Nh

HTa

b

c

ARAA

KHSNSG

1

kgra

T

0h

Geothermics 47 (2013) 61– 68

Contents lists available at SciVerse ScienceDirect

Geothermics

jou rn al h om epa ge: www.elsev ier .com/ locate /geothermics

umerical simulation and sensitivity study of double-layer Slinky-coilorizontal ground heat exchangers

ikari Fujii a,∗, Shohei Yamasakia, Takahiro Maeharaa,akashi Ishikamib,1, Naokatsu Chouc,2

Department of Earth Resources Engineering, Kyushu University, JapanMitsubishi Materials Techno Corporation, JapanKyushu Electric Power Co., Inc., Japan

a r t i c l e i n f o

rticle history:eceived 5 July 2012ccepted 27 February 2013vailable online 30 March 2013

eywords:orizontal ground heat exchangerlinky-coilumerical simulationensitivity studyround-source heat pump

a b s t r a c t

In ground-source heat pump (GSHP) systems, the application of Slinky-coil horizontal ground heatexchangers (HGHEs) greatly reduces the initial costs for the system since the HGHEs can be constructedusing common excavation machines. Though HGHEs have been commonly used in the United Statesand Canada, where abundant land space is available for installing HGHEs, the reduction of the land arearequirement is important for the wider application of the system to other regions of the world. For thispurpose, the introduction of a double-layer Slinky-coil HGHE is considered an effective choice if the heatexchange rates are much more than those for single-layer HGHEs.

In this study, long-term cooling and heating tests, using single-layer and double-layer Slinky-coilHGHEs as the heat source, were conducted in Fukuoka, Japan to compare their heat exchange capacities.The tests showed that the heat exchange capacity of HGHEs per unit land area is remarkably enhancedby the introduction of double-layer HGHEs. Numerical simulation models were then developed for theHGHEs on the basis of the procedures of Fujii et al. (2012) after modifications of surface boundary con-ditions. The models could successfully reproduce the temperature behaviors of the heat medium (heatcarrier fluid) and ground temperatures in the cooling and heating tests, demonstrating the reliability ofthe numerical model for double-layer Slinky-coil HGHEs.

Using the model, sensitivity studies were performed to optimize the design of the double-layer HGHEs.The results of the sensitivity study on installation depth showed that the optimum depth of the upper

layer was 1.5 m in case that the depth of the lower layer was fixed at 2.0 m. The preferable direction ofheat-medium circulation was then investigated and it was concluded that circulation from the upperlayer to the lower layer is the most suitable direction. Finally, the influence of the reflectance of the landsurface was investigated by changing the albedo of the land surface as 0.1, 0.3 and 0.6. The numericalsimulations showed that lower albedo is preferable in heating operations, while higher albedo is favorablein cooling operations.

© 2013 Elsevier Ltd. All rights reserved.

. Introduction

Slinky-coil horizontal ground heat exchangers (HGHEs) arenown to be a cost-effective choice to reduce the initial cost of

round-source heat pump (GSHP) systems since HGHEs do notequire drilling machines to construct. However, there are far fewerpplications of Slinky-coil HGHEs than applications of vertical GHEs

∗ Corresponding author at: 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan.el.: +81 92 802 3343; fax: +81 92 802 3343.

E-mail address: fujii@mine.kyushu-u.ac.jp (H. Fujii).1 1-14-16 Kudankita, Chiyoda-ku, Tokyo 100-8117, Japan.2 1-10-1 Takagisehigashi, Saga 849-0922, Japan.

375-6505/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.geothermics.2013.02.006

owing to the large land area required to bury the heat-exchangepipes. Therefore, to promote the use of HGHEs in locations withlimited space, the heat exchange rate per unit land area should beimproved according to an optimum design of the HGHEs.

The application of double-layer Slinky-coil HGHEs is consideredone of the most promising ways to enhance the heat exchange rateper unit area. Hence, the effectiveness of the double-layer HGHEsneeds to be proved though field tests or numerical simulations.Several investigations have been carried out to develop analyticaland numerical models of single-layer straight HGHEs (Mei, 1986;

Piechowski, 1998; Esen et al., 2007; Koyun et al., 2009; Pulat et al.,2009; Demir et al., 2009; Philippe et al., 2010; Bottarelli and DiFederico, 2010; Bennaza et al., 2011; Fontaine et al., 2011). In con-trast, the research on Slinky-coil HGHEs has been limited. Fujii et al.

62 H. Fujii et al. / Geothermics 47 (2013) 61– 68

of the

(twbss(rtesHspvhodrhm

doet

Fig. 1. Schematics

2010) presented the results of thermal response tests and long-erm cooling and heating tests on two sets of Slinky-coil HGHEsith different loop angles and compared their heat-exchange capa-

ilities. Wu et al. (2010) developed a 3D numerical model toimulate the performance of Slinky-coil HGHEs and conductedensitivity studies on coil diameters and pitches. Congedo et al.2012) developed 3D numerical models for straight, Slinky and spi-al HGHEs and compared their performances. Using fine mesheso model the Slinky-coils, however, the numerical models by Wut al. (2010) or by Congedo et al. (2012) could handle only small-cale models and were not applicable for the modeling of field-scaleGHEs. Fujii et al. (2012) then developed a full-field numerical

imulation model of single-layer Slinky-coil HGHEs applying a sim-lified shape of Slinky-coils using numerical software, FEFLOW, andalidated the model using the results of short-term and long-termeat-exchange tests. Li et al. (2012) developed an analytical modelf Slinky-coil HGHEs using the moving line source theory and vali-ated the model using laboratory experiments. Through the aboveesearches, the optimum designs of the single-layer Slinky-coilsave been rigorously studied. However, there has been no experi-ental or analytical research on double-layer Slinky-coil HGHEs.In this study, numerical simulation models of single-layer and

ouble-layer Slinky-coil HGHEs are constructed after modificationsf the numerical model developed by Fujii et al. (2012). The mod-ls are validated using the results of long-term cooling and heatingests of single-layer and double-layer Slinky-coil HGHEs, which

field test facility.

were carried out from December 2010 to February 2011 in Fukuoka,Japan. Using the developed numerical model, sensitivity studies areperformed to optimize the design of the double-layer Slinky-coilHGHEs. In the analysis, a fixed heat load is given to the heat mediumassuming the operation of heat pumps, and the temperatures ofthe heat medium are calculated to evaluate the energy efficiencyof the heat pump. First, the optimum depth of the upper layer isinvestigated using a fixed depth of the lower layer of 2.0 m. Thepreferable direction of heat-medium circulation, upper to lower,lower to upper or in parallel, is then determined from the simu-lation results. Finally, the effect of reflectance (albedo) at the landsurface is investigated by changing the albedo as 0.1, 0.3 and 0.6,assuming various types of land uses.

2. Field tests of Slinky-coil ground heat exchangers

From September 2010 to February 2011, long-term cooling andheating tests on single-layer and double-layer Slinky-coil HGHEswere carried out in Itoshima City, Fukuoka, Japan. The local annualtemperatures used were the 10-year averages (2000–2009) for theannual average temperature (16.5 ◦C) and for the monthly averagetemperatures in February (7.1 ◦C) and August (27.5 ◦C). The field site

is located on a hill on which several greenhouses have been built togrow orchids and other plants. The type of soil is sandy clay, whichis widespread in the shallow ground. The groundwater level wasmeasured in a 30-m-deep observation well as being approximately

H. Fujii et al. / Geothermics 47 (2013) 61– 68 63

3tsodsM

t2aadhinIltetflaa0tTtal1thphptctaw

wpt

Fig. 2. Photograph of Loop 2 under construction.

m below the ground surface throughout the year, indicating thathe shallow ground soil was unsaturated. For the determination ofoil properties, a soil sample was taken at the bottom of a trenchf 1.5 m deep located adjacent to the Slinky-coils. The thermal con-uctivity of the soil sample was measured as 1.16 W/m/K using aingle-probe type thermal conductivity meter (Kyoto Electronicsanufacturing Co., Ltd., QTM-500) in the laboratory.Fig. 1 is a schematic drawing of the experiment facility. Two

ypes of Slinky-coil HGHEs were constructed, namely Loops 1 and. Loop 1 is a single-layer HGHE with a land space of nearly 125 m2,

total coil length of 100 m and a trench depth of 1.5 m. Loop 2 is double-layer HGHE occupying the similar land space as Loop 1oes, whose upper layer (1.0 m deep) and lower layer (2.0 m deep)ave a total coil length of 100 m each. In this paper, the coil length

s defined as the length of coil-shaped heat exchangers, and doesot mean the total length of polyethylene pipes when laid straight.

n both loops, water is used as heat medium considering the mildocal climate. The flow of heat medium (18 L/min) is divided intowo lines (9 L/min each) before the inlet of the HGHEs. In Loop 1,ach flow of heat medium circulates though half of the coils andhe flows merge at the outlet as shown in the figure. In Loop 2, eachow of heat medium passes into upper and lower layers in parallelnd the two flows merge at the outlet. The diameter of the loopnd the inner and outer diameters of the polyethylene pipes are.8, 0.034, and 0.024 m, respectively. The thermal conductivity ofhe polyethylene pipe is given as 0.35 W/m/K in the specifications.he dimensions of the trenches are given in Fig. 1. The shape of therench in Loop 2 is not rectangular owing to the limited land avail-bility. The ratio of pipe length [=(length of polyethylene pipe)/(coilength)] is set as 5.0 in each loop, which indicates that 500 and000 m of polyethylene pipes were used in Loops 1 and 2, respec-ively. Each HGHE is connected to a small water–air heat pumpaving 6 kW heating capacity and 5 kW cooling capacity. The heatumps are installed in greenhouses for growing orchids to cool andeat the houses. All surface piping between the HGHEs and heatumps is insulated using polystyrene covers. Fig. 2 is a photographaken during the construction of Loop 1. After laying the coils andovering them with soil, water was sprayed on the soil to reducehe void space around the coils. The trenches were then back-filled,nd the soil was compressed with power shovels. The land surfaceas not paved but left as a soil surface.

Circulation rates were controlled using flow control valves and

ere measured using electromagnetic flow meters (Keyence Cor-oration, FD-M10AY, accuracy ± 0.18 L/min). The temperatures ofhe heat medium were measured at the inlet and outlet of heat

Fig. 3. Farfield temperatures and ambient temperature during the field test period.

pumps (Sensors A–D in Fig. 1) and at the outlets of the upper andlower layers in Loop 2 (Sensors E and F in Fig. 1) using thermo-resistance thermometers (Pt100 �, accuracy ±0.15 ◦C). Groundtemperatures were also measured using Pt100 � in the central partof Loop 2 at Point G (0.75 m deep) and Point H (1.5 m deep) as shownin Fig. 1. Farfield temperatures were measured in an observationhole drilled approximately 10 m from the loops at depths of 0.25,0.75, 1.5 and 2.5 m using Pt100 �. For data collection, flow metersand Pt100 � thermometers were connected to a high-accuracy datalogger (Keyence Corporation, GR-3500). The accuracies of the log-ger are ±0.3 ◦C for Pt100 � and ±0.04 L/min for flow meter, whichgives the combined accuracies of ±0.45 ◦C for temperature mea-surement and ±0.22 L/min for flow measurement, respectively).Meteorological data (i.e., ambient temperature, precipitation, windvelocity, and solar radiation) were also recorded every hour at thefield test site. Fig. 3 shows ground temperatures and ambient tem-peratures measured during the field test period. As the ambienttemperature decreased in winter, ground temperatures decreasedat all depths. With an increase in sensor depth, the variation in tem-perature decayed and the delay of temperature change increasedsince heat conduction from the land surface to deep sensors tooklonger.

The long-term cooling and heating tests were carried out bymaintaining the greenhouse temperatures at designated temper-atures using the heat pumps in each greenhouse. The designatedtemperatures were changed during the day and night for the opti-mum growth of the plants; they were 28 ◦C all day in the coolingoperations, and 25 ◦C (daytime) and 20 ◦C (nighttime) in heatingoperations. Fig. 4 shows the daily average heat exchange rates perunit coil length in each loop and the ratio of the heat exchangerate in the lower loop to the total heat exchange rate in Loop 2.Fig. 5 shows the temperature of the heat medium at the inlet ofheat pumps (i.e., the temperature at the outlet of the HGHEs). Theheat pumps were operated in cooling mode from September 1 toOctober 10, and then in heating mode until the end of the field test.Hence, the heat exchange rates in Fig. 4 indicate heat rejection ratesin the cooling period, while they indicate heat extraction rates in theheating period. As shown in Fig. 4, heat exchange rates decreasedapproaching mid-winter since the coefficient of performance (COP)of heat pumps dropped with the decrease in heat-medium temper-

atures. As shown in Fig. 5, the heat-medium temperatures rapidlydecreased after the heating operation began on October 10. Com-paring the two loops, Loop 2 had superior performance over Loop 1

64 H. Fujii et al. / Geotherm

FrL

so1pfhwl

rp2Oilil

lip

Fa

ig. 4. Daily average heat exchange rates per unit coil length in each loop and theatio of the heat exchange rate in the lower loop to the total heat exchange rate inoop 2.

ince the total length of the heat exchangers of Loop 2 is twice thatf Loop 1. The operation of Loop 1 was forced to stop on January0 because the heat-medium temperature at the outlet of the heatump reached the preset lower limit of 3 ◦C, which was determinedor the heat pump operation using water as the heat medium. Theeat pump in Loop 2 continued a stable operation throughout theinter owing to the higher heat exchange capacity of the double-

ayer HGHEs, and no shut-down was observed.At the beginning of the heating period, in Loop 2, the lower/total

atio of heat exchange rates was close to 0.5 since the ground tem-eratures at the depth of the upper and lower layers (1.0 m and.0 m) are close to each other in October 2010 as shown in Fig. 3.n the other hand, the ratio gradually increased and exceeded 0.75

n January 2011, which indicates that the contribution of the loweroop became more prominent in mid-winter since the temperaturen the upper layer deceased more quickly than that in the lowerayer.

For the evaluation of economy, the installation costs for each

oop were calculated by summing the excavation and backfill-ng cost of the trenches and the material cost for heat exchangeipes. The local excavation and backfilling cost in September, 2010

ig. 5. Temperature of the heat medium at the inlet of heat pumps in Loops 1nd 2.

ics 47 (2013) 61– 68

was 3380 Yen/m3 (42.25 US$/m3 using 1 US$ = 80 Yen). The price ofpolyethylene pipe used in the loops was 376 Yen/m (4.70 US$/m).Then, the installation cost of Loop 1 is totalized as 822,626 Yen(10,283 US$) assuming the land space of 125 m2, the excavationdepths of 1.5 m and the pipe length of 500 m. Similarly, the installa-tion cost of Loop 2 is calculated as 1,221,000 Yen (15,263 US$) usingthe land space of 125 m2, the excavation depths of 2.0 m and thepipe length of 1000 m. The above evaluation shows that the installa-tion cost of Loop 2 is higher than that of Loop 1 by 48.4% based on thesame surface land area. Since the energy cost of the GSHP systemdepends on the COP of heat pumps, the use of double-layer HGHEswould enable lower energy cost than single-layer HGHEs due tothe better heat exchange performance. The choice between single-layer and double-layer should be determined carefully consideringthe land availability, the local climate and the operation scenariosof the GSHP system.

3. Numerical modeling

Fujii et al. (2012) developed a three-dimensional (3D) numer-ical model including the Slinky-coil HGHEs and the surroundingground using a finite-element numerical simulator, FEFLOW Ver.5.4 (Diersch, 2005). In their modeling procedures, the Slinky-coilHGHEs were simplified as a thin flat plate whose width was setequal to the diameter of the loops. The thickness of the flow pathwas determined so as to equalize the total inner volume of the heat-exchange pipes with the volume of the flow path in the model. Theflow path was modeled as a porous medium with porosity of 1.0and high hydraulic conductivity of 1.0 m/s. The thermal conductiv-ity of the polyethylene pipe is an important parameter in matchingthe rate of heat exchange between the heat medium and surround-ing soil. Because the actual surface areas of the Slinky-coil heatexchangers are much smaller than those of the plate-like pipe, thethermal conductivity of the pipes was adjusted to match the actualand calculated heat exchange rates by trial and error in the history-matching calculations. The flow path was surrounded by grids withvery low permeability (hydraulic conductivity = 1.0 × 10−15 m/s) toavoid leakage of the heat medium into the surrounding soil. Theheat medium was injected at the inlet of the flow path and producedat the outlet of the flow path.

In this research, the same modeling procedures as above wereapplied to calculate the heat transfer in the ground and pipes. As theinitial condition of the numerical model, the measured ground tem-peratures at the test site were input for each layer. All peripheraland bottom boundaries were defined as having a constant temper-ature with reference to the measured values. On the other hand,Fujii et al. (2012) considered the energy balance at the land sur-face using Eq. (1) and calculated the temperature distribution inthe shallow ground.

Q = Rsol + Rsky − Rsurf − Hsurf − Lsurf, (1)

where Q is the heat flux at the land surface (W/m2); Rsol is the totalsolar radiation (W/m2); Rsky is the downward longwave radiation(W/m2); Rsurf is the upward longwave radiation from the groundsurface (W/m2); Hsurf is the sensible heat flux (W/m2); Lsurf is thelatent heat flux (W/m2).

This equation was programmed and coupled with FEFLOW as anouter module. However, the coupling of the outer routine signifi-cantly reduced the computation speed of FEFLOW, which made thesimulations quite time-consuming. In this research, the calculationof the energy balance at the land surface was simplified using the

sol–air temperature (SAT) as shown in Eq. (2).

SAT = �0 + 1˛0((1 − ˛s)J − εJeh)

, (2)

H. Fujii et al. / Geothermics 47 (2013) 61– 68 65

wpase

ewitcccw

˛

˛

f

J

w(nen

Ttctbgtnnp2i2

oLaai

The thermal conductivity of the polyethylene pipe, the matchingparameter in history matching, was determined as 0.027 W/m/K,which is close to the value of 0.025–0.045 W/m/K used by Fujii et al.(2012).

Fig. 6. 3D view of the numerical simulation model for Loop 2.

here SAT is the sol–air temperature (◦C); �0 is the ambient tem-erature (◦C); ˛0 is the coefficient of overall heat transfer betweenir and soil (W/m2/K); ˛s is the albedo (=0.3 for soil); J is the totalolar radiation (W/m2); ε is the longwave emissivity (–); Jeh is theffective emission (W/m2).

As described in Section 2, meteorological data including ambi-nt temperature, precipitation, wind velocity, solar radiation, etc.,ere recorded every hour at the field test site. The parameters used

n Eq. (2) were determined based on these measurements in eachime step. The coefficient of overall heat transfer, ˛0, is the sum ofonvective heat transfer coefficient, ˛c, and radiative heat transferoefficient, ˛r. A value of 5.1 (W/m2/K) was used for ˛r and ˛c wasalculated by the Jürges equation (Eqs. (3) and (4)) with respect toind velocity v (m/s):

c = 5.8 + 3.9v (v < 5 m/s) (3)

c = 7.1 v0.78 (v > 5 m/s) (4)

Longwave emissivity, ε, was determined as 0.9, a common valueor soil. The effective emission, Jeh, was calculated using Eq. (5):

eh = �(273.16 + �0)4(0.474 + 0.076f 1/2) (5)

here � and f are the Stephen–Bolzmann constant=5.67 × 10−8 W/m2/K4) and water vapor pressure (mmHg)ear land surface. Using the above equations, SAT was calculatedach time step and was input as the surface temperature of theumerical model.

A 3D view of the numerical model for Loop 2 is shown in Fig. 6.he grids including the Slinky-coils are surrounded by white lines inhe figure. The inlet and outlet of the HGHE are shown with red cir-les. Outside the HGHEs, peripheral grids of 10 m width are definedo leave enough distance between the HGHEs and the outer modeloundaries to eliminate the influence of the latter. Finite-elementrids are refined near the HGHE and are coarsened as they approachhe outer boundaries. A total of 22 and 28 layers were defined in theumerical models of Loops 1 and 2, respectively, which are refinedear the plane including the HGHEs. The total numbers of elementser layer are 4222 and 6462 in the numerical models of Loops 1 and, respectively. The number of elements in Loop 2 is larger than that

n Loop 1 since the HGHE domain of Loop 1 is rectangle while Loop has an irregular shape as shown in Fig. 1.

Fig. 7a and b compare the measured and simulated temperaturesf the heat medium at the outlet of the upper and lower layers in

oop 2, respectively, during the cooling and heating tests. Fig. 8and b compare the measured and simulated ground temperaturest Sensors G (0.75 m) and F (1.5 m) in Loop 2, respectively, dur-ng the same period. Reasonably good agreement of heat-medium

Fig. 7. Comparison of measured and simulated temperatures of the heat mediumat the outlet of the upper layer (a) and the lower layer (b) in Loop 2.

and ground temperatures was obtained after history matching.

Fig. 8. Comparison of measured and simulated ground temperatures at Sensors G(a) and F (b) in Loop 2.

66 H. Fujii et al. / Geothermics 47 (2013) 61– 68

tlSsurcd

4

peta(osHHTwfia9mr2

tF

Fig. 9. 3D view of the numerical simulation model used in the case studies.

The results of history matching for Loop 1 are not presented inhis paper since the validity of the modeling approach for the single-ayer Slinky-coil HGHE has already been shown by Fujii et al. (2012).imilar good agreement was observed between the measured andimulated heat-medium temperatures and ground temperaturessing the same thermal conductivity for polyethylene pipes. Theseesults indicate that the modified modeling approach using SATan reproduce the temperature performance of single-layer andouble-layer HGHEs.

. Sensitivity studies

Using the developed numerical model, sensitivity studies wereerformed for optimizing the design of double-layer HGHEs. In allvaluations, fixed heat disposal/extraction rates were assigned tohe HGHEs and the chronological changes in heat-medium temper-tures were calculated for cooling and heating operations. Higherlower) temperatures of the heat medium in the heating (cooling)perations allow better performance of the heat pumps. For theimulation, a simplified 3D numerical model of the double-layerGHE was constructed as shown in Fig. 9. The total length of theGHE, surrounded with white lines in the figure, was set as 50 m.he same thermophysical properties of soil, pipes and heat mediumere used in the model. The heat disposal and extraction rates werexed at 38 W per unit coil length and the operation time was sets 12 h per day. The circulation rate of the heat medium was set as

L/min in all simulations. Surface boundary conditions were deter-ined by the SAT, which was calculated using the weather data

ecorded at the test site during the field test period from 2010 to

011.

First, the optimum depth of the double-layer HGHEs was inves-igated during the heating operations from December 2010 toebruary 2011. According to the Japanese domestic regulations, the

Fig. 10. Heat-medium temperatures at the inlet of the HGHE when the depth of theupper layer is varied as 0.5, 1.0 and 1.5 m.

excavation of trenches deeper than 2.0 m requires additional safetymeasures, which makes the excavation quite expensive. Hence, theoptimum depth of the upper layer was investigated assuming afixed depth of the lower layer at 2.0 m. Fig. 10 shows the heat-medium temperatures at the inlet of the HGHE when the depthof the upper layer is varied as 0.5, 1.0 and 1.5 m. When the upperlayer is set at shallower depth, the temperatures fluctuate moreand the temperatures are much lower than in the case of deeperinstallations since the ground temperature is strongly affected bythe ambient temperature. On the other hand, the burial depth of1.5 m provided stable performance and maximized the tempera-tures during the period from the beginning of December to themiddle of February, in which high heat demand is expected. Theburial depth of 0.5 m provided better performance at the end ofFebruary, but this is considered negligible since the heat demandreduces in this period. The averages of the heat-medium temper-ature at the HGHE inlet during the whole simulation period were7.1 ◦C, 8.1 ◦C and 8.5 ◦C for burial depths of 0.5, 1.0 and 1.5 m, respec-tively. The above results show that the optimum depths of thedouble-layer HGHE are 1.5 and 2.0 m.

Next, the preferable direction of heat medium circulation wasinvestigated by changing the flow direction as (i) from the upperlayer to the lower layer, (ii) from the lower layer to the upper layer,and (iii) in parallel. The depths of the layers were fixed at 1.5 and2.0 m according to the results of the first sensitivity study. Fig. 11shows the heat-medium temperatures at the inlet of the HGHE inthe above three cases. The averages of the heat medium tempera-tures at the HGHE inlet during this period were 8.9 ◦C, 8.1 ◦C and8.5 ◦C in Cases (i)–(iii), respectively. Case (ii) gave the worst result,while Case (i) maximized the temperatures. These results can beexplained by the difference in heat exchange capacity between thetwo layers resulting from the temperature difference in each layer.In Case (ii), the heat medium collects heat from the lower layer first,but collects only limited heat from the upper layer since the tem-perature of the upper layer is lower than that of the lower layer inwinter. In Case (i), both upper and lower layers contribute to heatextraction, which results in the better performance. Case (iii) is con-sidered to have performance between the performances of Cases (i)and (ii). These results demonstrate that circulation from the upper

layer to the lower layer is the optimum circulation direction.

Finally, the effects of land surface types were investigated usingdifferent albedo (reflectance) values in Eq. (2). The examined values

H. Fujii et al. / Geothermics 47 (2013) 61– 68 67

Fig. 11. Heat-medium temperatures at the inlet of the HGHE calculated by changingtl

or0ta(of2

Hotiatlgoaraomautoat7tt

ic1wtad

of heat-medium circulation was also investigated and it was con-

he flow direction as (i) from the upper layer to the lower layer, (ii) from the lowerayer to the upper layer, and (iii) in parallel.

f albedo were 0.1 (asphalt pavement), 0.3 (soil) and 0.6 (high-eflectance asphalt). As shown in the previous section, an albedo of.3 was used successfully in the simulation of cooling and heatingests. Since the effects of albedo are expected to differ in coolingnd heating operations, heat rejection (cooling) and heat extractionheating) simulations were carried out individually for each valuef albedo. The periods of the cooling and heating simulations wererom June to September 2011 and from December 2010 to February011, respectively.

Fig. 12a shows the heat-medium temperatures at the inlet of theGHE for each albedo value in cooling operation. At the beginningf the cooling period, the difference in the heat-medium tempera-ures in each case was small since the simulations used the samenitial distribution of ground temperatures. In mid-summer, largelbedo values became more preferable since they retard the rise ofhe ground temperature. The amplitude of temperature change wasarger when the albedo was small since the temperature of shallowround was more strongly affected by the weather than in the casef using a large albedo. Fig. 12b shows the heat-medium temper-tures for each albedo value in heating operation. Contrary to theesults for the cooling operation, a small albedo was determineds more preferable since the low reflectance helps the absorptionf solar energy into the land surface, which maintains the heat-edium temperature at a higher level. This trend became clearer

pproaching late winter. Between the cases of using albedo val-es of 0.1 and 0.6, the differences in the average temperatures ofhe heat medium were 1.7 ◦C and 0.7 ◦C in the cooling and heatingperations, respectively, which implies that the values of albedore more influential in summer. This is explained by the fact thathe daily average solar radiation at the test site was 14.7 and.1 MJ/day/m2 in the cooling and heating periods, respectively, andhe temperature of the ground is thus more likely to be affected byhe albedo during the cooling operations in summer.

Assuming heating operation with a water temperature of 35 ◦Cn the secondary side of heat pumps, the performance curve of aommonly used water-water heat pump (Sunpot Co., Ltd. GSHP-001, heating/cooling capacity: 10 kW) gives COPs of 3.7, 4.1, 4.8hen the inlet water temperatures are 0 ◦C, 5 ◦C and 10 ◦C, respec-

ively. The above sensitivity studies showed that there could be temperature difference close to 2 ◦C in the inlet temperaturesepending on the designs of HGHEs. Hence, an improper design

Fig. 12. Heat-medium temperatures at the inlet of the HGHE for albedo = 0.1, 0.3and 0.6 in cooling operation (a) and heating operation (b).

of HGHEs results in the significantly lower COP values than that inwell-designed cases. The importance of the careful design of HGHEsbased on weather conditions and operation scenarios is confirmedthrough the sensitivity studies.

5. Conclusions

This study conducted long-term cooling and heating tests tocompare the heat exchange capacities of double-layer Slinky-coilHGHEs with single-layer HGHEs. The tests showed that the heatexchange capacity of HGHEs per unit land area is remarkablyenhanced by the introduction of double-layer HGHEs. Numeri-cal simulation models were then developed for the two types ofHGHEs on the basis of the modeling procedures of Fujii et al.(2012) after some modifications of surface boundary conditionsusing sol–air temperatures (SATs). The models reproduced the tem-perature behaviors of the heat medium and ground temperaturesin the cooling and heating tests, which validated the reliability ofthe numerical models.

Using the numerical model, sensitivity studies were performedto optimize the design of the double-layer HGHEs. The sensitivitystudy showed that the optimum depth of the upper layer was 1.5 mwhen the lower layer was fixed at 2.0 m. The preferable direction

cluded that circulation from the upper layer to the lower layer ismost suitable. Finally, the influence of the reflectance of the landsurface was examined using values of 0.1, 0.3 and 0.6 for the albedo.

6 therm

Tacei

A

DE(Ro

R

B

B

C

D

8 H. Fujii et al. / Geo

he numerical simulations showed that a lower albedo is prefer-ble in heating operations, while a higher albedo is favorable inooling operations. The albedo was more influential for the heat-xchange performance in summer than that in winter since theres more solar radiation in summer.

cknowledgements

This work was partly funded by the Climate Change Research,evelopment and Demonstration Program of the Ministry of thenvironment, Japan and a Grant-in-Aid of Scientific Research (A)No. 23246155) and a Grant-in-Aid of Challenging Exploratoryesearch (No. 24656543) from the Japan Society for the Promotionf Science.

eferences

ennaza, A., Blanco, E., Aichouba, M., Luis, J., Laouedj, S., 2011. Numerical investiga-tion of horizontal ground coupled heat exchanger. Energy Procedia 6, 29–35.

ottarelli, M., Di Federico, V., 2010. Adoption of GSHP system at the residentialmicro-scale: field analysis and energetic performance. In: Proceedings of the2010 World Geothermal Congress, Bali, Indonesia, April 25–29, 2010, Paper No.2922, 6 p.

ongedo, P.M., Colangelo, G., Starace, G., 2012. CFD simulation of horizontal ground

heat exchangers: a comparison among different configurations. Applied Ther-mal Engineering 33–34, 24–32.

emir, H., Koyun, A., Temir, G., 2009. Heat transfer of horizontal parallel pipe groundheat exchanger and experimental verification. Applied Thermal Engineering 29,224–233.

ics 47 (2013) 61– 68

Diersch, H.J.G., 2005. FEFLOW Reference Manual. WASY GmbH, Berlin, Germany292.Esen, H., Inari, M., Esen, M., 2007. Numerical and experimental analysis of a

horizontal ground-couples heat pump system. Building and Environment 42,1126–1134.

Fontaine, P., Marcotte, D., Pasquier, P., Thibodeau, D., 2011. Modeling of horizon-tal geoexchange systems for building heating and permafrost stabilization.Geothermics 40, 211–220.

Fujii, H., Okubo, H., Cho, N., Ohyama, K., 2010. Field tests of horizontal groundheat exchangers. In: Proceedings of the 2010 World Geothermal Congress, Bali,Indonesia, April 25–29, 2010, Paper No. 2904, 10 p.

Fujii, H., Nishi, K., Komaniwa, Y., Chou, N., 2012. Numerical modeling of Slinky-coilhorizontal ground heat exchangers. Geothermics 41, 55–62.

Koyun, A., Demir, H., Torun, Z., 2009. Experimental study of heat transfer of buriedfinned pipes for ground source heart pump applications. International Commu-nication in Heat and Mass Transfer 36, 739–743.

Li, H., Nagano, K., Lai, Y., 2012. Heat transfer of horizontal spiral heat exchangerunder groundwater advection. International Journal of Heat and Mass Transfer55, 6819–6831.

Mei, V.C., 1986. Horizontal ground–coil heat exchanger, theoretical and experimen-tal analysis. Technical report ORNL/CON-193, Oak Ridge National Laboratory, 47p.

Philippe, M., Marchio, D., Lesueur, H., Vrain, A., 2010. An evaluation of ground ther-mal properties measure accuracy by thermal response test of horizontal groundheat exchangers. In: Proceedings of the 2010 World Geothermal Congress, Bali,Indonesia, April 25–29, 2010, Paper No. 2929, 10 p.

Piechowski, M., 1998. Heat and mass transfer model of a ground heat exchanger:validation and sensitivity analysis. International Journal of Energy Research 22,965–979.

Pulat, E., Coskun, S., Unlu, K., Yamankaradeniz, N., 2009. Experimental study of hori-

zontal ground source heat pump performance for mid climate in Turkey. Energy34, 1284–1295.

Wu, Y., Gan, G., Verhoef, A., Vidale, P.L., Gonzalez, R.G., 2010. Experimental mea-surement and numerical simulation of horizontal-coupled Slinky ground heatexchangers. Applied Thermal Engineering 30, 2574–2583.