Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 108 (2014) 340–347
www.elsevier.com/locate/solener

Analysis of the convective heat exchange effect on the

undisturbed ground temperature
Mohamed Ouzzane a,⇑, Parham Eslami-Nejad a, Zine Aidoun a, Louis Lamarche b
a
CanmetENERGY Natural Resources Canada, 1615 Lionel Boulet Blvd., P.O. Box 4800, Varennes, Québec J3X1S6, Canada
b
Department of Mechanical Engineering, École de technologie supérieure, 1100 Rue Notre-Dame ouest, Montréal, Québec H3C1K3, Canada

Received 12 February 2014; received in revised form 17 June 2014; accepted 15 July 2014

Communicated by: Associate Editor Ruzhu Wang

Abstract

The ground temperature is an important parameter for several applications such as ground source heat pumps, agricultural green-
houses and ground energy storage systems. This paper describes a numerical model based on 1D transient heat conduction equation,
using the energy balance on the soil surface as a boundary condition. The absorbed solar radiation by the soil, the convection heat trans-
fer between the soil and the ambient air, as well as the long wave radiation exchange between the soil and the sky have been considered.
An hourly simulation over a whole year (8760 h) with 1 h time step has been conducted using real meteorological data including global
solar radiation, ambient dry bulb and dew point temperatures as well as the wind velocity. The model has been validated against mea-
surements and analytical calculations for a site located in Montreal (Canada). The model is applied to investigate the effect of convective
heat flux, calculated using three different correlations on the deep ground temperature for different climates. It has been found that in
general, McAdams (1954) and Kusuma (2004) correlations can be used for different climates with relatively good agreement between
measurements and calculations.
Ó 2014 Crown Copyright and Elsevier Ltd. All rights reserved.

Keywords: Ground temperature; Convection; Heat conduction; Undisturbed ground temperature; Solar radiation

1. Introduction the ground temperature about a mean value, as a function

The prediction of the ground temperature profile is
required in various energy applications such as: ground
source heat pumps, energy storage, heating and cooling
of buildings and agricultural greenhouses. Several works
have been done to find a function describing the soil tem-
perature profile. Among all correlations, the most popular
one is that proposed by Kasuda and Archenbach (1965).
At each certain depth, it gives a sinusoidal variation of
⇑ Corresponding author. Tel.: +1 450 6524636.
E-mail addresses: Mohamed.ouzzane@nrcan.gc.ca (M. Ouzzane),
Parham.eslaminejad@nrcan.gc.ca (P. Eslami-Nejad), Zine.aidoun@
nrcan.gc.ca (Z. Aidoun), Louis.lamarche@etsmtl.ca (L. Lamarche).
of time. The amplitude of the variation decreases with
depth at a rate that depends on the properties of the soil
until it gives a single value for deep ground temperature.
This correlation is used by several softwares such as
TRNSYS (2005). Numerous other methods have been pro-
posed to simulate the soil temperature, including numerical
methods (Herb et al., 2008; Zoras et al., 2012;
Mihalakakou et al., 1995), analytical and semi-empirical
methods (Lin, 1980; Droulia et al., 2009; Al-Temeemi
and Harris, 2001 and Elias et al., 2004), purely empirical
methods (Zheng et al., 1993) and statistical methods that
employ intelligent algorithms such as the neural network
algorithm (Tabari et al., 2011). It should be noticed that
most of these analytical and numerical methods are not

http://dx.doi.org/10.1016/j.solener.2014.07.015
0038-092X/Ó 2014 Crown Copyright and Elsevier Ltd. All rights reserved.
 M. Ouzzane et al. / Solar Energy 108 (2014) 340–347
Nomenclature
a thermal diffusivity (m2/s)
CH bulk coefficient of the globe
Cp specific heat at constant pressure (J/kg K)
g gravitational acceleration (m/s2)
h convective heat transfer coefficient (W/m2 K)
t time (s)
T temperature (K)
V velocity (m/s)
z depth coordinate (m)
Greek symbols
a absorption coefficient
e emissivity
k thermal conductivity (W/m K)
m kinematic viscosity (m2/s)
q density (kg/m3)
always able to provide reliable and actual prediction of the
ground temperature distribution because of difficulties in
the accurate determination of the ground surface boundary
conditions and the actual thermal properties of the soil.
Other several works have been done based on the energy
balance on the ground surface (Khatry et al., 1978;
Cellier et al., 1996; Mihalakakou et al., 1997;
Mihalakakou, 2002; Thiers, 2008 and Okada and
Kusaka, 2013). An explicit expression of the ground tem-
perature as a function of time and depth was derived by
Khatry et al. 1978, based on the solar radiation and the
atmospheric temperature. The equation was used to inves-
tigate the daily and annual variation of the ground temper-
ature in Kuwait. Mihalakakou et al. (1997) and Herb et al.
(2008) investigated the effect of the land on the surface tem-
perature by using the energy balance as a boundary condi-
tion on the ground surface. Different types of land were
considered: bare soil, short and tall grass, a forest and
two agricultural crops (corn and soybeans). Thiers (2008)
and Mihalakakou et al. (1995) studied the thermal interac-
tion between a building and the ground.
In practice, using simple semi-empirical equations with
measurements of the ground surface temperature evolu-
tion at a given location is the most popular and credible
method for calculating the average ground temperature
at a given depth and a day of the year. However, this
method cannot be used everywhere due to the fact that
the measurement of the ground surface temperature is
not available in meteorological data of the site. Indeed,
the last category of studies in which the energy balance
is applied as a boundary condition at the ground surface
is applicable anywhere due to the availability of the
required data. However, it constitutes more modeling
complexities. As noticed by different studies, the
341
r Stefan-Boltzmann constant = 5.67  108
2 4
(W/m K )
U_ density of heat flux (W/m2)
Subscripts
air air
amb ambient
c convective
dp dew point
g ground
gs ground surface
net net
r radiative
sky sky
wind wind
boundary condition at the ground surface consists of
three terms including solar radiation, heat losses to the
cold sky by long wave radiation and convective heat
transfer between the ambient air and the soil surface.
The first two terms with all related coefficients are formu-
lized in different references. However several correlations
have been proposed to calculate the convective heat trans-
fer coefficient that they all have not been developed for
heat transfer between the ambient air and the soil surface.
Since the convective heat exchange between the air ambi-
ent and the soil surface constitutes an important portion
of the overall energy balance at the ground surface, it is
important for designers and scientists to be aware of the
impact on their calculations. This may give a good sight
to choose the appropriate correlation for their works.
For example, Mihalakakou et al. (1997) and Thiers
(2008) used Mostrel and Givoni’s correlation (1982) while
Deru (2003), Duffie and Beckman (2006) and Lee and
Strand (2006) used McAdams correlation (1954).
Palyvos (2008) also reviewed a large number of convec-
tive coefficient correlations with a linear, a power law and
boundary layer form. He presented the conditions under
which the correlations have been produced. Rabadiya
and Kirar (2012) presented a comparative assessment of
various correlations and developed an improved equation
in the form of experimentally validated correlation for
wind loss coefficient. Most of the works cited previously
have been conducted for a finite surface, while Kroger
(2002) study is among a few works developed specially
for an infinite horizontal surface. Based on experimental
data, this study developed a correlation for convective heat
transfer coefficient between the natural environment and
an infinite horizontal surface subjected to a constant tem-
perature or constant heat flux.
342 M. Ouzzane et al. / Solar Energy 108 (2014) 340–347
The purpose of the present work is to analyse the effect
of the convective heat flux on the ground temperature pro-
file using a validated numerical transient model. Three dif-
ferent correlations for the forced convective heat coefficient
calculation are compared.
2. Theoretical model
The surface of the ground is heated by solar radiation
when the sky is clear. The surface also loses heat to the cold
sky by long wave radiation. By conduction, the heat is
transferred to the lower soil layers, neglecting water move-
ment in the soil. On the surface, both sensible and latent
heat transfer, occur. In the proposed model, the latent heat
part is neglected and heat transfers from the surface to the
ambient air by convection. Temperature changes in the soil
are essentially driven by transient one dimensional heat
conduction. The medium is considered isotropic and
homogenous with uniform properties. The heat diffusion
equation is given by:
2 2006) in degree Kelvin as a function of dew point temper-
1 @T @ T
¼ 2
a @t @z
where T is the soil temperature (K), t the time (s), z the
depth (m), and a, is the thermal diffusivity (m2/s).
The calculation domain and the different heat exchanges
are presented in Fig. 1. A sufficient depth of 30 m was
chosen to meet the adiabatic conditions in the lower limits
(Eq. (3)). The initial and boundary conditions are
expressed by the following equations:
Fig. 1. Schematic representation of the problem.
T ðz; t ¼ 0Þ ¼ T 0 0 6 z 6 30 m ð2Þ
The heat flux is assumed to be positive if directed down-
ward in the ground and negative when is directed upwards.
@T ðz ¼ 30 m;tÞ
¼0 t>0 ð3Þ
@z

@T 
kg   ¼ U_ net t > 0 ð4Þ
@z z¼0 m
U_ net ¼ U_ solar  a þ U_ c þ U_ r ð5Þ
U_ c þ U_ r ¼ hc ðT amb  T gs Þ þ e:r:ðT 4sky  T 4gs Þ ð6Þ

@T 
kg :  ¼ hc ðT amb  T gs Þ þ U_ solar  a  hr ðT gs  T sky Þ
@z z¼0m
ð7Þ
T sky ¼ T amb ½0:711 þ 0:0056  T dp þ 0:000073  T 2dp
þ0:013 cosð15  tÞ
0:25
ð8Þ
Eq. (8) gives the sky temperature (Duffie and Beckman,
ð1Þ ature (Tdp), dry ambient temperature (Tamb) and hour from
midnight (t).
Tsky and Tamb are in degrees Kelvin and Tdp is in degrees
Celsius.
The diffusion equation was solved by numerical method
(CFD), using PHOENICS commercial software. The diffu-
sion equation (Eq. (1)) was discretized and integrated,
using the control volume method. The calculation domain
was divided into 180 small control volumes along the z
direction. The grid distribution employed is non-uniform,
providing a closely spaced mesh in region of pronounced
temperature variation near the ground surface.
Starting with a uniform temperature in the entire
domain, calculations were performed for a duration of over
50 years with a time step of one hour , until a perfect global
convergence is reached. The same typical meteorological
data were successively applied on the ground surface for
each simulation year. The results then give the typical
ground temperature profile, including the deep ground
temperature (undisturbed ground temperature).
The convective heat transfer coefficient (Eqs. (10)–(13)
and (15)) has an important effect on the ground tempera-
ture as a solution of the diffusion Eq. (1). This parameter
depends essentially on the wind velocity. According to
the wind velocity and the temperature difference between
the ground surface and the ambience, three different modes
of convection heat transfer are possible: natural, forced
and mixed convection (Table 1). When the ambient air is
warmer than the ground surface (Tgs  Tamb < 0) the heat
flux is assumed to be positive (flux is downward) and the
energy goes inside the ground. When the ambient air is
colder than the ground surface (Tgs  Tamb > 0) the heat
flux is assumed to be negative (flux is upward) and the
energy goes from the soil to the ambient air. When there
is no wind (Vwind = 0) and the Tgs  Tamb < 0, the heat is
transferred only by conduction and the convective heat
 M. Ouzzane et al. / Solar Energy 108 (2014) 340–347 343

Table 1
Different possible cases of heat transfer between the ground surface and ambience.
Case number Tgs  Tamb Vwind (m/s) Heat transfer mode Heat flux direction Duration per year (h)
1 Tgs  Tamb > 0 0 Natural convection Negative heat flux 115
2 Tgs  Tamb < 0 0 Conduction Positive heat flux 569
3 Tgs  Tamb > 0 0 < Vwind 6 2.0 Mixed convection Negative heat flux 209
4 Tgs  Tamb < 0 0 < Vwind 6 2.0 Forced convection Positive heat flux 728
5 Tgs  Tamb > 0 Vwind > 2.0 Forced convection Negative heat flux 2301
6 Tgs  Tamb < 0 Vwind > 2.0 Forced convection Positive heat flux 4838

transfer coefficient is set equal to zero. For Tgs  Tamb < 0 2.2. Mixed convection
and 0 < Vwind 6 2.0 (case number 4 in Table 1) it is
assumed that the heat is transferred mostly by forced con- The mixed convection is considered when the wind
vection. According to the values in the last column of the velocity is low and the magnitude of the natural convection
Table 1, the forced convection mode is dominant with is similar to that of forced convection. The density of the
7867 h per year which represents 89.8% of the total period convective heat flux is calculated by the following equation
for the site of Montreal. Several correlations of the convec- (Okada and Kusaka, 2013):
tive heat transfer coefficient are available in the literature.
U_ c ¼ qair  C pair  C H  V wind  ðT gs  T amb Þ ð14Þ
Based on tests and experiments, these correlations pro-
duced as a function of wind speed. There are correlations The following formulation is used to calculate the
derived from wind tunnel measurements or model studies exchange coefficient of heat CH. It is proposed by Kondo
on relatively small plates and bluff bodies obstacles. Some and cited by (Okada and Kusaka, 2013).
others were developed using full scale field data measured
on actual building facades and roofs. C H ¼ 0:0065  ðT gs  T amb Þ1=3 for V wind 6 2 m=s ð15Þ

2.1. Forced convection 2.3. Natural convection

In forced convection regime, the three following correla- When there is no wind and the temperature difference
tions have been used to calculate heat transfer coefficient: between the ground and the ambient air is positive the nat-
ural convective heat flux is calculated by the following
2.1.1. Kusuma correlation equation (Kusuma, 2004):
The convective heat transfer calculated using Kusuma  1=3
g  aair
correlation (2004) was obtained for ground surface and U_ c ¼ qair  Cpair  C s ðT gs  T amb Þ4=3 ð16Þ
air. The wind velocity is measured at 10 m from the ground mair  T amb
surface. where Cs = 0.2 is a constant given by Townsend and lies in
the range 0.1–0.24 according to Deardorff and Willis
U_ c ¼ qair :C pair :C H :V wind :ðT gs  T amb Þ ð9Þ
(1985), aair, is thermal diffusivity and mair is the kinematic
The non dimensional exchange coefficient of heat CH is viscosity.
calculated as:
C H ¼ 0:025:V 0:7
wind for 0:5 m=s 6 V wind 6 7 m=s ð10Þ

2.1.2. Givoni correlation

Based on the analysis of various experimental data
involving exposed and wind-screened radiator, Mostrel
and Givoni (1982) proposed the following formula for
convective heat transfer coefficient:
hc ¼ 0:5 þ 1:2V 0:5
wind ð11Þ

2.1.3. McAdams correlation

McAdams correlation (1954) was derived from
experimental data obtained in a wind tunnel facility for
horizontal plate.
hc ¼ 5:7 þ 3:8V 0:5
wind for V wind 6 4:88 m=s ð12Þ
0:78
Fig. 2. Convective heat transfer coefficient versus the wind velocity for the
hc ¼ 7:2V wind for V wind > 4:88 m=s ð13Þ three selected correlations.
344 M. Ouzzane et al. / Solar Energy 108 (2014) 340–347
Fig. 2 presents the evolution of the convection heat
transfer coefficient versus the wind velocity for the three
different correlations with forced convection. Givoni corre-
lation gives the lowest value of the convective heat transfer
coefficient with the lowest slope representing the smallest
effect of wind. Kasuma provides convective coefficient
values which are significantly higher than those given by
Givoni. From a wind velocity of 2 m/s, Kasuma and
McAdams correlations demonstrate nearly similar slopes.
2.4. Analytical correlation for the ground temperature profile
Among the familiar correlations used to calculate the
ground temperature profile is that proposed by Kasuda
and Archenbach (1965). This correlation (Eq. (17)) is com-
monly used by several commercial software, such as
TRNSYS (2005), DOE-2 (1982) and RETScreen (2005).
It gives the ground temperature as a function of time of
the year and the depth below the ground surface. It has
been used in this study for validation.
 rffiffiffiffiffiffiffiffiffiffiffiffiffi " rffiffiffiffiffiffiffiffiffi#!
p 2p
T ¼ T z¼0  As  exp Z   cos tnow  tshift 
365  a 365
where T z¼0 is the mean surface temperature, As is the
amplitude of the surface temperature, tshift is the day of
the year which corresponds to the minimum surface tem-
perature and tnow is the current day of the year.
3. Validation with Kasuda correlation
As mentioned before, the diffusion Eq. (1) has been
solved using the CFD method. Boundary conditions
(Eqs. (3) and (7)) and the initial condition (Eq. (2)), have
been applied with the ground characteristics and the
weather data of Montreal, Canada. McAdams equation
is used to calculate the convection heat transfer. The
ground surface temperature profile obtained by CFD is fit-
ted by the equation below:
 
2p
T z¼0 ¼ 9:5  19:0  cos ½tnow  20 ð18Þ
365
where T z¼0 ¼ 9:5  C, As = 19.0 °C and tshift = 20 days.
Then, this data was used in Kasuda Eq. (17) to calculate
the ground temperature profiles for the spring and autumn
equinoxes. Fig. 3 shows the comparison between results of
CFD and that of Kasuda correlation. As presented in this
figure the agreement is excellent.
4. Analysis and discussion
4.1. Effect of the convection heat transfer on the ground
temperature
Eqs. (9)–(16) have been selected for the calculation of
the forced, mixed and natural convective heat transfer
Fig. 3. Ground temperature profile: comparison between CFD and
Kasuda correlation using the same ground surface temperature calculated
by the CFD model.
coefficients which are applied to calculate the ground tem-
Z 365 perature for five different sites. In this section, we are inter-
2 pa
ested to deep ground temperature or undisturbed ground
ð17Þ temperature. The selected cities present different climates
from: cold, Montreal in Canada to warm, Kiln in USA.
The meteorological data, the soil characteristics as well as
the undisturbed ground temperature for the five sites are
given in Table 2. The measured undisturbed ground tem-
perature data was obtained from Thermal Response Test
(TRT) reports at five locations. A TRT is an in-situ mea-
surement method used to determine the thermal properties
of the ground (the effective thermal conductivity of the
ground, the thermal resistance of the borehole and undis-
turbed ground temperature). The test consists of a constant
heat injection in the ground (50–80 W per meter of depth)
by circulating fluid (usually water) inside HDPE (High
Density Polyethylene) U-tube during 36–72 h approxi-
mately. The pipe size (3/4–1½ in.) and the borehole depth
(50–250 m) vary depending on the project design as the test
borehole is used with other boreholes for the project. The
evolution of the transient temperature of the fluid at the
inlet and at the outlet of the borehole is measured as well
as the flow rate and the energy injected. The theoretical line
source model is used for the analysis and the calculation of
the thermal properties of the ground.
As shown in Table 2, Montreal is the coldest site where
the annual average solar energy is the highest after Kiln
and Oklahoma. The two European cities; Amsterdam
and Brussels have almost the same annual average ambient
temperature with stronger wind and higher solar radiation
for Amsterdam.
The comparison between the results of three correlations
shows that Givoni gives the highest deep ground tempera-
ture prediction, especially for warmer climates (Oklahoma
and Kiln) where the average ambient temperature is
15.1 °C and 19.5 °C respectively. This is due to the fact that
the convective heat transfer is underestimated by the Givoni
 M. Ouzzane et al. / Solar Energy 108 (2014) 340–347 345

Table 2
Meteorological data, physical properties of the ground and undisturbed ground temperature of five different sites.
Montreal Amsterdam Brussels Oklahoma Kiln (USA)
(Canada) (Netherlands) (Belgium) (USA)
Thermal conductivity (W/m K) 2.65 1.88 1.86 3.42 2.07
Thermal diffusivity (m2/s)  10+6 1.10 0.81 0.76 1.38 0.82
Annual average ambient temperature (C) 6.3 10.02 10.24 15.1 19.5
Annual average wind velocity (m/s) 4.1 5.35 4.35 5.4 3.1
Annual global solar energy density (kW h/m2/ 1406.6 982.49 918.15 1733.36 1677.7
year)
Undisturbed ground temperature (C) 9.4 13.8 12.0 17.2 21.7
(measurement)

correlation as shown in Fig. 2. As shown in Fig. 4, the

ground temperature for Amsterdam is higher than Brussels
due to the higher solar radiation, whereas the ambient tem-
perature of Brussels is slightly higher. In order to have bet-
ter understanding of what causes the differences between
measured deep ground temperature values and calculated
ones, the contribution of convection (positive and negative
parts), solar radiation and long wave radiation flux for Giv-
oni, McAdams and Kusuma convection correlations is pre-
sented separately in Figs. 5–7 respectively. It is obvious that
since convection heat transfer from the ground surface can
change the surface temperature, it has an impact on the cal-
culated values of other terms such as the radiation to sky.
The solar heat flux is always positive, representing the heat
gain by the ground, while the long wave radiation heat flux
is always negative, indicating heat loss from the ground. On
the contrary, the convective flux can be positive or negative Fig. 5. Annual energy density on the ground: solar, long wave radiation
with sky and positive and negative convection, using Givoni correlation
depending on the temperature difference between the
for the convective heat coefficient calculation.
ground surface and the ambient air. As shown in Fig. 4,
for cold and moderate climates (Montreal, Amsterdam
and Brussels), the agreement between measurements and
calculations based on Givoni correlation is relatively satis-
factory. However, the difference between calculations using
Givoni correlation and measurements is significant for hot
climates (Oklahoma and Kiln). As evidenced by Figs. 5–7,

Fig. 6. Annual energy density on the ground: solar, long wave radiation
with sky and positive and negative convection, using McAdams correla-
tion for the convective heat coefficient calculation.

the net convection heat flux for Amsterdam and Brussels

Fig. 4. Ground temperature: comparison between measurements and
numerical results using Givoni, McAdams and Kusuma correlations of
convective heat coefficient.
is relatively small leading to negligible convection contribu-
tion. In other words, the negative portion compensates the
positive portion of the convection heat flux. This confirms
firstly why all the correlations give similar values and sec-
ondly that Givoni can give a good prediction of the deep
346 M. Ouzzane et al. / Solar Energy 108 (2014) 340–347
Fig. 7. Annual energy density on the ground: solar, long wave radiation
with sky and positive and negative convection, using Kusuma correlation
for the convective heat coefficient calculation.
ground temperature for these two sites. However for the
two warmer sites (Oklahoma and Kiln) as well as the colder
site (Montreal), the negative portion of convective heat flux
is dominant over the positive portion with the higher abso-
lute value calculated using McAdams and Kusuma correla-
tions. Although Givoni correlation is still good for
Montreal, it gives significantly higher values than measured
data for Oklahoma and Kiln. This can be explained for
Montreal, by low ambient temperature and relatively high
annual solar energy, mostly leading to relatively high
ground surface temperature. In fact, the temperature differ-
ence between the ground surface and the ambient air is
increased so that it lessens the contribution of the convec-
tive coefficient. Therefore, underestimation of the convec-
tive coefficient using Givoni correlation in a weather
conditions like Montreal is not seen as much as that of
Oklahoma and Kiln. This is due to the fact that in Okla-
homa and Kiln, high annual solar energy together with high
ambient temperature present a lower temperature difference
(Tgs–Tamb). In general, McAdams and Kusuma correlations
give relatively good agreement between measurements and
calculations for all sites. However, for hot climates (Okla-
homa and Kiln) Kusuma correlation gives better results
than McAdams. Although Kusuma correlation gives the
highest convective heat transfer coefficient (Fig. 2) which
might overestimates the convection heat flux, the overesti-
mated positive portion compensates the overestimated neg-
ative portion leading to this good agreement. It has to be
mentioned that the McAdams correlation gives better
results for cold and moderate climates (Montreal,
Amsterdam and Brussels).
4.2. Comparison between results obtained from the numerical
model and measurement
Following these analysis, McAdams correlation was
selected for the validation of the CFD model. Fig. 8 shows
Fig. 8. Ground temperature profile: comparison between measured and
CFD results for Varennes site (near Montreal) using McAdams
correlation.
the validation results of the ground temperature profile for
a small city of Varennes (near Montreal) in Canada with
the physical properties of the ground obtained from the
(TRT) report: k = 2.65 W/m K and a = 0.0948 m2/day.
The simulation was performed for one typical year using
one hour time step which gives a total of 8760 h. The mete-
orological input data are: dry bulb temperature, dew point
temperature, wind velocity and global horizontal solar flux.
Thirteen (13) thermocouples of T type inserted inside a
46 m metallic probe have been used to measure the ground
temperature at different depths. Because of the important
variation of the temperature near the ground surface, more
thermocouples are located in this region. The measured
data obtained in September 10th at 10:00 am are compared
with the numerical results using McAdams correlation for
the calculation of the convective heat transfer coefficient.
Two different regions are observed; near the ground sur-
face, up to 7 m and deep until 46 m. Near the ground sur-
face, the agreement between measurement and calculation
is very good. However, after 7 m, it shows about 1.5 °C
temperature difference which then reduces to less than
1.0 °C as it goes deeper into the ground. The difference
between experiments and calculations can be explained
by nonhomogeneous thermal properties of the ground
(proven by the TRT report), water movement at some lev-
els in the ground or the uncertainty of heat flux calculation
at the ground surface.
5. Conclusion
A numerical model has been developed to analyse the
convective heat exchange effect on the deep ground temper-
ature under realistic ground surface conditions including
convection, solar radiation and long wave radiation to
sky. The model has been validated against both analytical
solution of 1D heat conduction and experimental results
and a good agreement has been demonstrated. Five sites
 M. Ouzzane et al. / Solar Energy 108 (2014) 340–347
with different weather conditions were chosen, in which
Montreal in Canada and Kiln in USA represent the cold
and the hot climates respectively while Amsterdam, Brus-
sels and Oklahoma are sited in between. As the main objec-
tive of this paper, the model is used to investigate the effect
of convective heat flux, calculated using three different cor-
relations (Givoni, McAdams and Kusuma) on the deep
ground temperature for different climates. Calculated
values were compared against measured deep ground tem-
perature data and it was found that in some sites, the dif-
ference becomes important, which stresses the importance
of using reliable convective heat transfer coefficients.
Among three correlations, Givoni correlation calculates
the lowest convective heat transfer rate, while Kusuma cor-
relation gives the highest value. For Brussels, where the net
convective heat exchange is relatively insignificant, all three
correlations calculate almost the same deep ground temper-
ature. However, for sites with relatively high annual solar
energy density (Montreal, Oklahoma and Kiln) Givoni
overestimates significantly the deep ground temperature
since the convective energy loss from the ground surface
is underestimated. In fact, the convective heat transfer
rate is influenced by both the temperature difference and
the convective heat transfer coefficient. Therefore, it can
be concluded that in sites with high annual solar energy
density (mostly leading to high ground surface tempera-
ture) and high ambient temperature, Kusuma provides a
better estimation of the deep ground temperature.
However, in cold climates with relatively high solar energy
density, McAdams or Givoni correlation is better. Finally,
Givoni can also be used for sites with relatively low solar
energy density and moderate annual average ambient tem-
perature. However, In general, McAdams and Kusuma
correlations can give better estimation over a relatively
wide weather conditions.
References
Al-Temeemi, A.A., Harris, D.J., 2001. The generation of subsurface
temperature profiles for Kuwait. Energ. Build. 33, 837–841.
Cellier, P., Richard, G., Robin, P., 1996. Partition of sensible heat fluxes
into bare soil and the atmosphere. Agric. For. Meteorol. 82, 245–265.
Deardorff, J.W., Willis, D.W., 1985. Further results from the laboratory
model of the convective planetary boundary. Bound.-Lay. Meteorol.
33 (3), 205–236.
Deru M., 2003. A model for ground-coupled heat and moisture transfer
from buildings. Technical Report, National Renewable Energy Lab-
oratory. June, NREL/TP-550-33954. <http://www.nrel.gov/docs/
fy03osti/33954.pdf>.
DOE-2, 1982. Technical information center. Engineers Manual Version
2.1A, Department of Energy, USA.
Droulia, F., Lykoudis, S., Tsiros, I., Alvertos, N., Akylas, E., Garofalakis,
I., 2009. Ground temperature estimations using simplified analytical
and semi-empirical approaches. Sol. Energ. 83 (2), 211–219.
Duffie, J.A., Beckman, W.A., 2006. Solar Engineering of Thermal
Processes, third ed. John Wiley & Sons, Inc., ISBN-13 978-0-471-
69867-8 (Printed in USA).
347
Elias, E.A., Cichota, R., Torriani, H.H., Torriani, H.H., De Jong Van
Lier, Q., 2004. Analytical soil-temperature models correction for
temporal variation of daily amplitude. Soil Sci. Soc. Am. J. 68, 784–
788.
Herb, W.R., Janke, B., Mohseni, O., Stefan, H.G., 2008. Ground surface
temperature simulation for different land covers. J. Hydrol. 356, 327–
343.
Kasuda, T., Archenbach, P.R., 1965. Earth temperature and thermal
diffusivity at selected stations in the United States. ASHRAE Trans. 71
(1).
Khatry, A.K., Sodha, M.S., Malik, M.A., 1978. Periodic variation of
ground temperature with depth. Sol. Energ. 20 (1978), 425–427.
Kroger, D.G., 2002. Convection heat transfer between a horizontal
surface and the natural environment. R&D J. 18 (3).
Kusuma, Rao G., 2004. Estimation of the exchange coefficient of heat
during low wind convective conditions. Bound. Layer Meteorol. 111,
247–273.
Lee, K.H., Strand, R.K., 2006. Implementation of an earth tube system
into EnergyPlus. In: SimBuild 2006 Conference, 2–4 August 2006,
MIT, Cambridge, Massachusetts, IBPSA-USA.
Lin, J., 1980. On the force-restore method for prediction of ground surface
temperature. J. Geophys. Res. 85, 3251–3254.
McAdams, W.H., 1954. Heat Transmission. McGraw-Hill, New York.
Mihalakakou, G., Santamouris, M., Asimakopoulos, D.N., Argiriou, A.,
1995. On the ground temperature buildings. Sol. Energ. 55 (5), 355–
362.
Mihalakakou, G., Santamouris, M., Lewis, J.O., Asimakopoulos, D.N.,
1997. On the application of the energy balance equation to predict
ground temperature profiles. Sol. Energ. 60, 181–190.
Mihalakakou, G., 2002. On estimating soil surface temperature profiles.
Energ. Build. 34, 251–259.
Mostrel, M., Givoni, B., 1982. Windscreens in radiant cooling. Passive
Sol. J. 1, 229–238.
Okada, M., Kusaka, H., 2013. Proposal of a new equation to estimate
globe temperature in an urban park environment. J. Agri. Meteorol. 69
(1), 23–32.
Palyvos, J.A., 2008. A survey of wind convection coefficient correlations
for building envelope energy system’s modeling. Appl. Therm. Eng. 28,
801–808.
Rabadiya, A.V., Kirar, R., 2012. Comparative analysis of wind loss
coefficient (Wind Heat Transfer Coefficient) for solar flat plate
collector. Int. J. Emerg. Technol. Adv. Eng. 2 (9).
RETScreen 4, 2005. Clean energy project analysis: RESTscreen Engineer-
ing & cases textbook, Ground-Source Heat Pump Project Analysis
Chapter, CanmetENERGY at Varennes, Natural Resources Canada.
Tabari, H., Sabziparvar, A.A., Ahmadi, M., 2011. Comparison of
artificial neural network and multivariate linear regression methods
for estimation of daily soil temperature in an arid region. Meteorol.
Atmos. Phys. 110, 135–142.
Thiers S, 2008. Bilans énergétiques et environnementaux de bâtiments à
énergie positive. PhD Thesis, École National Supérieure des Mines de
Paris (France). <http://pastel.archives-ouvertes.fr/docs/00/50/10/65/
PDF/These_S_Thiers.pdf>.
TRNSYS, 2005. Version 16.A. Transient system simulation program solar
energy laboratory. University of Wisconsin, Madison, USA.
Zoras, S., Dimoudi, A., Kosmopoulos, P., 2012. Analysis of conductive
temperature variation due to multi-roomunderground interaction.
Energ. Build. 55, 433–438.
Zheng, D., Hunt Jr., E.R., Running, S.W., 1993. A daily soil temperature
model based on air temperature and precipitation for continental
applications. Clim. Res. 2, 183–191.