You are on page 1of 37

Accepted Manuscript

Horizontal Ground Heat Exchangers Modelling

Louis Lamarche

PII: S1359-4311(18)37474-X
DOI: https://doi.org/10.1016/j.applthermaleng.2019.04.006
Reference: ATE 13596

To appear in: Applied Thermal Engineering

Received Date: 5 December 2018


Revised Date: 20 February 2019
Accepted Date: 2 April 2019

Please cite this article as: L. Lamarche, Horizontal Ground Heat Exchangers Modelling, Applied Thermal
Engineering (2019), doi: https://doi.org/10.1016/j.applthermaleng.2019.04.006

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Horizontal Ground Heat Exchangers Modelling

Louis Lamarchea,*
aÉcole
de technologie supérieure,
1100 Notre-Dame Streetr West, Montreal, Canada, H3C 1K3

Abstract

Geothermal heat exchangers applied to building air conditioning and heating systems are
often categorized as vertical and horizontal configurations. In this article, an analytical
model is presented to simulate on an hourly basis, ground heat exchangers having a
horizontal configuration. The model is based on a new formalism of the finite line source
approach that allows faster simulation. The model presents a novel approach to take into
the effect of seasonal temperature variations especially when the inlet temperatures in the
ground heat exchanger are imposed. It also takes into account the effect of the air-soil
boundary condition and the thermal interference between the different portions of pipes in
the buried field. The temperature as well as the heat flux distribution along the ground heat
exchanger is part of the solution and not a priori assumed as in the case of previous models.
The model does not consider the possible phase change due to ice formation or the moisture
content variation associated with some horizontal configurations. It is based on recent
studies related to vertical boreholes and adapted to horizontal systems. It is intended to be a
valuable tool to simulate on an hourly basis, the time response of the ground heat exchanger
coupled with a building equipped with heat pumps.

*
Tel: 514-396-8858 Email address: louis.lamarche@etsmtl.ca (Louis Lamarche)
Preprint submitted to Elsevier Science June 13, 2007
Key words: Ground source heat pump, Horizontal ground heat exchanger, Finite line
source.

2
Nomenclature

A Area (m2)
Cp Specific heat capacity (J-Kg-1 –K-1)
D Diameter (m)
E1 Exponential integral function
h Heat transfer coefficient (W m −2K −1) or thermal response factor.
H Pipe segment length (m)
k Thermal conductivity (W m −1K −1)
I Infinite line source function
L Pipe length per trench (m)
m Mass flow rate (kg-s-1)
q Heat load (W )
q’ Heat flux per unit length (W m −1)
q” Heat flux per unit area (W m −2)
r Radial coordinate (m)
R Unit length thermal resistance (K m W −1)
Sp Group of variables defined in Eq. A6
Sq Group of variables defined in Eq. A6
t Time (s)
T Temperature (K)
u Fluid velocity (m s-1)
X Group of variables defined in Eq. 29
Y Group of variables defined in Eq. 31
x-y- z Cartesian coordinates system (m)

Greek letters
2 −1
α Thermal diffusivity (m s )
 Reduced thermal conductance (Eq. 19)
 Group of variables (defined in Eq. 11) (K m W −1)
 Group of variables (defined in Eq. 19)
 Time integral variable
 Dimensionless temperature (defined in Eq. 17)
o Exit dimensionless temperature (defined in Eq. 20)
 Annual frequency of temperature variations
 Group of variables (defined in Eq. 61)

Subscripts

3
p At the pipe radius
f Fluid
o Far-field value
s Soil values

1. Introduction

Ground-Coupled Heat Pumps (GCHP) using a secondary loop ground heat exchanger are
often categorized as vertical and horizontal loop systems, although hybrid systems such as
inclined or pile systems, are also available. Vertical borehole systems are the most popular
technology in Europe, but in North America, horizontal systems continue to be largely used
[1, 2]. This observation notwithstanding, there is general agreement that the scientific
literature on horizontal systems is much more sparse than that on vertical ones [3]. Design
approaches often continue to be associated with rules-of-thumb and approximate charts
[3]. Even though some common physical aspects link both technologies, there are some
specific factors associated with horizontal loops that make the design of such systems
unique. The following is a non-exhaustive list of such factors:
1. The ground surface has a greater influence on the thermal response than for vertical
systems.
2. Seasonal temperature variations, which are often neglected in vertical systems, are
very important in horizontal ones.
3. Horizontal piping lay in trenches in the subsurface overburden where the moisture
content can influence the thermal performance of the heat exchanger.
4. Phase change due to freezing and thawing may occur.
5. Several configurations are possible, including inline, slinky, etc., which cause pipe-
to-pipe interference.

Given the above factors, horizontal loops are more complicated to analyse, even though
they may be easier to install. Several studies have looked at these systems. A well-known
report on horizontal systems [4] by Claesson and Dunand is often used as a reference even
though several aspects shown in the list above not addressed. Mei [5] develop a numerical
model for horizontal systems, which was extended by Piechowski [6] to include mass

4
transfer due to moisture content. More recent studies relying on numerical simulations
include the work of Kupiec et al. [7], Kayaci and Demir [8] Congedo et al. [9], Gan [10]
and Soni et al. [11]. Various configurations of the horizontal heat exchanger, including
inline piping, slinky or spiral configurations have also recently been analysed [9, 12-14].
Some experimental works on horizontal configurations are also available [13, 15]. In the
case of vertical boreholes, significant recent research effort has been devoted to the
analytical method [16-18]. A recent study by Fontaine et al. [19] follows similar steps as in
the case of horizontal loops. In their approach, they split a horizontal loop into several
discrete sections, and then extend the Claesson and Dunand [4] method developed for the
steady-state analysis case to the unsteady case. To that end, they use the finite line source
(FLS) solution to evaluate the thermal response of the discrete pipe sections in the case of
the time-varying heat exchange, but assuming that at each time step, the heat transfer rate
variation along the pipe follows the steady-state Claesson and Dunand solution. In this
paper, we follow similar steps, but the heat transfer rate distribution along the pipe is not
known a priori; rather, it is part of the final solution. Another simplification is also used for
the FLS implementation. The model addresses some, but not all, of the horizontal loop
distinct characteristics taken into account. The impact of the ground boundary and the
seasonal temperature variation are however covered in a rigorous manner. The soil is
assumed to be homogeneous, and for now, no mass transfer is assumed. The possible phase
change associated with freezing is not taken into account. It must be pointed out that in the
Fontaine model, the freezing of the permafrost is discussed, but not solved explicitly.

If no phase change occurs and the ground is assumed homogeneous, the temperature
variation in the ground can easily be modelled if one assumes a sinusoidal variation of the
temperature at the air-soil boundary. A well-known solution in this regard was given by
Kusuda and Achenback [20] . Assuming that the surface temperature is known and given
by:


Ts (t )  To - A cos  (t - tshift )  (1)

The solution in the ground is:

5
    z 365 
Tground ( z, t )  To - A exp  -z  cos   (t - tshift - )  (2)
 365   2  
where To is the mean temperature far from the boundary,  is the annual frequency, A is

the variation amplitude of the temperature at the surface and tshift is the coldest day of the
year. In Kusuda’s relation, the surface temperature at z=0 is assumed to be known and
follow a periodical variation with an annual frequency. Other models exist in which the air
temperature is assumed known instead of the surface temperature [21]. In this case, it was
shown by Badache at al [21] that the mean soil temperature To can be found from the air
mean temperature by the following expression:

1
To =  hrTamb  hrad Tsky   G  c hconv f (1  rh )  (3)
he  

where :

Tamb : mean air temperature


Tsky : mean sky temperature
hconv : mean air convection coefficient
hevap : mean evaporation coefficient
hrad : mean radiative coefficient
he  hconv  hevap  hrad
hr  hconv  rh hevap
G : mean solar radiation
g : mean ground solar absortion coefficient
rh : mean air relative humidity
c, f : empirical constant for the evaporation flux
More details on the numerical values of the constants can be found in [21] and [22]. In the
case whwere the meteorological data does not follow a simple sinusoidal pattern, it is
possible to decompose the signal into several Fourier terms where the previous description
would be the main mode [23].
2. Ground Models

6
Fig. 1: Horizontal heat exchanger

Fig. 1 gives an example of a simple piping in the ground exchanging heat with a heat
transfer fluid. Assuming a homogeneous ground, the solution is given by the heat
conduction equation:

1 T
  2T (4)
 t
T
T  x, y, 0   Ts (t ) , k  q p  t 
n r  rp

Due to the linearity of the heat equation, the solution is usually found from the
superposition principle. The last problem can be seen as the superposition of the following
problem:

Fig. 2: Horizontal heat exchanger

is replaced by the following equivalent problem:

7
Fig. 3: Horizontal heat exchanger

So that the final solution is:

T  t , x, y, z   1T  t , x, y, z   2T  t , z 
(5)
2
Using the Kusuda assumption, T is simply given by Eq. 2. As stated in the Introduction,
this temperature field can be evaluated differently, or even measured, the premise here is
that it is known. The main problem is to find the other solution. A well-known solution is
based on the assumption that the heat flux is the same for all pipes and uniform along the
pipe length. Using this assumption, early ground response factors [24] were found using the
theory of images, forcing the homogeneous boundary condition to be fulfilled. The solution
at the periphery of the first pipe for the case shown above would then be:

1 q   rp   z2  z1   2 z1   z2  z1  
T p1(t )  I  I  I   I  (6a)
2 ks   2  t   2 t   2 t   2  t  
where the last two terms represent the contribution of the two images. q’ is the heat transfer
rate per unit length of pipe, not the trench. Under the assumption of uniform heat flux, the
response factor I(X) is known to be the classical Infinite Line Source (ILS) solution [25]:

1 r
I X   E1( X 2 ) , X (6b)
2 2 t
And E1 is the first-order exponential integral.
The solution at the second pipe would be:

8
1 q   rp   z2  z1   2 z2   z2  z1  
T p 2 (t )  I  I  I   I  (6c)
2 ks   2  t   2 t   2 t   2  t  

The response factor of the two-pipe arrangement can then be approximated as


1
T p1 (t )  1T p 2 (t )
1
T p (t )   qRs
2
(7)
1   rp   z2  z1   z2  z1    z   z2   
Rs  I  I   I   0.5  I  1  I  
2 ks   2  t   2  t   2  t    t    t  
Several “soil resistances” are given in graphical form in [24] for different trench
configurations based on this principle. The mean fluid temperature is found from the pipe
resistance:

log  o 
R d
eq1
T f  1T p  qR p  q( Rs  R p ) where R p 
1
  di (8)
h d i 2 k p

The simplified assumption of a uniform heat flux along the pipes was discarded by
Claesson and Dunand [4], at least in the steady-state case for some simple configurations.
In their analysis, they assumed a linear variation of the temperature along the pipes and
found that for the case shown in Fig. 1:

 2 
2 
 2 R1R2  2 R12  2 
T f  q    qR eq 2 (9)
 R1  R2  2 R12 
 
where

1  2z  1  2z  1 z z 
R1  log  1   R p , R2  log  2   R p , R12  log  1 2  (10)
2  ks  rpo  2  ks  rpo  2  ks  z2  z1 
   

 L (11)
mC p

And
9
 R  R    R  R  
q1  2q  2  q2  2q  2 
2 12 1 12
(12)
 R  R  2 R   R  R  2 R 
 1 2 12   1 2 12 

The justification of using a steady-state analysis for horizontal ground heat exchangers is
sometimes justified knowing that the time scales in horizontal configurations are much
shorter than for vertical boreholes. In this latter case, Eskilson [26] showed that the time
scale for the steady-state regime was:

L2
tc,v  (13a)
9
whereas for the horizontal configuration, it is in the order of:

10 zo 2
tc,h  (13b)

This can be seen in Fig. 4, where the steady-state thermal resistance for a single pipe is
compared with the unsteady one derived from the ILS approach.

Fig. 4: Comparison between the unsteady and steady-state soil resistances for a single pipe for
zo/rp = 50

10
Fontaine et al. [19] extended the approach of Claesson and Dunand to the unsteady case. In
their approach, they could not solve the whole problem as a global unsteady thermal
response factor. Instead, they split the horizontal piping into segments. Mutual thermal
responses were evaluated with the finite line source (FLS) model. In their solution, the total
heat exchange through the heat exchanger varied with time, but they assumed that, at a
given time, the heat transfer distribution along the pipe followed the steady distribution
found with the Claesson and Dunand distribution. In this paper, an approach similar to the
one followed by Fontaine et al. is used, but no assumption is made regarding the heat
distribution along the pipe. Instead, this distribution is part of the solution. The model is
also based on the FLS model, but uses modified mathematical expressions of the FLS that
were initially solved in the context of vertical boreholes and applied here in horizontal
configurations [16, 17, 27]. The model also takes into account seasonal temperature
variations that have a significant influence on the thermal behavior of horizontal
geothermal fields. The model assumes, however, that the ground is homogeneous and
single-phase.

3. New Model for Steady-state Simulations


Before looking at the transient case, it is interesting to revisit the steady case. The simple
two-pipe case (Fig 1) will be studied here. The Claesson and Dunand solution was found,
assuming a linear variation of the temperature along the pipe. It is known that this is often a
valid assumption, but for long pipes with low flow rate, it may lead to errors. The exact
solution can be easily found in some simple configurations solving the following
differential equations system:

dT1 (T  T ) (T  T )
mC p  q1  o 1  2 1
dx R1 
R12
(14)
dT (T  T ) (T  T )
mC p 2  q2  o 2  1 2
dx R2 
R12
This system is similar to a typical U-tube model inside a vertical heat exchanger (see, for
example [28]), except that the undisturbed ground temperature replaces the borehole
temperature, and that the resistance associated with each pipe is now different. The 

11
formalism is used as a reference in this case. Those resistances can easily be found from the
values given by Eq. 10. In fact, we have:

(T1  To )  q1 R1 q2 R12


(15)
(T2  To )  q1 R12 q2 R2
Inverting the last system will give:

R R  R2 R R  R2 R R  R2
R1  1 2 12 , R2  1 2 12 , R12

 1 2 12 (16)
R2  R12 R1  R12 R12
The solution to Eq. 14 is given based on the dimensionless temperature:

T  To
  (17)
Tin  To

T1  To  x(1   2 )   cosh( (1  x))   m sinh( (1  x)) 


 exp    (18a)
Tin  To  2   cosh( )   m sinh( ) 
T2  To  x(1   2 )   cosh( (1  x))   m sinh( (1  x)) 
 exp    (18a)
Tin  To  2   cosh( )   m sinh( ) 
with:

x   
x , 1  , 2  , 12 
L   
R1 R2 R12
(19)
1   2
m  ,   m
2
 212 m
2
It is easy to verify that in the case of 1   2 , the classical Zeng’s profile [29] is obtained.
The exit temperature is then:

T (0)  To  cosh( )   m sinh( )


o  2  (20)
Tin  To  cosh( )   m sinh( )
From which an effective resistance can be found:

 (1 o )  R1  R2 


Req3    cotanh( ) (21)
(1  o )  2 
 
This is the classical result found in a U-tube borehole if both resistances are equal, but with
the difference that now the heat rate is per unit length of pipe, and not per unit length of
borehole.
12
3.1 Simulation results
A simulation using the finite element software COMSOL Multiphysics™ was done to
compare the different steady-state solutions for the case just described.

Fig. 5: Mesh generated for horizontal ground heat exchanger

The geometry is the same as the one in Fig 1. The domain was split in half to take into
account the symmetry of the problem. The parameters of the simulation are given in Table
1. No thickness was simulated on the pipe. Instead the convection coefficient represent and
equivalent transfer coefficient:

 do 
1 1  di 
 log   (22)
 do h  di hconv  2 k pipe 
 

Table 1. GHE configuration

rb L u z1 z2 Cp q ks h
(m) (m) (m/s) (m) (m) (MJ/m3-K) (W) (W/mK) (W/m2K)

13
0.02 100 0.3 2 3 3.93 1000 8 1000

The results obtained by the finite element simulation are shown in Fig. 6, where they are
compared with the analytic variation given by Eq. 18 and the linear variation, assuming a
soil resistance given by Claesson and Dunand (Eq. 9) and the classical method of images of
Bose et al. (Eq. 8). In the latter case, the thermal response is unsteady, and only the steady-
state solution is shown.

Fig. 6: Fluid temperature along the heat exchanger

The corresponding resistances are given in Table 2, and the corresponding inlet and outlet
temperatures in Table 3. A grid sensitivity was performed ans is shown in Fig. 7 where it is
seen that the

Table 2. Equivalent resistance (m-K/W)


Req1 (Eq. 8) Req2 (Eq. 9) Req3 (Eq. 21) Req, comsol
.145 0.163 0.167 0.169

Table 3. Inlet and Outlet temperatures


Req1 Req2 Req3 Req, comsol
Tfin Tfout Tfin Tfout Tfin Tfout Tfin Tfout
14
-1.08 -0.40 -1.15 -0.4 -1.17 -0.50 -1.18 -0.51

Fig. 7: Impact of grid refinement

4. Unsteady Analysis

4.1 Finite Line source model for horizontal pipes


The method of images described in the previous section is applicable for the unsteady case,
where it is again assumed that the heat flux is uniform along the pipe. In order to take into
account the variation of the heat flux, Fontaine et al. suggested splitting the pipe into
segments, with the heat transfer rate varying from segment to segment. At each time step,
the total heat transfer is assumed to be known, and the variation along the pipe is assumed
to vary exponentially. From the known heat transfer distribution along the pipe, the fluid
temperature variation is calculated with the finite line source model (FLS). In its most
general form, the FLS model is given [30] by:

15
(xi,i,i)
ground surface

(x,,) di

A1 (x,y,z)

d
A2

Fig. 8: Segment-to-segment temperature thermal influence.

t    d 2    di 2  
 q   4 (t  )   4 (t  )  
T12  t    dA2  dA1  e d
3/2 
e (23)
0   (t   ) 
8 k H2  
 
A common simplification is often done by integration over time. Doing so will result in:

 
 erfc  d  erfc  di
  
T12  t  
q  2 t    2  t  
4 k H 2  dA2  dA1 
 d di 
(24)
 
 
which is the form used by Fontaine et al. in their model. This model involves a double
integral, which has been simplified in the case of vertical boreholes by Lamarche and
Beauchamp [31] and by Claesson and Javed [27]. The latter solution is of particular
interest. The idea is to keep the time integral and invert the order of integration. In the case
of a vertical borehole, the solution becomes:
16
d H    r 2  ( z  )2    r 2  ( z  )2  
 q
H t
Tb  t    dz  d   e 4 (t  )   e 4 ( t  )   d  (25)
  (t   )   
3/2
8k H d d 0
 
with r 2  x2  y 2


q 2ierf ( Hs)2ierf ( Hs  2ds) ierf (2 Hs  2ds)  ierf (2ds)
Tb  t    e r s
2 2
ds (26)
4 k H 1 4 t
s2

1
with ierf ( x) x  erf ( x)  (1  e x )
2


Details of the derivation can be found in [17]. The generalization of the previous expression
for the influence of a vertical segment on another vertical segment was given by Cimmino
and Bernier [32]

Fig. 9: Vertical segment-to-segment thermal response factor

q q
T12  hFLS   (r , t , H1 , H 2 , d2  d1 )  (r , t , H1 , H 2 , H1  d1  d2 )  (27a)
2 ks 2 ks

 r 2 s 2
1 e ds
(r , t , H1 , H 2 , z ) 
2H 2  s 2
1
4 t
 ierf ((z H 2 ) s) ierf (z s)  ierf ((z  H1 ) s) ierf ((z  H 2  H1 ) s)  (27b)

The generalization of the segment-to-segment temperature response in the general inclined


case was studied by Lazarotto [33], where the author simplified the double integral
involved, but could not reduce it into a single integral in the general case. Although it may
17
not seem possible to simplify the thermal response in the general case, it can be done in the
horizontal configuration.

Fig. 10: Horizontal segment-to-segment thermal response factor

Looking at Fig. 10, it can be shown that the generalization of Cimmino’s expression is
given by (Appendix 1):

q q
T12, hor  hFLS ,hor ,1,2    (r1 , t , H1 , H 2 , x)   (r2 , t , H1 , H 2 , x)  (28)
2 k 2 k
where
r1  rp , segment on the same pipe section
r1   y1  y2   d1  d 2 
2 2
, otherwise

r2  2di , segment on the same pipe section


r2   y1  y2   d1  d 2 
2 2
, otherwise

4.2 Unsteady Temperature history


Knowing the response factor associated with a horizontal piping arrangement, it is possible
to calculate the time evolution of the average pipe temperature over a pipe segment. This
calculation assumes knowledge of the heat flux associated with every segment. Again, in
Fontaine’s model, the total heat transfer time history is known and the variation along the
pipe is assumed to follow an exponential variation. Since the heat transfer varies with time,

18
a convolution scheme has to be performed to estimate the temperature evolution. This
convolution can be very time-consuming, and Fontaine uses the Fast Fourier Transform
(FFT) approach proposed by Marcotte and Pasquier [34] to accelerate the convolution
evaluation. The FFT is a very powerful approach that allows the time domain convolution
product to be replaced by a simple product in frequency domain. I order to use the FFT
approach,the heat transfer history has to be known a priori, which is not a trivial factor,
especially in the case of horizontal piping, where the seasonal temperature variation can
interfere with the process. The proposed model follows the path already proposed by the
author in the case of mixed vertical boreholes [16], where any arrangements of vertical
borehole in series and in parallel can be simulated. The idea is to split a horizontal pipe into
several sections, each of which can be seen as a pipe in series with the previous section.
Referring again to the typical configuration shown in Fig 1, splitting the pipe into ns
segments can be analysed by the tools given, assuming ns pipes in series. Two major
differences have to be confronted in the case of horizontal systems: the response factors
have to be changed and the seasonal temperature variation has to be taken into account. The
first problem has been covered in the previous section, and the second one will now be
discussed. The description will be done with only one parallel path, but the general case is
given in [16]. It can be shown that the pipe temperature at a given time can be evaluated
with the one at a previous time step in a form given by:

ns
T p,i  S p,i   Sq,ij X j ( T fi, j  Tp, j ) (29)
j 1

where the expressions of S p,i , Sq,ij are given in [16] and reproduced in Appendix 2. We
also have:
qi  X i (T fi,i  T p,i )

 mC p i (30)
X (1  o,i )
Hi

where o,i represents the temperature variation along the segment:

T fo,i  T p,i
 o, i  (31)
T fi,i  T p,i

Assuming a linear variation, we have:


19
1  Yi Hi
 o, i  , Yi  (32)
1  Yi 2(mC p )i R p

Assuming an exponential variation, the result is:


o,i  exp(2Yi ) (33)
Again, these assumptions are for a single segment, not the entire pipe. In the case of np
parallel pipes, where all the inlet temperatures are known, Eq. 29 represents a system of np
equations for the np unknowns. For the case of ns series arrangement, only the first
temperature is known and the (na = ns-1) others are unknowns. The new equations that have
to be added to the system are:

T fi,i  T fo,i 1 o,i 1 T fi,i 1  Tp,i 1 (1  o,i 1)

A* ×T* = B* (34)

 A(ns  ns ) UR(ns  na )  BU  Tp 


A*    , B*    , T*    (35)
LL(na  ns ) LR(na  na )  BL  TL 
1  X1Sq,11 , X 2 Sq,12 ,... 
 
 X1Sq,21 , 1  X 2 Sq,22 , ....
A (36)
.... 
 
 

T fi,2 
 
TL  T fi,3  (37)
 
 
  X 2 Sq,12 ,  X 3Sq,13 ,... 
 
UR    X 2 Sq,22 ,  X 3Sq,23 ,... (38)
 
 
 (1  o,1) 0 0 ...
 
LL   0  (1  o,2 )  (39)
 
 

20
 1 0 0 0...
  1 0 0... 
LR  
o,2
(40)
 0  o,3 1 0... 
 
 

 S p,1  X1Sq,11 T fin,1 


 
BU   S p,2  X1Sq,21 T fin,1  (41)
 
 
T fi,1 
 
BL  0  (42)
 
 
Once the inlet temperatures are known, the heat transfer distribution is evaluated with Eq.
30, the Sp coefficients are updated (Eqs. A6-A7), and the scheme moves to the next time
step. In the case where the inlet temperature is not known, but the total heat transfer is
imposed, the system is modified:


q  (mC p ) T fi,1  T fo, ns  (43)

T fo, ns  T p, ns
o, ns  (44)
T fi, ns  T p, ns

q
T fi,1  o, ns T fi, ns  Tp, ns (1  o,ns )  (45)
mC p

A ×T = B (46)

 A(ns  ns ) UR(ns  ns )  BU  Tp 


A    , B    , T    (47)
LL(ns  ns ) LR(ns  ns )  BL  TL 
T fi,1 
 
TL  T fi,2  (48)
 
 

21
  X1Sq,11,  X 2 Sq,12 ,... 
 
UR    X1Sq,21,  X 2 Sq,22 ,... (49)
 
 
0 0 0 ...  (1  o, ns ) 
 
LL  0  (1  o,1)  (50)
 
 
 1 0 0 0...  o, ns 
 
 o,1 1 0 0... 
LR    (51)
0  o,2 1 0...
 
 

 S p,1 
 
BU   S p,2  (52)
 
 
q 
 mC p 
 
BL  0  (53)
 
 
 
5. Impact of Seasonal Temperature Variation
The effect of seasonal variation is quite simple when the heat flux is known, and less trivial
when it is part of the solution. The heat transfer on a pipe segment becomes:

T f , i  T p ,i T f ,i  (1Tp,i  2Tp,i ) T f ,i  1Tp,i


qi    (54)
Rp Rp Rp

where T  T  2T and 2T are given by Kusuda equation (Eq. 2) or any other model used
for the seasonal temperature variations. System of Eqs. 29 becomes:

qi  X i (T fi,i  1Tp,i ) (55)

ns
T p,i  S p,i   Sq,ij X j ( T fi, j  1T p, j )
1
(56)
j 1

22
which means that the results given in the previous section will give the pipe temperature
contribution 1T and the reduced fluid temperature. Particular attention is paid to the
complementary system:

T fi,i  o,i 1 T fi,i 1  T p,i 1 (1  o,i 1 )

T fi,i  2T1 p,i  o,i 1T fi,i 1  (1  o,i 1 )(1Tp,i 1  2Tp,i 1 )  2Tp,i (57)

T fi,i  o,i 1T fi,i 1  (1  o,i 1 ) 1T p,i 1  ( 2T p,i 1  2Tp,i )

The last term in Eq. 57 is zero if the two segments are at the same depth, but different from
zero otherwise. Eq. 45 should also be modified:
q
T fi,1  o, ns T fi, ns  1T p, ns (1  o,ns )   ( 2T p, ns  2T p,1)
mC p

where the last term is zero again if the inlet and outlet are at the same depth.

6. Simulation Results
The model just described was compared with a finite element simulation on COMSOL
Multiphysics. The geometry studied is the same as the one shown in Fig 1, and the grid is
the same as in the steady-state analysis.

Table 4. GHE configuration

rb L u z1 z2 (Cp)f ks (Cp)s h ns
(m) (m) (m/s) (m) (m) (MJ/m3-K) (W/mK) (MJ/m3-K) (W/m2K)
0.02 100 0.3 2 3 3.93 8 2.0 100 8

6.1 Inlet temperature imposed


In the first simulation, the inlet temperature will be imposed at the inlet of the upward pipe
section shown in Fig. 1. A linear profile as shown In Fig. 11 will be imposed during half of
the year, and a constant temperature is imposed for the rest of the year. In the numerical
simulation, a time-varying temperature profile given by Eq. 1 is imposed at the surface of
the domain (z = 0), with A = 10 and tshift = 0 (January 1st is the coldest day of the year). A
5-hour time step was chosen for the numerical simulation, while a 1- hour-time step is
taken during the analytic simulation. Eight segments (4 on each pipe section) were chosen

23
for the analytic simulation. Fig. 11 shows the outlet temperature variation during the year,
and Fig. 12, the total heat exchange with the ground.

Fig. 11: Fluid temperature during one year (test1).

Fig. 12: Total heat transfer during one year(test1).

Fig. 13 shows the temperature variation and Fig. 14 the heat transfer per unit length along
the pipe for two different simulation times: after half a year and after one year.

24
Fig. 13: Temperature variation along the pipe for two different times(test1).

Fig. 14: Heat transfer per unit length variation along the pipe for two different times (test1).

The impact of the number of segments was studied for this particular test. The number of
segments in each pipe section was varied and the difference between the total heat
exchange in the ground was compared. Ths difference is defined as:

25
Qanalytical   qanalytical (t ) dt
Qcomsol   qcomsol (t ) dt (58)

Qanalytical  Qcomsol
difference(%)  100
Qcomsol

between the analytical model and the numerical simuation was evaluated and the result is
shown in Fig. 15.

Fig. 15: Influence of the number of segments on the difference between analytical and numerical
solution.

6.2 Total heat transfer imposed


As a second test, the total heat flux was imposed to test the systems of Eqs. 46-53. A
constant heat extraction of 5000 W was imposed for the first part of the year (t < 4380
hours), while no heat exchange (restitution period) was imposed for the rest of the year. All
other parameters were the same as the one given in Table 4.
The inlet and outlet temperatures are shown in Fig. 16.

26
Fig. 16: Fluid temperature during one year (test2).

The temperature and heat flux along the pipe are now given just before the end of the heat
injection (t = 4375 h), and 100 hours after the power is turned off (t = 5380 h), and shown
in Fig. 16. The heat transfer per unit length is shown in Fig. 17. It can be seen that when the
total heat exchange is zero, the local heat transfer is not, due in part to the temperature
difference between the ground temperature around the pipe.

Fig. 17: Temperature variation along the pipe for two different times(test2).

27
Fig. 18: Heat transfer per unit length variation along the pipe for two different times (test2)

Table 5 CPU time


Test1 (Comsol) Test1 (Analytic) Test2 (Comsol) Test2 (Analytic)
18 min 2.2 sec 21 min 2.2 sec

Finally, the calculation times for both approaches are compared in Table 5, where an
important reduction in the calculation time is observed using the proposed model.
6.3 Comparison with simpler models
As stated in section 2, the ILS model was often used to generate unsteady soil resistances
for horizontal piping. In the case where the fluid temperature is imposed, it is not trivial to
apply the model but in the case where the heat transfer is imposed, superposition of the
solution given by the method of images and the seasonal variation is possible. For example
the second test analyzed in the last subsection can be compared with the method of images.
From Eq. 8:
1
T f  qRs  R p

With Rs given by Eq. 7 for the two-pipe arrangement. The inlet-outlet temperatures are
then

28
1 q q
T f ,o  1T f  , 1
T f ,o  1T f  (59a)
2mC p 2mC p

T f ,o  1T f ,o  2Tp,o , T f ,i  1T f ,i , 2Tp,i (59b)

Comparison of the inlet-outlet temperatures thus obtained is compared with the numerical
simulation in Fig. 19. A zoom in the vicinity of the end of the heat pulse is done to better
see the difference in the models.

Fig.19a: Outlet temperature given by Eq. 59 Fig.19b: Inlet temperature given by Eq.59
Although the general trend of the temperature evolution is correctly predicted with the ILS
model, larger variations are observed near the regions where the heat flux varies. Most of
the differences come from the assumption that the heat transfer per unit length is uniform.
The rigorous approach was described in the previous sections. However a simple variation
of the method of images using the ILS can be done if we allow the heat transfer to vary
from one pipe to the other. A possibility is to assume that, even in the unsteady case, the
heat transfer distribution resembles the steady case given by the Claesson and Dunand [4]
relation (Eq. 12). Rewriting those equations

 R  R    R  R  
q1  2q  2    q q2  2q  2    q
2 12 1 12
(60)
 R  R  2 R  1  R  R  2 R  2
 1 2 12   1 2 12 
The unsteady soil resistance given by Eq. 7 can now be rewritten as:

Rs*  1R1   2 R2 (61)


29
where R1 and R2 are given by Eq. 6. Replacing this new resistance in Eq. 59

Fig.20a: Outlet temperature given by Eq.62 Fig.20b: Inlet temperature given by Eq.62

q q
T f ,o  qRs*  R p 
1
, 1
T f ,o  qRs*  R p  (62a)
2mC p 2mC p

T f ,o  1T f ,o  2Tp,o , T f ,i  1T f ,i , 2Tp,i (62b)

would give the results shown in Fig. 20:

It is seen that even though, the unsteadiness is still not well predicted, the overall
comparison is better.

7. Conclusion

An analytical model based on a new formulation of the finite line source associated to
horizontal configurations was developed to simulate the heat transfer between a horizontal
heat exchanger and the surrounding ground. The model was compared to a finite element
simulation in the case of a simple configuration and display excellent agreement given that
the model is 500 0times faster than the numerical simulation. While the configuration may
30
be simple, it illustrates a very important aspect of horizontal systems, namely, different
local ground temperatures around pipes at different heights and how this can affect the
thermal behavior of the ground exchanger. The model can easily be extended to different
inline configurations, which can have parallel branches as well. Extensions to slinky or
spiral configurations can also be considered, but in that case, the thermal response factor
between pipe sections would be more complex.. An extension of the classical work of
Claesson and Dunand was also presented as part of this study. It is a future goal to use it to
provide potentially better guidelines for horizontal design procedures.

31
Acknowledgments

We would like to thank the Natural Sciences and Engineering Research Council of Canada
(NSERC), which partly financed this research (RGPIN-2014-06240).

Appendix 1
Referring to Fig. we have:

x  H 2 H1    r 2  (x x )2    rim 2  (x x )2  
t
 q 1   4 (t  )   4 (t  )  
Ti  j  t    dx     (t   ) 3/2 e
d x e d
8k H
x 0 0    
 
(A1)
where r represents the perpendicular distance between the source line and the surface of
interest. In the case where two segments are on the same pipe section, we then have:
r  rp , rim  2 zi

In the case where the two segments are in different pipe sections:

r 2  ( yi  y j )2  ( zi  z j )2 , 2
rim  ( yi  y j )2  ( zi  z j )2 (A2)

Using the usual change of variable s  1 leads to:


4 (t   )

 x  H 2 H1
q e r s  e ri s  d s dx  e s ( x x ) dx
2
Ti  j  t    
2 2 2 2 2 2

  (A3)
4 k H 2  1 x 0
4 t

 x  H 2
q er s  eri s  d s
Ti  j  t    
2 2 2 2

  s (erf ( sx)  erf ( s( x  H1) ) dx (A4)


4 k H 2 1
x
4 t

e r s  e ri s
2 2 2 2

q
Ti j  t     ierf ((x H 2 )s) ierf (x s) 
4 k H 2 1 s2 (A5)
4 t
ierf ((x  H1 ) s) ierf ((x  H 2  H1 ) s  ds

Appendix 2

32
t
t
rp2

1 Nz  zn 2 t
e
1 Nz
  
1  e zn t uij ( zn )zn
2

S p ,i  Fi (t , zn )zn , Sq,ij  (A6)


ks n 1 ks n 1

 
nserie
Fi (t  t , zn )  e zn t Fi (t , zn )   qj (t  t ) 1  e  zn t uij ( zn )
2 2

j 1 (A7)
Fi (t  0, zn )  0
z
uij ( z )   L 1(hFLS , hor ,ij ) (A8)

Details on the derivation of these expressions are given in [16]. The choice of the integral
variable z is discussed in [35].

References

[1] Lund JW, Boyd TL. Direct utilization of geothermal energy 2015 worldwide review. Geothermics.
2016;60:66-93.
[2] Raymond J, Malo M, Tanguay D, Grasby S, Bakhteyar F. Direct utilization of geothermal energy from
coast to coast: a review of current applications and research in Canada. Conference Direct utilization of
geothermal energy from coast to coast: a review of current applications and research in Canada.
[3] Rees S. Advances in ground-source heat pump systems. Duxford, UK: Woodhead Publishing is an imprint
of Elsevier, 2016.
[4] Claesson J, Dunand A. Heat extraction from ground by horizontal pipes. Swedish Council of building
Research. Stokholm; 1983.
[5] Mei VC. Horizontal ground-coil heat exchanger theoretical and experimental analysis. United States:
Report No. ORNL/CON-193, Dec, Oak Ridge National Laboratory, Oak Ridge; 1986.
[6] Piechowski M. Heat and mass transfer model of a ground heat exchanger: theoretical development.
International Journal of Energy Research. 1999;23(7):571-88.
[7] Kupiec K, Larwa B, Gwadera M. Heat transfer in horizontal ground heat exchangers. Applied Thermal
Engineering. 2015;75:270-6.
[8] Kayaci N, Demir H. Numerical modelling of transient soil temperature distribution for horizontal ground
heat exchanger of ground source heat pump. Geothermics. 2018;73:33-47.
[9] Congedo P, Colangelo G, Starace G. CFD simulations of horizontal ground heat exchangers: A
comparison among different configurations. Applied Thermal Engineering. 2012;33:24-32.
[10] Gan G. Dynamic thermal modelling of horizontal ground-source heat pumps. International Journal of
Low-Carbon Technologies. 2013;8(2):95-105.
[11] Soni SK, Pandey M, Bartaria VN. Ground coupled heat exchangers: A review and applications.
Renewable and Sustainable Energy Reviews. 2015;47:83-92.
[12] Xiong Z, Fisher DE, Spitler JD. Development and validation of a Slinkyb " ground heat exchanger
model. Applied Energy. 2015;141:57-69.
[13] Yoon S, Lee S-R, Go G-H. Evaluation of thermal efficiency in different types of horizontal ground heat
exchangers. Energy and Buildings. 2015;105:100-5.
[14] Fujii H, Nishi K, Komaniwa Y, Chou N. Numerical modeling of slinky-coil horizontal ground heat
exchangers. Geothermics. 2012;41:55-62.
33
[15] Esen H, Inalli M, Esen M, Pihtili K. Energy and exergy analysis of a ground-coupled heat pump system
with two horizontal ground heat exchangers. Building and Environment. 2007;42(10):3606-15.
[16] Lamarche L. Mixed arrangement of multiple input-output borehole systems. Applied Thermal
Engineering. 2017;124:466-76.
[17] Cimmino M, Bernier M. A semi-analytical method to generate g-functions for geothermal bore fields.
International Journal of Heat and Mass Transfer. 2014;70:641-50.
[18] Lamarche L. G-function generation using a piecewise-linear profile applied to ground heat exchangers.
International Journal of Heat and Mass Transfer. 2017;115:354-60.
[19] Fontaine P-O, Marcotte D, Pasquier P, Thibodeau D. Modeling of horizontal geoexchange systems for
building heating and permafrost stabilization. Geothermics. 2011;40(3):211-20.
[20] Kusuda T, Achenbach PR. Earth temperature and thermal diffusivity at selected stations in the United
States. National Bureau of Standards Gaithersburg MD; 1965.
[21] Badache M, Eslami-Nejad P, Ouzzane M, Aidoun Z, Lamarche L. A new modeling approach for
improved ground temperature profile determination. Renewable Energy. 2016;85:436-44.
[22] Mihalakakou G. On estimating soil surface temperature profiles. Energy and Buildings. 2002;34(3):251-
9.
[23] Jacovides C, Mihalakakou G, Santamouris M, Lewis J. On the ground temperature profile for passive
cooling applications in buildings. Solar energy. 1996;57(3):167-75.
[24] Bose JE, Parker JD, McQuiston FC, American Society of Heating R, Engineers A-C. Design/data
Manual for Closed-loop Ground-coupled Heat Pump Systems: American Society of Heating, Refrigerating,
and Air-Conditioning Engineers, 1985.
[25] Carslaw HS, Jaeger JC. Conduction of Heat in Solids. 2nd ed1959.
[26] Eskilson P. Thermal analysis of heat extraction boreholes1987.
[27] Claesson J, Javed S. An analytical method to calculate borehole fluid temperatures for time-scales from
minutes to decades. Conference An analytical method to calculate borehole fluid temperatures for time-scales
from minutes to decades, vol. 117. p. 279-88.
[28] Lamarche L, Kajl S, Beauchamp B. A review of methods to evaluate borehole thermal resistances in
geothermal heat-pump systems. Geothermics. 2010;39(2):187-200.
[29] Zeng H, Diao N, Fang Z. Heat transfer analysis of boreholes in vertical ground heat exchangers.
International Journal of Heat and Mass Transfer. 2003;46(23):4467-81.
[30] Lamarche L. Analytical g-function for inclined boreholes in ground-source heat pump systems.
Geothermics. 2011;40(4):241-9.
[31] Lamarche L, Beauchamp B. A new contribution to the finite line-source model for geothermal boreholes.
Energy and Buildings. 2007;39(2):188-98.
[32] Cimmino M. Fast calculation of the g-functions of geothermal borehole fields using similarities in the
evaluation of the finite line source solution. Journal of Building Performance Simulation. 2018:1-14.
[33] Lazzarotto A. A methodology for the calculation of response functions for geothermal fields with
arbitrarily oriented boreholes b. Renewable Energy. 2016;86:1380-93.
[34] Marcotte D, Pasquier P. Fast fluid and ground temperature computation for geothermal ground-loop heat
exchanger systems. Geothermics. 2008;37(6):651-65.
[35] Lamarche L, Beauchamp B. A fast algorithm for the simulation of GCHP systems. ASHRAE
Transactions. 2007;113(1):470-6.

34
List of Figures
Fig. 1: Horizontal heat exchanger
Fig. 2: Horizontal heat exchanger cross-section

Fig. 3: Horizontal heat exchanger equivalent problem

Fig. 4: Comparison between the unsteady and steady-steady soil resistance for a single pipe
for zo/rp = 50.

Fig. 5: Mesh generated for horizontal ground heat exchanger.

Fig. 6: Fluid temperature along the heat exchanger.

Fig. 7: Impact of grid refinement

Fig. 8: Segment-to-segment temperature thermal infleunce.

Fig. 9: Vertical segment to segment thermal response factor

Fig.10: Horizontal segment to segment thermal response factor.

Fig. 11: Fluid temperature during one year (test1)

Fig.12: Total heat transfer during one year(test1).

Fig.13: Temperature variation along the pipe for two different times (test1).

Fig.14: Heat transfer per unit length variation along the pipe for two different times (test1).

Fig. 15: Influence of the number of segments on the difference between analytical and
numerical solution.

Fig. 16: Fluid temperature during one year (test2).

Fig.17: Temperature variation along the pipe for two different times (test2).

Fig.18: Heat transfer per unit length variation along the pipe for two different times (test2).

Fig.19a: Outlet temperature given by Eq.59.


Fig.19b: Inlet temperature given by Eq.59.
Fig.20a: Outlet temperature given by Eq.62.
Fig.20b: Inlet temperature given by Eq.62.

35
Highlights

 New model for horizontal ground heat exchanger


 Analytical model which is very fast and accurate with respect to numerical
simulation
 A comparison with the classical trench model is performed.
 Particularity of horizontal systems are discussedPossible applications of the model
are discussed.

36

You might also like