusoids with
the same pulse u:

692

X. Moch et al. / Applied Thermal Engineering 73 (2014) 689e696

1
0
r
Tamp;ext
uA
@
T z; t Tmean 
exp z
2a
1 a4
2
r

u
 4a0 cos ut  tc z
2a
r


u p
acos ut  tc z

2a 4
r


u p
a2 cos ut  tc z

2a 2
3
r


u
3p
5

a3 cos ut  tc z
2a
4


p
 lrcp u
where a h
p
tot
The term expz u=2a implies a sharp decrease of the temperature amplitude for high pulsations u of the air temperature. For
example, let us consider a daily variation of 10  C. With a diffusivity
a 106 C m2/s, the daily temperature amplitude is lower than 1  C
at z  0.4 m and even lower than 0.1  C when z  0.8 m. In a
general case, daily temperature variations are not measurable at
the top of the exchanger, located 1 m below the ground level.
If we consider that the atmospheric temperature is the sum of a
1-year-period sinusoid and of a 1-day-period sinusoid (describing,
respectively, the fact that temperatures are lower at night and
during winter), only the 1-year-period sinusoid needs be taken into
account to evaluate the underground temperatures: in the
following, u stands for the annual pulse.
p
The order of magnitude of lrcp u=htot is 0.06, ranging from 0 to
0.2. Except for particular cases, such as very good insulation on a
ground with high thermal effusivity, the sinusoid in a0 is the only
term to be considered in the expression of Tz; t (the other funcTamp;ext
tions in a1, a2, and a3 are negligible). 1a
4 xTamp and as a consequence, the temperature at the ground level can be likened to the
air temperature. Moreover, at signicant depths, thermal variations
are already damped: the difference between reality and the model
generally cannot be measured. Moreover, it should be remembered
that htot is actually a function of time, mainly due to wind action,
but also to sunlight, plantings in the ground and atmospheric and
ground humidity. Therefore, it does not make sense to expect a high
level of accuracy of the input data.
When the underground is thermally not disturbed, the natural
~ z; t in the underground can be deduced from Eq. (1):
heat ux 4

~ z; t l
4

r

r 

p
vT~
u
p
u
E uTamp cos ut  tc
z
exp
z
vz
2a
4
2a

temperature is T(0,t) (h is the convective heat transfer coefcient

between the ground and atmosphere). The additional convective
~
ux is thus hT0; t  T0;
t.
~
Similarly, when the ground temperature is T0;
t, the ground
~ t4 , while its value is sT(0,t)4 when the ground
radiation is sT0;
temperature is T(0,t). The additional radiative ux is
~
~
~
sT0; t4  T0;
t4 x4sT0;
t3 T0; t  T0;
t when T0; tx
~
T0; t. In order to implement a constant coefcient in the simula~ t3 with T 3
tions, we approximate T0;
mean , since Tamp Tmean (the
unit is Kelvin).
As a consequence, as long as the ground temperature is close
enough to the undisturbed temperature, the additional ux has a
~ t, with htot h43 sT3mean. This includes
value of htot T0; t  T0;
convection and radiation on the upper boundary. From the results
of [13], we can consider that an average value is
htot 15 W m2 K1. Fig. 3 summarizes these boundary conditions
(the exchanger depicted on the left is the horizontal rings model,
while the exchanger depicted on the right is the annular cylindrical
conduit one, cf. Section 3).
2.3. Summary
The underground is described in a 2D axisymmetric way. It is
modeled as a cylinder, with height ranging from z < 0 to 0 and
radius ranging from 0 to r. The conditions applied to the underground are:
1. Initial condition at t tini

r
 

 r
u
u
 cos u tini  tc z
Tz; tini Tmean  Tamp exp z
2a
2a
(3)
2. Boundary conditions:

axial symmetry at r 0;

adiabatic condition at r r;

heat ux condition at z z


 
 r
r 
p
u
u
p
z cos u t tc
z
4z ;t E uTamp exp
2a
2a
4
(4)

heat ux condition at z 0


p
p
~ t
htot T0; t  T0;
40; t E uTamp cos ut  tc
4

(5)

(2)
p
with E lrcp the thermal effusivity of the underground.
~ z ; t is applied as a boundary condition at z z. On
This ux 4
the ground, 4
~ 0; t is given in the same way. However, an additional
heat ux must be taken into account when the temperature T(r,0,t)
~
differs from T0;
t, i.e, when the ground surface is warmer or colder
than it would be without the exchanger.
For simplicity reasons, T(r,0,t) will be written T(0,t) in the
following.
This additional ux refers to the inuence of the difference of
temperature at ground level on the thermal exchange with the
~ t, where htot
atmosphere. Its value is htot T0; t  T0;
(W m2 K1) takes into account convection and radiation at z 0.
~ t for the ground
Indeed, when the temperature is equal to T0;
and to Text(t) for the atmosphere, the convective ux is
~
hT0;
t  Text t, while it is h(T(0,t)Text(t)) when the ground

Fig. 3. Boundary conditions for 2D-axisymmetric models (left: horizontal rings; right:
annular cylindrical conduit).

X. Moch et al. / Applied Thermal Engineering 73 (2014) 689e696

693

2.4. Underground freezing

3.1. General points for both models

Below 0  C, the underground water will freeze. We take into

account the thermal consequences of this phenomenon by making
l and rcp depend on the local temperature. Moreover, latent heat L
is released. For a given state (liquid or solid), we keep considering l
and rcp as constant.
It is well-known that the liquidesolid phase change of water is
accompanied by expansion, and thus leads to ground displacement.
A complete description of this phenomenon is complex and would
need to take into account an induced displacement of the air contained in the underground. The mechanical aspects such as the
variation of volume due to the freezing of water are not taken into
account in this work.

Junctions between the heat pump and the extremities of the

exchangers are not taken into account. It is generally assumed that
the heat transfer uid ows from the upper to the lower bound of
the spiral coil.
The thermal resistance between the underground and the heat
transfer uid is calculated through the geometry of the wall, the
thermal properties of the material constituting the exchanger, the
thermal conductivity of the heat transfer uid, and the Nusselt
number. The total thermal resistance of the object modeled has to
be the same as the thermal resistance of the real object, i.e, the sum
of the thermal resistance of the pipe, Rth,pipe, and the ow, Rth,ow:

Rth;pipe
2.4.1. Physical principle
To make the calculation easier, we consider a progressive
freezing of the water contained in the underground, between 0  C
and 1  C. The volumic ratio of water in the underground is constant and equal to k 20%: there is neither migration of water nor
convective transport. While the transition between liquid water
and ice is occurring, latent heat is released. Moreover, the thermal
properties of the underground evolve, since thermal conductivity
and heat capacity are not the same for liquid water and ice.

2.4.2. Modeling
The underground temperatures indicate the ratio of frozen
water. For numerical reasons, we use a smoothed Heaviside function to describe the fraction h of liquid water in the liquid state.
The thermal conductivity of the mixture such as its volumic heat
capacity is calculated with the thermal conductivities (respectively,
volumic heat capacities) of its components:

lmix 1  klmat khlliq:wat k1  hlice

(6)

 
 
 
 
rcp mix 1  k rcp mat kh rcp liq:wat k1  h rcp ice

Rth;flow

 
ln rrei
2pLtot lpipe
1
pNuLtot lf

(9)

(10)

Moreover, there exists a thermal resistance of external contact

Rth,ext, due to the potential development of an air layer between the
pipe and the underground, inherent to freezing and thawing cycles
during extraction. Considering the other thermal resistances, we
assume that this resistance can be neglected.
The order of magnitude of the volumetric ow rate in the
exchanger is 4 L min1. The heat transfer uid is mono-propylene
glycol, so that the kinematic viscosity should remain between
2  106 and 10  106 m2/s, depending on the temperature and
the concentration. Through a cross-section 2ri 20.4  103 m, the
Reynolds number is Re < 2000: the ow in the exchanger is
assumed to be laminar, as is the case in Ref. [13].
We use Nu 4.36: it matches with a constant ux condition.
This hypothesis is realistic: assuming a steady state for the heat
pump, the temperature of the uid varies quasi-linearly with the
distance from the inlet.
3.2. Horizontal rings

(7)
Moreover, the phase change is taken into account in the heat
equation written as:

  vT
vhk
rLliq:wat
rcp mix Vlmix VT 
vt
vt
that is:


Vlmix VT

 
dh
krLliq:wat
rcp mix
dT

vT
vt

Thus, we dene the equivalent heat capacity of the underground

as:

 
 
dh
krLliq:wat
rcp mix;eq rcp mix
dT

(8)

Since thermal properties of the underground may be substantially modied locally with freezing, the mesh around the
exchanger has to be rened.

3. Modeling the helical heat exchanger

Two models for the heat exchanger are compared in this paper,
namely horizontal rings and the annular cylindrical conduit.

3.2.1. Principle
It appears quite natural to represent the helical geometry with
several horizontal rings. Using the cylindrical symmetry, the study
becomes 2D axisymmetric, as reported by Ref. [13]. Thus it is
possible to keep the radius of the exchanger, its pitch, and the
radius of the pipe (internal and external). Since the number of rings
is a natural number, the height of the exchanger as well as the total
length of the pipe generally cannot be preserved, but the lower the
pitch, the better the approximations are.
The model consists of N rings, numbered from 1 (inlet ring) to
N (outlet ring). To calculate the temperature of the heat transfer
uid in a ring i, one must consider its temperature in the previous
ring (i1) such as the thermal ux on the latter ring i1. On the rst
ring, the inlet temperature is calculated using the outlet temperature and the geothermal power of the heat pump.
A spontaneous method to calculate the way the temperature
evolves in the rings consists in performing an enthalpy balance (see
Eq. (12) below). Numerical problems arise with this solution:
indeed, ux and temperature are strongly related, and solvers are
generally not developed to solve such strongly coupled boundary
conditions. Therefore, a direct calculation was developed, based
upon the temperature of the underground next to the rings and
considering the pipe as a resistive-only material (see Eq. (11)
below). It does not take into account the heat capacity of the pipe
and the uid, whose value is about 150  103 J/K (42 Wh K1): the

694

X. Moch et al. / Applied Thermal Engineering 73 (2014) 689e696

power released by the cooling of the uid and the pipe is ignored in
comparison with the geothermal power. This method is detailed in
the following sections.

resistance between the heat transfer uid and the underground is

retained.

3.2.2. Notations for the horizontal ring model

Tfi is the temperature of the heat transfer uid entering ring i.
Thus, the inlet temperature is Tf1 and the outlet temperature is TfN1 .
Tsi is the mean temperature of the underground next to the ring i.
Rth,ring N(Rth,pipe Rth,ow) is the thermal resistance of one
ring, taking into account the corresponding uid ow estimated
using the Nusselt number with a uniform surface heat ux for
circular tubes (Nu 4.36).
qv is the volumetric ow in the exchanger, (rcp)f is the volumetric heat capacity of the heat transfer uid, P is the incoming
geothermal power (into the exchanger), and Fi is the incoming
geothermal heat ux incoming into ring i.
Finally, let b be dened by:

3.3.2. Heat transfer uid ow

Since the model is 2D axisymmetric, the ow cannot have an
orthoradial component: it is vertical. Its velocity corresponds to the
vertical component of the velocity of the uid inside the real helical
exchanger. Since the height is kept, the transit time is also retained.
Moreover, since the uid volume is kept, the volumic ow rate is
also retained.
The boundary conditions are much easier to express than those
corresponding to the horizontal rings. At each time step, only two
parameters are needed: the geothermal power P to be extracted
and the outlet temperature Tout (computed at the lower side of the
cylinder).
These parameters enable the calculation of the inlet temperature Tin to be applied to the upper side of the cylinder:

1
b 
rcp f qv Rth;ring

P
Tin Tout   
rcp f qv
3.4. Numerical run

3.2.3. Fluid temperature

The model uses the following equations:

Tsi  Tfi Rth;ring Fi

(11)

F
Tfi1  Tfi  i
rcp f qv

(12)

N
X

Fi P

(13)

i1

When considering a known inlet temperature, Eqs. (11) and (12)

are sufcient to calculate the values of Tfi easily. When considering
a known geothermal power, we need to take into account Eq. (13).
In this case, it is quite easy to obtain the theoretical value of the
outlet temperature using the geothermal power and the underground temperatures:

"

N
X
b
TfN1 
1  bNi Tsi

N
N
rcp f qv 1  b  1
1  b  1 i1

P


1  bN

(14)
For i from N down to 1, Tfi is given by:

Tfi

Tfi1  bTsi
1b

(16)

(15)

3.3. Annular cylindrical conduit

This model appears in Refs. [12,13,14] but is not detailed. We will
undertake it in this part.
3.3.1. Size
We choose not to keep the exchange area between the
exchanger and the underground. The cylinder has the same (mean)
radius and the same height as the exchanger, so that the underground volume in the middle of the exchanger is kept.
The ow section is chosen so that that the volume of the heat
transfer uid is kept. This is the only inuence of the pitch.
The thickness of the wall is not very important, as long as it is
low compared to the radius of the cylinder. The value of the thermal
conductivity of the cylinder walls is set so that the total thermal

The simulations are carried out on a PC Intel Duo Core at

3.32 GHz with 3.25 Gb RAM, and version 4.2 of COMSOL Multiphysics. The transient heat equation is solved thanks to the niteelements theory and the solver is Pardiso (relative tolerance:
104; absolute tolerance: 105).
The annular cylindrical conduit also requires substantial RAM.
However, the calculation over 1 year lasted about 30 s, i.e, about 40
times faster than the horizontal ring model with direct calculation.
With a very ne mesh next to the cylinder, the calculation lasted a
little more than 1 min.
4. Comparison with a validated model
Rabin et al. ([14]) provide results for a 6-m-high exchanger
buried at a depth of 10 m. The underground has a constant temperature of 20  C and is insulated at a distance r 3 m from the
axis of the exchanger. Water is injected at 70  C at a ow-rate of
0.04 m3/h for 150 days, owing upwards. Immediately after, water
is injected at 20  C with the same ow rate for 30 days. Results are
given for two types of soil:

l 1.8 W m1 K1 and rcp 3.18  106 J m3 K1 (case 1);

l 0.8 W m1 K1 and rcp 2.5  106 J m3 K1 (case 2).
The models presented in sections 3.2 and 3.3 were adapted in
order to make the heat transfer uid go upwards. r 3 m was
specied for both models. Fig. 4 compares the results in these soils.
Both models show quite a good match with the results of Rabin
et al.; the outlet temperatures are slightly higher due to the thermal
resistance. Moreover, the simulations run signicantly faster when
modeling the exchanger as an annular ring conduit. This model will
be used in the following.
5. Underground freezing
5.1. Theoretical approach
To prove the value of taking into account the freezing of the
underground water content, three simulations are done. At the
beginning, the underground temperature is homogeneous at 10  C,
and the ambient one remains equal to 10  C during the whole
simulation. A constant power, large enough to make the

X. Moch et al. / Applied Thermal Engineering 73 (2014) 689e696

695

When the heat pump stops, the liquid inside the exchangers
continue to ow in the closed loop, leading to the homogenization
of the temperature along the exchanger. The temperature of the
surrounding soil increases leading to an increased temperature of
the liquid owing into the tubes. The ice core reaches quickly 0  C
just after the stop of the heat pump because of its high thermal
conductivity. The fusion of the ice core leads to a lower temperature
of the soil around the exchanger compared to the one observed
with dry soil 2.5 days after the heat-pump stop.
Therefore, the freezing phenomenon must be taken into account
when modeling the underground: it would not sufce to consider
only the thermal parameters of the underground at the beginning,
when the entire water content is liquid.
5.2. Experimental approach

Fig. 4. Comparison with [14].

underground freeze, is continuously extracted over 3 days. Then

the extraction is stopped, but the circulator still works for 11 days
(relaxation time).
Three soils are simulated, each of them containing a different
volumetric ratio of water (0%, 20%, 50%). These soils are chosen so
that the thermal parameters are the same when the whole water
content is liquid. Fig. 5 shows how the outlet temperature evolves
(when geothermal power is extracted, the inlet temperature is
about 3.5 K lower).
The temperature signicantly decreases when the heat pump
works. The heat is rst extracted by transient conduction. When the
temperature goes below zero, liquidesolid phase change occurs,
and an ice core grows around the exchanger. The latent heat of
fusion released by ice formation associated with the increase of the
heat conduction due to crystallization reduces the temperature
decrease of the ground surrounding the exchangers. Those effects
are all the most important than the liquid fraction included into the
ground is high.

Fig. 5. Inuence of underground freezing on outlet temperatures.

An experimental platform was created a few kilometers away

ry (Savoy, the French Alps). Helical heat exchangers
from Chambe
were connected to a heat pump. The exchangers are H 2.4 m
high and R 0.5 m wide. The platform is a dozen meters away
from a river, the Leysse, and the underground is mainly
composed of silt and sand. Water was found at a depth of 3.5 m
or 4 m, just below the lowest part of the exchanger. According to
[19] and [20], we rst estimate the volumetric water content at
k 0.4.
For this experiment, the heat pump worked steadily for several
days, until temperatures below 0  C could be reached. Temperatures were measured at several places, and the geothermal power
was calculated using the ow rate. This power was then used as a
working condition for the model.
Moreover, the atmospheric temperature values over a mean
year were used to obtain the Tmean, Tamp and tc values. We measured
the evolution of natural temperatures in the underground, i.e,
temperatures not disturbed by the geothermal exchangers.
Although we cannot provide the value of diffusivity a, these temperatures were in coherence with theoretical temperatures given
by Eq. (1) with ax1.1106 m2/s.
Therefore, we used a value l 2.6 W m2 K1 and
rcp 2.3  106 J m3 K1 for the simulation. The value of thermal
conductivity is slightly high according to [17], but reasonable.
Fig. 6 compares the experimental and simulated temperatures
at the following points:

Fig. 6. Comparison between simulation and experimental work.

696

X. Moch et al. / Applied Thermal Engineering 73 (2014) 689e696

Outlet temperature of the heat transfer uid;

Axis (at mid-height);

Observation 1 m from the axis (at mid-height).
The temperatures on the axis retain a value of 0  C for a long
time (more than 200 h). This proves the necessity of taking into
account the effect of freezing.
We probably overestimated the water content, since the simulated temperatures on the axis remain at 0  C for a longer time in
the simulation than in reality. Conversely, the model underestimates the outlet temperatures when the water freezes. This
may be a consequence of the way we modeled the freezing of water,
between 0  C and 1  C: indeed, experimental and simulated outlet
temperatures converge again when all the water is solid (after
400 h). We could not check this hypothesis by reducing this interval, because of a lack of computer memory. At a distance of 1 m
from the axis, the model overestimates the temperatures: this effect may be related to poor estimation of the thermal properties for
the underground or to a ow of water in the underground involving
a heat transfer by advection.
The model is not perfect and should be improved, but these
results are in accordance with experimental results.
6. Conclusion
In this paper, we have presented the thermal interaction between the underground and a helical heat exchanger. We have seen
how the phenomena in the underground can be described (natural
variation of temperature, freezing), as well as two ways of modeling
heat transfers with the exchanger (horizontal rings, annular cylindrical conduit).
Simulations and comparison with existing results prove that
both models can give useful results. Therefore, we recommend
using the annular cylindrical conduit, which is easier to use and
runs signicantly faster.
Finally, we showed, by simulation and experimental work, the
effect of freezing in the underground.
A number of questions remain and need to be further studied. Is
there a thermal unload of the underground over several years when
the heating and cooling needs are not well-balanced? What is the
inuence of all the parameters over the thermal results? Can the
real cycles of a heat pump be simulated? These questions will be
treated in another paper.

References
[1] T.Y. Ozudogru, C.G. Olgun, A. Senol, 3D numerical modeling of vertical
geothermal heat exchangers, Geothermics 51 (2014) 312e334.
[2] V. Trillat-Berdal, B. Souyri, G. Achard, Coupling of geothermal heat pumps
with thermal solar collectors, Appl. Therm. Eng. 27 (2007) 1750e1755.
[3] Stuart J. Self, Bale V. Reddy, Marc A. Rosen, Geothermal heat pump systems:
status review and comparison with other heating options, Appl. Energy 101
(2013) 341e348.
[4] M. Bouchi-Lamontagne. Geothermal energy: Which technics for which use?
(France e in French). Technical report, ADEME and BRGM.
[5] Xavier Moch, Marc Palomares, Fabrice Claudon, Bernard Souyri, and Benot
ry, France, August
Stutz. Geothermal Helical Heat Exchanger, BS2013, Chambe
26e28, (2013), 2949e2955.
[6] Y. Bi, L. Chen, C. Wu, Ground heat exchanger temperature distribution analysis
and experimental verication, Appl. Therm. Eng. 22 (2002) 183e189.
[7] D.A. Alkhasova, M.G. Alishaev, Calculating the gain in the efciency of a
borehole heat exchanger due to furnishing its heat-transfer surface with
longitudinal nning, Therm. Eng. 56 (11) (2009) 970e976.
[8] Y. Shang, S. Li, H. Li, Analysis of geo-temperature recovery under intermittent
operation of ground-source heat pump, Energy Build. 43 (2011) 935e943.
[9] Valentin Trillat-Berdal, Energy integration in Buildings: Combined Use of Solar
Energy and Low-temperature Geothermal Energy (in French), PhD thesis,
 de Savoie, France, 2006.
Universite
[10] M. Philippe, Development and Experimental Validation of Horizontal and
Vertical Geothermal Heat Exchangers for Heating Residential Buildings (in
French), PhD thesis, Ecole des Mines de Paris, France, 2010.
[11] L.R. Ingersoll, O.J. Zobel, A.C. Ingersoll, Heat Conduction with Engineering,
Geological, and Other Applications, Mc Graw-Hill, 1954.
[12] C. Doughty, A. Nir, C.F. Tsang, Seasonal Thermal Energy Storage in Unsatured
Soils: Model Development and Field Validation, University of California, june
1991.
[13] Y. Rabin, E. Korin, E. Sher, A simplied model for helical heat exchanger for
long-term energy storage in soil, in: Springer-Verlag (Ed.), Design and Operation of Heat Exchangers, Eurotherm, 1991, pp. 305e314.
[14] Y. Rabin, E. Korin, Thermal analysis of a helical heat exchanger for ground
thermal energy storage in arid zones, Int. J. Heat Mass Transf. 39 (5) (1996)
1051e1065.
[15] R. Nebbali, S. Makhlouf, Determining the distribution of temperature eld in
the ground, by a semi-analytical model. boundary conditions for the purposes
nergies renouvelof simulation of a greenhouse crop (in French), Rev. des e
ables (2007) 255e258. CER07 Oujda.
[16] M. Deveughele, P. Vercamer, Thermal use of close underground with a heat
pump (in French), Bull. Int. Assoc. Eng. Geol. 28 (1983) 81e89.
[17] VDI 4640. Technical report Verein Deutscher Ingenieure, VDI-Gesellschaft
Energietechnik Fachausschuss Regenerative Energien (Germany), December
2000.
[18] Xavier Moch, Theoretical and Experimental Study of Geothermal Helical Heat
Exchangers: Warming and Cooling with Heat Pumps and Sizing Installations
 de Grenoble, France, February 2013.
(in French), PhD thesis, Universite
[19] M. Th. van Genuchten, A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils, Soil Sci. Soc. Am. J. 44 (1980) 892e898.
sten, A. Lilly, A. Nemes, C. Le Bas, Development and use of a database
[20] J.H.M. Wo
of hydraulic properties of European soils, Geoderma 90 (1999) 169e185.