Available online at www.sciencedirect.

com

Energy and Buildings 40 (2008) 1060–1066

www.elsevier.com/locate/enbuild

Numerical analysis and experimental validation of heat transfer

in ground heat exchangers in alternative operation modes
P. Cui a,*, H. Yang a, Z. Fang b
a
Renewable Energy Research Group, The Hong Kong Polytechnic University, Hong Kong, China
b
Shandong Jianzhu University, Jinan, China
Received 19 July 2007; accepted 1 October 2007

Abstract
A finite element numerical model has been developed for the simulation of the ground heat exchangers (GHEs) in alternative operation modes
over a short time period for ground-coupled heat pump applications. Comparisons between the numerical and analytical results show that the finite
line-source model is not capable of modeling the GHEs within a few hours because of the line-source assumption. On the other hand, the
experiments with respect to the alternative cooling and heating modes have been undertaken during a short-time period. The comparisons show a
reasonable agreement between the numerical and the measured data. The results illustrate that the finite element numerical model can be used to
simulate the heat transfer behavior of the GHEs in short time scales instead of the typical finite line-source model. Finally, the variation of the U-
tube pipe wall temperatures demonstrates that the discontinuous operation mode and the alternative cooling/heating modes can effectively alleviate
the heat buildup in the surrounding soil.
# 2007 Elsevier B.V. All rights reserved.

Keywords: Ground heat exchanger; Numerical heat transfer; Alternative operations

1. Introduction hot water heating according to different requirements of

owners. In summer, the system can be converted alternatively
In recent years, a large number of residential and into the heating mode for a few hours to produce hot water
commercial buildings have been installed with ground-coupled when the cooling requirement is insignificant or zero. In winter
heat pump systems (GCHPs) for space cooling, heating and or some transitional seasons when the cooling requirement is
even hot-water supply because of their higher efficiency, low unnecessary, the main function of the system is to satisfy the
maintenance cost and environmental friendliness. Most of the domestic hot water demand. Such systems usually run in at least
GCHPs use vertical ground heat exchangers (GHEs) which two operating modes during a short time period, which causes
usually offer higher energy performance than the horizontal that the heat transferred by the GHE to/from the surrounding
GHEs due to the less temperature fluctuation in the ground. soil varies accordingly. As a result, the corresponding
However, in hot or warm climates, the performance of the fluctuations are observed in temperatures of the surrounding
conventional GCHPs with vertical GHEs descends gradually soil and the heat carrier (water) to the heat pump, which directly
with operating time since a bulk of excess heat from cooling- impacts the operating performance of the GCHP system.
dominated buildings increases the ground temperature. Therefore, it is vital to design and analyze the thermal
One economical way to solve this problem is to employ the performance of the GHEs in short-time scales. For the last two
cooling/heating alternative operation modes in short time decades, a number of researchers have developed many design/
scales, which can be implemented through the triple-function simulation techniques including the empirical methods,
GCHP systems. The system can provide cooling, heating and detailed analytical or numerical approaches [1]. However,
the majority of the present approaches found in the literature are
inadequate in designing or simulating the GHEs in short-term
* Corresponding author. Tel.: +852 2766 7801; fax: +852 2774 6146.
operation. Most practical models use the Kelvin line-source
E-mail addresses: 04901782R@polyu.edu.hk (P. Cui), theory, in which the ground is regarded as an infinite medium
behxyang@polyu.edu.hk (H. Yang), fangzh@sdjzu.edu.cn (Z. Fang). and the borehole as an infinite line source [2,3]. This model is
0378-7788/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.enbuild.2007.10.005
 P. Cui et al. / Energy and Buildings 40 (2008) 1060–1066 1061

the two legs of the U-tubes, which introduces significant errors

Nomenclature in determining system short time-step behavior.
This present research is focused on the study of the transient
a thermal diffusivity (m2/s)
heat transfer behavior of the GHE in alternative operation
Fo Fourier number
modes in short-time scale. The numerical finite element method
H borehole depth (m)
(FEM) is used to analyze the temperature distribution in the
k thermal conductivity (w/m 8C)
borehole field.
ql heat flux per unit length of borehole (W/m)
r radius (m)
2. Model development
s space between the pipe center and the borehole
center (m)
The modeling of the transient heat transfer process in a
T temperature (8C)
GHE, especially in a short time period, is a complex
T0 ground far field temperature (8C)
mathematical problem. Simplifying assumptions are necessary
t time (h or s)
to obtain the solution. The ground is regarded to be
u angular coordinate
homogeneous in its thermal properties and initial temperature.
Q dimensionless temperature
The heat flux through the top and the end of the borehole is
Subscripts neglected because the size of the borehole diameter is much
b borehole smaller than its depth and the operation time concerned is very
down down pipe short. Based on the two assumptions, a two-dimensional
far far-field boundary transient heat conduction model is employed to represent to the
g grout real heat transfer process in the borehole domain.
in inside radius of the pipe
out outside radius of the pipe 2.1. Boundary conditions
p pipe
s surrounding soil For numerical heat transfer calculation, an outer boundary
up up pipe condition must be given to obtain the closed-form solution to
the heat transfer problem. The outer extent of the domain
defined here is large enough to make sure the boundary
temperature maintains consistently at the value of the far-field
proved to be valid only for operation periods of a few hours to a temperature over the concerned time, i.e., the amount of heat
few months because of the assumption of infinite line source. A flux at the outer edge of the domain is zero or insignificantly
major progress was made by Eskilson [4] to account for the small.
finite length of the borehole. In Eskilson’s model, the thermal The convective heat transfer between the water in the U-tube
capacitance of the borehole elements such as the pipe wall and and the pipe wall of the U-tube is not simulated in the model
the grout are neglected. Numerical computation with the finite- and the convective heat transfer resistance due to water flow
difference method is used on a radial-axial coordinate system to inside the U-tube pipes will be approximately considered in the
obtain the temperature distribution of a single borehole with conductive heat transfer of the pipes. Therefore, the inner
finite length. Zeng et al. [5] proposed the analytical solution for boundary condition at the inner pipe wall is set to be the heat
the finite line source model on the base of Eskilson’s approach. flux. It is noticed that the heat flux and fluid temperatures along
However, this model is not applicable to modeling or the two legs of the U-tube are actually varied due to the thermal
simulating the GHEs in short time intervals (usually less than interference between the two legs. Zeng et al. [9] proposed the
a few hours) because of the assumption of line source. quasi-three dimensional transient model which illustrates the
In recent years, a number of researchers have paid more fluid temperature variation along the flow direction in the two
attention to analyzing the effect of the short-term performance pipes of a U-tube. In the case of the project mentioned below,
of the GHEs using various numerical methods. Yavuzturk et al. about 60% of the total heat flux in a single borehole is
[6] have recently developed the short time step model for transferred through the down pipe and only 40% is released
vertical GHEs based on the finite volume numerical method. In through the up pipe when Fo ¼ at=rb2 ¼ 4:5 (about 6 h for a
this model, a ‘‘pie-sector’’ approximation of the circular pipe typical borehole). Fig. 1 illustrates the dimensionless fluid
geometry was implemented and the pipe wall heat flux temperature distribution in the down and up pipes of a U-tube in
boundary conditions were assumed to be the source terms cooling mode of the case.
because of the restriction of the code used. Muraya et al. [7]
developed a transient finite-element model of the heat transfer 2.2. Governing equations
around a vertical GHE with the aim of studying the thermal
interference between the adjacent legs of the U-tube. In order to simplify the analysis, a symmetrical disposal of
Sutton et al. [8] proposed the analytical algorithm to take the two legs of the U-tube inside the borehole is assumed. Then,
into account the short time behavior on the borehole with the only half of the borehole domain is modeled because of the
assumption of a single ‘‘equivalent diameter’’ pipe that replaces axisymmetric configuration, as shown in Fig. 2. An adiabatic
1062 P. Cui et al. / Energy and Buildings 40 (2008) 1060–1066

the pipe section:

Qi ¼ f ðr  ; u; Fo; kp ; ap ; kg ; ag ; rp;in

 
; rp;out ; s Þ (3)

 
where rp;in ¼ r p;in =r b ; rp;out ¼ r p;out =r b ; s ¼ s=r b ; rfar

¼
r far =r b s means the space between the pipe center and the
borehole center and rfar means the far-field boundary away from
the borehole center.

2.3. Finite element method

In this research, the numerical FEM is employed to solve the

problem and the commercial code ANSYS is used to perform the
transient numerical simulations of heat transfer in the borehole
domain. An important advantage of the FEM is the ability to
Fig. 1. Fluid temperature profiles of the two pipes along the borehole depth at
deal with the heat transfer process with arbitrary geometries
Fo = 4.5.
and non-homogenous media. After building the geometric
model and defining the necessary parameters including the
boundary condition is applied to the symmetric plane on the material thermal properties, the element size and type as well as
center of the borehole. The borehole domain is physically the boundary conditions, the ANSYS program can automatically
divided into three regions, the inner one is the pipe wall; the generate a finite element model which consists of nodes and
middle is the grout backfilled in the borehole and the outer elements, as shown in Fig. 2. The region nearest to the pipes,
region is the soil surrounding the borehole. where the temperature gradient is the steepest, is meshed finer
The governing equation for the each region in the borehole to enable the temperatures to be accurately predicted, as shown
domain can be represented with cylindrical coordinates. in Fig. 2(b). In contrast, a coarser mesh is sufficient for the
region outside the borehole. After the finite element model is
1 @T i @2 T i 1 @T i 1 @2 T i built with the mesh discretisation, the heat transfer problem can
¼ 2 þ þ (1)
ai @t @r r @r r 2 @u2 be solved numerically.
where the subscript i = p means the pipe; i = g denotes the
grout; i = s is the soil. 3. Comparison with finite line source method
Introducing the following dimensionless variables:
The finite line-source method in a semi-infinite medium has
 Ti  T0  r as t  ai been proved to be more accurate to describe the heat conduction
Qi ¼ 2pks ; r ¼ ; Fo ¼ 2 ; ai ¼ ;
ql rb rb as process of the vertical borehole for their long-term operation.
ki However, it is not capable of modeling or simulating the GHE
ki  ¼ in a short-time scale. In order to make a meaningful comparison
ks
between the finite element (numerical) method and the finite
Eq. (1) can be readily transformed into the nondimensionalized line source (analytical) method, a test case is established. All
expression: the necessary dimensionless variables including the geometric
1 @Qi  @2 Qi  1 @Qi  1 @2 Qi  parameters and the material thermal properties are given, as

¼ 2 þ  
þ 2 (2) shown in Table 1. The case was simulated respectively using the
ai @Fo @r r @r r @u2
numerical and analytical methods over a period of operation
Taking the boundary conditions into consideration for each time (Fo = 4e  2  4e + 2), which may be correspond to the
section, the dimensionless temperature rise in each region can time steps between about 4 min and 670 h for a typical
be determined by a number of dimensionless parameters.For borehole. It should be noted that the borehole depth H is only

Fig. 2. (a) Meshed model of the borehole domain. (b) Magnification of the mesh adjacent to the U-tube.
 P. Cui et al. / Energy and Buildings 40 (2008) 1060–1066 1063

Table 1 The relative error achieves the greatest value of nearly 100% at
Dimensionless parameters for the test case
the beginning of the simulation. The significant relative error
kp ¼ 0:25 ap ¼ 0:23 kg ¼ 0:8 ag ¼ 0:67 proves that the finite line-source method is not accurate enough
rp;in ¼ 0:24 r p;out ¼ 0:29 H/r = 2000 
rfar ¼ 73 for the simulation or design of the GHEs for short time
operation. However, after a few hours the deviation between the
two values is rapidly decreasing, which indicates that the
influence of the line-source assumption on the heat transfer
process is reduced gradually with time. The maximum relative
error is found to be less than 4% after the short-time of the
initiative period. This indicates that the finite line-source
method is completely acceptable for the practicable engineer-
ing except for the case of short-time simulation/design.

4. Experimental validation

4.1. Project description

To validate the numerical model for the heat transfer

simulation of the GHEs in short-time scale, operation data from
a pilot project was collected. The project was built to offer
Fig. 3. comparison of the numerical and analytical results. space cooling and domestic hot water for a house in Hong Kong
in 2006 for research purposes, as shown in Fig. 5. The system
considered in the analytical method although its effect is contains a water-to-water heat pump and a GHE that consists of
insignificant on the analytical solution in a short-time scale. two vertical and two inclined boreholes. The heat pump unit has
Fig. 3 illustrates the variations of the average dimensionless the cooling and heating capacity of 4.5 and 4.9 kW,
temperature rise on the borehole wall obtained from the respectively. A fan coil unit is located on the wall of the
numerical and analytical methods respectively. It can be seen house, which has three wind speeds. The major purpose of the
from Fig. 3 that the general variation trends of the numerical heating mode in this project is to produce domestic hot water
and analytical results show a good agreement during the since there is little space heating requirement for buildings in
operating time. To better analyze the difference between the Hong Kong.
two methods, the relative error which is defined in Eq. (4) is As stated above, the borehole field in the project consists of
also calculated and depicted in Fig. 4. two vertical and two inclined boreholes with a tilted angle of
208 arranged in a rectangular configuration. The distance
Analytical value  Numerical value
Relative error ¼  100% between two adjacent boreholes on the ground surface is about
Numerical value  Initial value
4 m. Each borehole has the diameter of 110 mm and the depth
(4)
of 30 m. The U-tubes are the high-density polyethylene pipe
It can be seen from Figs. 3 and 4 the analytical value of the with the outer diameter 32 mm (SDR 11). The horizontal
temperature rise on the borehole wall is markedly smaller than supply and return headers are buried at a depth of 1 m. It is
the numerical value at the beginning of the test case (generally noted that the purpose for drilling the inclined boreholes in this
within an hour for a typical borehole). This is due to the
assumption of the line source in the center of the borehole,
which neglects the physical size of the U-tube in the borehole.

Fig. 4. relative error profile between the numerical and analytical results. Fig. 5. An on-site photo of the project.
1064 P. Cui et al. / Energy and Buildings 40 (2008) 1060–1066

Fig. 7. Average heat transfer rate per unit length of borehole.

flux is observed in cooling mode, which had an average value of

about 60 W/m. However, the heat transfer rate in the heating
mode was approximated to be 45 W/m, which is due to the
Fig. 6. Distribution of temperature sensors along the U-tube. lower heat requirement for water heating.
The comparisons between the predicted and measured pipe
wall temperatures in the two chosen boreholes are presented in
project is to alleviate the thermal interference among them in Figs. 8 and 9. The relative errors between the predicted and
the ground while occupying less land area on the ground surface
than the common vertical GHEs [10]. However, the advantages
of the inclined boreholes will exhibit at least after one year’s
operation; therefore, it is not discussed in this paper.
The local average ground temperature a few meters below
the ground is about 22 8C during the test period. The thermal
properties of the local ground were estimated according to the
geological report offered by the drilling contractor. The Pt100
temperature sensors with an accuracy of 0.28C were bounded
tightly on the U-tubes. The two vertical boreholes (noted as B1
and B2) were chosen for the experiments. Fig. 6 shows the
distribution of the temperature sensors along the U-tubes in the
two boreholes.
A series of experiments were conducted under the
alternative cooling and heating modes during May 17th to
18th in 2007, as shown in Table 2. Fig. 8. Comparison of the predicted and measured up pipe wall temperatures.

5. Results and discussion

Fig. 7 presents the average heat transfer rate per unit length
of the borehole with operating time, where the heat rejection is
shown as positive and heat extraction as negative. A higher heat

Table 2
Operating schedule of the system during the experiments
No. Operating Time Operating mode
I 17th 11:00–14:27 Cooling mode
II 17th 14:32–14:58 Heating mode
III 17th 15:00–16:54 Cooling mode
IV 17th 16:55–18th 8:59 Off-time
V 18th 9:00–10:55 Cooling Mode
VI 18th 11:07–11:39 Heating mode
Fig. 9. Comparison of the predicted and measured down pipe wall tempera-
VII 18th 12:44–16:00 Cooling Mode
tures.
 P. Cui et al. / Energy and Buildings 40 (2008) 1060–1066 1065

average values of COP in cooling and heating modes are

obtained to about 3.5 and 4, respectively. It can be noticed
from Fig. 11 that the system performance is significantly
increased through the alternations of cooling and heating
modes. The higher COP in heating mode, which contributed
to the relatively higher ground temperature, also indicates that
the GCHP system is favorable for hot water heating in hot
climates.

6. Conclusions

A finite element numerical model has been developed for the

simulation of the GHEs in alternative operation modes over a
short time period. Comparisons between the numerical and
Fig. 10. Relative errors between predicted and measured pipe wall tempera- analytical results show that the finite line-source model is not
tures. capable of modeling the GHEs within a few hours because of
the line-source assumption.
A pilot project of the GCHP system with DHW supply has
measured temperatures are also presented in Fig. 10. The been developed for research purposes. The experiments with
predicted and measured temperatures on the two U-tubes show, respect to the alternative cooling and heating modes have been
on the whole, a great agreement during the operating time. It undertaken during a short-time period. The comparisons show a
should be noticed that the temperature deviations in the two reasonable agreement between the numerical and the measured
down pipes are obviously larger than that in the up pipes. This data. This illustrates that the finite element numerical model
phenomenon may be caused by several uncertain factors, such can be used to simulate the heat transfer behavior of the GHEs
as the heat flux sharing ratio between the two pipes of the U- in short time scale instead of the typical finite line-source
tubes or the temperature sensor sensitivity. The discrepancies model.
between the predicted and measured temperatures is in the Finally, the variation of the U-tube pipe wall temperatures
range of 6.5%. illustrates that the discontinuous operation mode and the
On the other hand, it can be detected from Figs. 8 and 9 that alternative cooling/heating modes can effectively alleviate the
the pipe wall temperatures oscillated significantly during the heat buildup in the surrounding soil. The operation modes can
cooling and heating alternative modes and then decreased ultimately improve the system performance. Therefore, the
gradually after the system was shut down. After 12 h of off-time discontinuous operation mode (such as operating during
at night, the pipe wall temperature was almost close to the daytime while shut down at night or vice versa) is also
initial value before the system started. This means both the recommended and it is feasible for commercial or residential
alternative modes and the discontinuous operation mode can buildings in this area.
significantly increase the heat transfer efficiency of the GHE
and thereby the performance of the whole system. Acknowledgements
Fig. 11 illustrates the variation of the COP with the
operation time during the periods, where the cooling COP is The work described in this paper is supported by a grant
noted as positive and the heating COP is negative. The from the Research Grants Council of the Hong Kong Special
Administrative Region, China (Project No. PolyU 5314/06E).

References

[1] J.D. Spitler, Ground-source heat pump system research past, present and
future, HVAC and R Research 11 (2) (2005) 165–167.
[2] L.R. Ingersoll, O.J. Zobel, A.C. Ingersoll, Heat Conduction with Engi-
neering, Geological, and Other Applications, McGraw-Hill, New York,
1954.
[3] D.P. Hart, R. Couvillion, Earth Coupled Heat Transfer, Publication of the
National Water Well Association, 1986.
[4] Eskilson P, Thermal analysis of heat extraction boreholes, Doctoral
Thesis, Department of Mathematical Physics, University of Lund, Lund,
Sweden, 1987.
[5] H.Y. Zeng, N.R. Diao, Z.H. Fang, A finite line-source model for boreholes
in geothermal heat exchangers, Heat Transfer Asian Research 31 (7)
(2002) 558–567.
[6] Yavuzturk Cenk, D. Spitler Jeffrey, J. Rees Simon, A Transient two-
Fig. 11. Variation of system COP versus operation time. dimensional finite volume model for the simulation of vertical U-tube
1066 P. Cui et al. / Energy and Buildings 40 (2008) 1060–1066

ground heat exchangers, ASHRAE Transactions 105 (2) (1999) 465– in a stratified geological regime, ASHRAE Transaction 108 (2) (2002)
474. 177–184.
[7] N.K. Muraya, D.L. O’Neal, W.M. Heffington, Thermal interference of [9] H.Y. Zeng, N.R. Diao, Z.H. Fang, Efficiency of vertical geothermal heat
adjacent legs in a vertical U-tube heat exchanger for a ground-coupled heat exchangers in ground source heat pump systems, Journal of Thermal
pump, ASHRAE Transactions 102 (2) (1996) 12–21. Science 12 (1) (2003) 77–81.
[8] M.G. Sutton, R.J. Couvillion, D.W. Nutter, R.K. Davis, Analgorithm for [10] P. Cui, H. Yang, Z. Fang, Heat transfer analysis of ground heat exchangers
approximating the performance of vertical bore heat exchangers installed with inclined boreholes, Applied Thermal Engineering 26 (2006) 1169–1175.