P. 1
Advances in Modeling of Ground Source Heat Pump Systems

Advances in Modeling of Ground Source Heat Pump Systems

|Views: 169|Likes:
Published by Luca Mazzarella

More info:

Published by: Luca Mazzarella on Nov 05, 2011
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

11/05/2011

pdf

text

original

Sections

  • 1. Introduction
  • 1.1. Overview of Ground-Source Heat Pump Systems
  • 1.1.1. Ground-Water Heat Pump Systems
  • Figure 1-2. A schematic of a ground-water heat pump system
  • 1.1.2. Ground-Coupled Heat Pump Systems
  • 1.1.2.1. Vertical Ground-Coupled Heat Pump Systems
  • 1.1.2.2. Horizontal Ground-Coupled Heat Pump Systems
  • 1.1.3. Surface-Water Heat Pump Systems
  • Figure 1-4. A schematic of a surface-water heat pump system
  • 1.1.4. Standing Column Well Systems
  • 1.2. Thesis Objectives and Scope
  • 1.3. The Overall Modeling Approach
  • Closed-Loop Ground-Source Heat Pump Systems
  • 2.1. Introduction
  • 2.2. Coupled Ground Water Flow and Heat Transport
  • 2.2.1. Ground-Water Flow
  • 2.2.2. Heat Transport in Ground Water
  • 2.2.3. Typical Hydraulic and Thermal Property Values for Soils and Rocks
  • Typical Values of Hydraulic and Thermal Properties of Soils and Rocks
  • 2.2.4. Conduction Versus Convection in Geologic Materials
  • 2.2.5. Numerical Ground-Water Flow and Heat Transport Models
  • 2.3. The Numerical Model
  • 2.3.1. Model Description
  • 2.3.2. The Finite Element Mesh
  • Figure 2-3. Finite element mesh representing a 16 borehole field
  • 2.3.3. Boundary Conditions
  • 2.3.4. Validation of the Numerical Model
  • 2.4. Results And Discussion
  • 2.4.1. Single-Borehole Simulations
  • 2.4.2. Simulated In-Situ Thermal Conductivity Tests
  • TABLE 2-3 Case Summary of Simulated In-Situ Ground Thermal Conductivity Tests
  • 2.4.3. Borehole Field Simulations
  • Summary of Borehole Field Design Parameters for Each Test Case
  • 2.5. Concluding Remarks and Recommendations for Future Work
  • 3.1. Introduction
  • 3.2. Heat Transfer In Ponds
  • 3.2.1. General Overview
  • 3.2.2. Existing Pond and Lake Models
  • Figure 3-1. Heat transfer mechanisms in shallow ponds
  • 3.3. Experimental Methods
  • 3.3.1. Pond Description and Data Collection
  • 3.3.2. Weather Instrumentation and Data Collection
  • 3.4. Model Development
  • 3.4.1. Governing Equations
  • 3.4.1.1. Solar Radiation Heat Gain
  • 3.4.1.2. Thermal Radiation Heat Transfer
  • 3.4.1.3. Convection Heat Transfer at the Pond Surface
  • 3.4.1.4. Heat Transfer to the Ground
  • 3.4.1.5. Heat Transfer Due to Ground Water Seepage
  • 3.1.4.6. Heat Transfer Due to Evaporation
  • 3.4.1.7. Heat Transfer Due to the Heat Exchange Fluid
  • 3.4.1.8. Solving the Overall Energy Balance Equation
  • 3.4.2. Computer Implementation
  • 3.5. Results and Discussion
  • 3.5.1. Model Comparison to Experimental Results with No Heat Rejection
  • 3.5.2. Model Comparison to Experimental Results with Heat Rejection
  • 3.5.3. Model Application
  • Table 3-1.Summary of Pond Performance for Example GSHP System Simulation
  • 3.6. Concluding Remarks and Recommendations for Future Work
  • 4.1. Introduction
  • 4.2. Heat Transfer in Pavement Slabs
  • 4.3. Experimental Methods
  • 4.3.1. Test Slab Description and Data Collection
  • 4.3.1.1. Bridge Deck Test Section
  • 4.3.1.2. Parking Lot Test Section
  • 4.3.2. Weather Instrumentation and Data Collection
  • 4.4. Model Development
  • 4.4.1. Governing Equations
  • 4.4.2. The Finite Difference Grid
  • 4.4.3. Boundary Conditions
  • 4.4.3.1. Solar Radiation Heat Flux
  • 4.4.3.2. Thermal Radiation Heat Flux
  • 4.4.3.3. Convection Heat Flux at the Pavement Surfaces
  • 4.4.3.4. Heat Flux Due to Rain and Snow
  • 4.4.3.5. Heat Transfer Due to the Heat Exchange Fluid
  • 4.4.4. Computer Implementation
  • 4.5. Results and Discussion
  • 4.5.3. Model Application
  • 5. Summary and Conclusions
  • REFERENCES

ADVANCES IN MODELING OF

GROUND-SOURCE HEAT
PUMP SYSTEMS
By
ANDREW D. CHIASSON
Bachelor of Applied Science
University of Windsor
Windsor, Ontario, Canada
1989
Master of Applied Science
University of Windsor
Windsor, Ontario, Canada
1992
Submitted to the Faculty of the
Graduate College of the
Oklahoma State University
in partial fulfillment of
the requirements for
the degree of
MASTER of SCIENCE
December 1999
ii
ADVANCES IN MODELING OF
GROUND-SOURCE HEAT
PUMP SYSTEMS
Thesis Approved:
Thesis Adviser
Dean of the Graduate College
iii
ACKNOWLEDGEMENTS
This thesis would not have been possible without the contributions of several
people. To all of these people, I wish to express my gratitude.
First and foremost, I wish to thank my thesis adviser, Dr. Jeffrey Spitler. His
expertise and respected reputation in the thermal sciences field allowed for the funding of
many projects and the opportunity for me to gain knowledge and experience in several
geothermal topics.
I also wish to thank to Dr. Simon Rees who served as an unofficial co-adviser
along the course of this thesis. His expertise in computational methods was quite
valuable in helping me to achieve the project goals.
Thanks also to Dr. Marvin Smith who provided input and experimental data
regarding the supplemental heat rejecter studies
The review and contribution to this thesis by the committee members, Dr. R.
Delahoussaye and Dr. A. Ghajar are appreciated.
Last, but not least, I thank my colleagues in B-10, both here and gone, namely
Cenk Yavuzturk and Nagendra Jain, for their ideas and suggestions along the way.
iv
TABLE OF CONTENTS
Chapter Page
1. Introduction.............................................................................................................. 1
1.1. Overview of Ground-Source Heat Pump Systems .............................................. 1
1.1.1. Ground-Water Heat Pump Systems ........................................................... 4
1.1.2. Ground-Coupled Heat Pump Systems........................................................ 5
1.1.2.1. Vertical Ground-Coupled Heat Pump Systems.................................. 6
1.1.2.2. Horizontal Ground-Coupled Heat Pump Systems............................ 10
1.1.3. Surface-Water Heat Pump Systems ......................................................... 12
1.1.4. Standing Column Well Systems .............................................................. 13
1.2. Thesis Objectives and Scope............................................................................ 15
1.3. The Overall Modeling Approach...................................................................... 17
2. A Preliminary Assessment of the Effects of Ground-Water Flow
on Closed-Loop Ground-Source Heat Pump Systems ......................................... 20
2.1. Introduction ..................................................................................................... 20
2.2. Coupled Ground-Water Flow and Heat Transport ............................................ 24
2.2.1. Ground-Water Flow................................................................................ 24
2.2.2. Heat Transport in Ground Water ............................................................. 26
2.2.3. Typical Hydraulic and Thermal Properties of Soils and Rocks ................ 27
2.2.4. Conduction versus Advection in Geologic Materials ............................... 30
2.2.5. Numerical Ground-Water Flow and Heat Transport Models.................... 33
2.3. The Numerical Model ...................................................................................... 37
2.3.1. Model Description................................................................................... 37
2.3.2. The Finite Element Mesh ........................................................................ 38
2.3.3. Boundary Conditions............................................................................... 39
2.3.4. Validation of the Numerical Model ......................................................... 41
2.4. Results and Discussion..................................................................................... 43
2.4.1. Single-Borehole Simulations ................................................................... 43
2.4.2. Simulated In-Situ Thermal Conductivity Tests ........................................ 45
2.4.3. Borehole Field Simulations ..................................................................... 49
2.5. Concluding Remarks and Recommendations for Future Work ......................... 57
3. A Model for Simulating the Performance of a Shallow Pond
as a Supplemental Heat Rejecter with Closed-Loop
Ground-Source Heat Pump Systems .................................................................... 60
v
Chapter Page
3.1. Introduction ..................................................................................................... 60
3.2. Heat Transfer In Ponds .................................................................................... 61
3.2.1. General Overview................................................................................... 61
3.2.2. Existing Pond and Lake Models .............................................................. 62
3.3. Experimental Methods ..................................................................................... 66
3.3.1. Pond Description and Data Collection..................................................... 66
3.3.2. Weather Instrumentation and Data Collection.......................................... 67
3.4. Model Development ........................................................................................ 68
3.4.1. Governing Equations............................................................................... 68
3.4.1.1. Solar Radiation Heat Gain .............................................................. 69
3.4.1.2. Thermal Radiation Heat Transfer.................................................... 71
3.4.1.3. Convection Heat Transfer at the Pond Surface................................ 72
3.4.1.4. Heat Transfer to the Ground ........................................................... 75
3.4.1.5. Heat Transfer Due to Ground Water Seepage ................................. 76
3.4.1.6. Heat Transfer Due to Evaporation................................................... 77
3.4.1.7. Heat Transfer Due to the Heat Exchange Fluid ............................... 79
3.4.1.8. Solving the Overall Energy Balance Equation................................. 83
3.4.2. Computer Implementation....................................................................... 84
3.5. Results and Discussion..................................................................................... 87
3.5.1. Model Comparison to Experimental Results With No Heat Rejection...... 87
3.5.2. Model Comparison to Experimental Results With Heat Rejection ........... 89
3.5.3. Model Application .................................................................................. 92
3.6. Concluding Remarks and Recommendations for Future Work ......................... 97
4. A Model for Simulating the Performance of a Pavement
Heating System as a Supplemental Heat Rejecter with
Closed-Loop Ground-Source Heat Pump Systems............................................... 99
4.1. Introduction ..................................................................................................... 99
4.2. Heat Transfer In Pavement Slabs ................................................................... 100
4.3. Experimental Methods ................................................................................... 101
4.3.1. Test Slab Description and Data Collection............................................. 101
4.3.1.1. Bridge Deck Test Section ............................................................. 101
4.3.1.2. Parking Lot Test Section............................................................... 104
4.3.2. Weather Instrumentation and Data Collection........................................ 105
4.4. Model Development ...................................................................................... 106
4.4.1. Governing Equations............................................................................. 106
4.4.2. The Finite Difference Grid .................................................................... 108
4.4.3. Boundary Conditions............................................................................. 108
4.4.3.1. Solar Radiation Heat Flux............................................................. 110
4.4.3.2. Thermal Radiation Heat Flux........................................................ 111
4.4.3.3. Convection Heat Flux at the Pavement Surfaces ........................... 111
4.4.3.4. Heat Flux Due to Rain and Snow.................................................. 113
4.4.3.5. Heat Transfer Due to the Heat Exchange Fluid ............................. 116
vi
Chapter Page
4.4.4. Computer Implementation..................................................................... 118
4.5. Results and Discussion................................................................................... 121
4.5.1. Model Comparison to Experimental Results With No Heat Rejection.... 121
4.5.2. Model Comparison to Experimental Results With Heat Rejection ......... 121
4.5.3. Model Application ................................................................................ 126
4.6. Concluding Remarks and Recommendations for Future Work ....................... 131
5. Summary and Conclusions................................................................................... 133
References.................................................................................................................. 140
APPENDIX A – Model and Experimental Uncertainty Analysis............................ 145
vii
LIST OF TABLES
Table Page
2-1. Typical Values of Hydraulic and Thermal Properties of Soils and Rocks............... 29
2-2. Peclet Numbers Corresponding to Typical Values of
Hydraulic and Thermal Properties of Soils and Rocks ........................................... 32
2-3. Case Summary of Simulated In-Situ Ground Thermal Conductivity Tests ............. 47
2-4. Summary of Borehole Field Design Parameters for Each Test Case....................... 50
3-1. Summary of Pond Performance for Example GSHP System Simulation................ 96
4-1. Summary of Pavement Heating System Performance for
Example GSHP System Simulation..................................................................... 131
viii
LIST OF FIGURES
Figure Page
1-1. Schematic of cycles in a GSHP system in cooling and heating modes...................... 2
1-2. A schematic of a ground-water heat pump system. .................................................. 5
1-3. A schematic of a vertical borehole and a horizontal
ground-coupled heat pump system........................................................................... 7
1-4. A schematic of a surface-water heat pump system. ................................................ 13
1-5. A schematic of a standing column well system. ..................................................... 14
2-1. Schematic of typical apparatus for in-situ
ground thermal conductivity testing....................................................................... 21
2-2. Finite element mesh representing a single borehole. .............................................. 38
2-3. Finite element mesh representing a 16 borehole field............................................. 39
2-4. Average borehole fluid temperature vs. time for
coarse sand, fine sand, and shale............................................................................ 44
2-5. Average borehole fluid temperatures for the 12 simulated
in-situ ground thermal conductivity test cases........................................................ 48
2-6. Predicted effective ground thermal conductivity values versus
ground-water flow velocity for the simulated in-situ thermal
conductivity tests................................................................................................... 49
2-7. Hourly ground loads for the test building............................................................... 51
2-8. Design borehole depths versus ground-water flow velocity for
the simulated in-situ thermal conductivity tests. .................................................... 52
ix
Figure Page
2-9. Annual maximum and minimum average borehole fluid
temperatures for the 16 borehole field simulations................................................. 54
2-10. Temperature contours for Case 8 for the end of September
of years 1, 2, 5, and 10. ....................................................................................... 56
3-1. Heat transfer mechanisms in shallow ponds........................................................... 63
3-2. Pond model component configuration. .................................................................. 85
3-3. Pond model computer algorithm. ........................................................................... 86
3-4. Comparison of observed and simulated average pond temperatures
with no heat rejection for the (a) 2-feet deep pond and
(b) 3.5 feet deep pond............................................................................................ 88
3-5. Comparison of observed and simulated heat exchange fluid return
temperatures for the (a) 2-feet deep pond and (b) 3.5 feet deep pond. .................... 90
3-6. Comparison of observed and simulated heat rejected to the
(a) 2-feet deep pond and (b) 3.5 feet deep pond. .................................................... 91
3-7. System schematic for the example model of a GSHP system with a
shallow pond supplemental heat rejecter................................................................ 92
3-8. Building thermal loads for the example building in Tulsa, OK............................... 94
3-9. Entering heat pump water temperatures for the example GSHP
system simulation with no pond and with a 2-feet deep, 6000 ft
2
pond. ................. 95
4-1. Serpentine pipe configuration in a hydronically-heated pavement
slab in (a) plan view and (b) cross-sectional view. ............................................... 102
4-2. Slinky pipe configuration in a hydronically-heated pavement
slab in (a) plan view and (b) cross-sectional view. ............................................... 102
4-3. Heat transfer mechanisms in hydronically-heated pavement slabs with
(a) no bottom exposure to the atmosphere and (b) bottom exposure
to the atmosphere. ............................................................................................... 103
4-4. Finite difference cell geometry and notation. ....................................................... 107
4-5. Model domain showing the finite-difference grid and boundary
conditions............................................................................................................ 109
x
Figure Page
4-6. Pavement heating model component configuration. ............................................. 119
4-7. Pavement heating model computer algorithm. ..................................................... 120
4-8. Comparison of observed and simulated slab surface temperatures
for the parking lot test section with no heat rejection to the slab. ......................... 122
4-9. Comparison of observed and simulated slab surface temperatures and
heat exchange fluid return temperatures for the bridge deck test section
for the nights of (a) March 7-8, 1996 and (b) March 24-23, 1996. ...................... 124
4-10. Comparison of observed and simulated (a) heat exchange fluid return
temperatures and (b) heat rejected to the parking lot test section. ...................... 125
4-11. System schematic for the example model of a GSHP system with a
pavement heating system supplemental heat rejecter. ........................................ 127
4-12. Entering heat pump water temperatures for the example GSHP
system simulation with no pavement heating and with a 24,000 ft
2
(2230 m
2
) parking lot with pavement heating. ................................................... 129
xi
NOMENCLATURE
Symbols
α = thermal diffusivity - ft
2
/hr (m
2
/s);
= solar absorptivity (-)
β = coefficient of thermal expansion - R
-1
(K
-1
)
ε = emissivity coefficient (-)
γ = Euler’s constant (0.577215664901)
µ = dynamic viscosity - lb
m
/ft-hr (Pa-s)
µ’ = extinction coefficient - ft
-1
(m
-1
)
ν = kinematic viscosity - ft
2
/hr (m
2
/s)
θ = angle of incidence of sun’s rays (radians)
ρ = density - lb/ft
3
(kg/m
3
)
ρ’ = reflectance of pond surface (-)
σ = Stephan-Boltzmann constant
= 0.1714 x 10-8 Btu/hr-ft
2
-
o
R
4
(5.67 x 10
-8
W/m
2
-K
4
)
τ = transmittance of solar radiation (-)
A = area - ft
2
(m
2
)
c,c
p
= specific heat capacity - Btu/lb
m
-
o
F (J/kg-
o
C)
d = depth - ft (m)
D = diffusion coefficient - ft
2
/s (m
2
/s)
D* = effective thermal diffusivity - ft
2
/s (m
2
/s)
ff = fouling factor - ft
2
-
o
F/hr-Btu (m
2
-
o
C/W)
g = acceleration due to gravity - ft/s
2
(m/s
2
)
h = heat or mass transfer coefficient - Btu/hr-ft
2
-
o
F (W/m
2
-
o
C);
= hydraulic head - ft (m)
H = borehole depth - ft (m)
i = hydraulic gradient - ft/ft (m/m)
I = solar radiation flux on horizontal - Btu/hr-ft
2
(W/m
2
)
K = hydraulic conductivity - ft/s (m/s)
k = thermal conductivity - Btu/hr-ft-
o
F (W/m-
o
C)
L = characteristic length - ft (m)
l = length term described by Eskilson (1987) - ft (m);
= distance between nodes or thickness – ft (m)
m& = mass flow rate– lb
m
/hr (kg/s)
" m& = mass flux – lb
m
/hr-ft
2
(kg/s-m
2
)
xii
Le = Lewis Number (-)
n = porosity (-)
N = quantity (-)
Nu = Nusselt Number (-)
P = perimeter - ft (m);
= pressure - atmospheres
Pe = Peclet number (-)
Pr = Prandtl Number (-)
q = heat transfer rate - Btu/hr (W);
= specific discharge - ft/s (m/s)
q* = ground thermal load - Btu/hr (W]
q ′ ′ = heat flux - Btu/hr-ft
2
(W/m
2
)
Q = volumetric flow rate - ft
3
/s (m
3
/s)
Q* = heat source/sink term -
o
F/s (
o
C/s)
r = radius - ft (m);
= unpolarized solar radiation component (-)
R = retardation coefficient (-);
= thermal resistance - ft
2
-
o
F/hr-Btu (m
2
-
o
C/W)
R* = ground-water recharge -s
-1
Ra = Rayleigh Number (-)
Re = Reynolds Number (-)
S
s
= specific storage coefficient - ft
-1
(m
-1
)
T = temperature -
o
F (
o
C)
t = time - s
T
om
= undisturbed mean temperature of the ground -
o
F (
o
C)
U = overall heat transfer coefficient - Btu/hr-ft
2
-
o
F (W/m
2
-
o
C)
V = volume - ft
3
(m
3
)
v = average linear groundwater velocity - ft/s (m/s)
= velocity - ft/s (m/s)
V* = volumetric flow rate - ft
3
/s (m
3
/s)
w = humidity ratio - lb
m
water/lb
m
dry air (kg water/kg dry air)
x,z = rectangular coordinates
∆t = time step - s
Subscripts
⊥ = perpendicular component
|| = parallel component
a = absorbed component of solar radiation
AB = transfer from material A (water) to material B (air)
b = beam radiation;
= borehole
c = convection
circuit = flow circuit or spool
d = diffuse radiation;
xiii
= diffusion
db = dry bulb
dp = dew point
eff = effective
fg = latent heat of vaporization
fluid = heat exchange fluid
i = nodal index;
= pipe inside
i,j = coordinate indices
if = latent heat of fusion
in = inlet
l = liquid phase
m,n = coordinate locations
o = pipe inside;
= background or initial
out = outlet
r = thermal radiation;
= refraction
s = solid phase
surf = surface
sw,b = steady-state, borehole wall
t = current time step;
= total
t-∆t = previous time step
w = water;
= injected/extracted water
wb = wet bulb
x,z = coordinate indices
(x,1) = coordinate index for a surface node
1
1. Introduction
1.1. Overview of Ground-Source Heat Pump Systems
Ground-source heat pump (GSHP) systems (also referred to as geothermal heat
pump systems, earth energy systems, and GeoExchange systems) have received
considerable attention in the recent decades as an alternative energy source for residential
and commercial space heating and cooling applications. GSHP applications are one of
three categories of geothermal energy resources as defined by ASHRAE (1995). These
categories are: (1) high-temperature (>302
o
F (>150
o
C) ) electric power production, (2)
intermediate- and low-temperature (<302
o
F (<150
o
C)) direct-use applications, and (3)
GSHP applications (generally <90
o
F (<32
o
C)). The GSHP applications are
distinguished from the others by the fact that they operate at relatively low temperatures.
The term “ground-source heat pump” has become an all-inclusive term to
describe a heat pump system that uses the earth, ground water, or surface water as a heat
source and/or sink. GSHP systems consist of three loops or cycles as shown in Figure 1-
1. The first loop is on the load side and is either an air/water loop or a water/water loop,
depending on the application. The second loop is the refrigerant loop inside a water-
source heat pump. Thermodynamically, there is no difference between the well-known
vapor-compression refrigeration cycle and the heat pump cycle; both systems absorb heat
at a low temperature level and reject it to a higher temperature level. The difference
between the two systems is that a refrigeration application is only concerned with the low
temperature effect produced at the evaporator, while a heat pump may be concerned with
2
both the cooling effect produced at the evaporator as well as the heating effect produced
at the condenser. In these dual-mode GSHP systems, a reversing valve is used to switch
between heating and cooling modes by reversing the refrigerant flow direction. The third
loop in the system is the ground loop in which water or an antifreeze solution exchanges
heat with the refrigerant and the earth.
(a) (b)
Figure 1-1. Schematic of cycles in a GSHP system in (a) cooling mode
and (b) heating mode.
Efficiencies of GSHP systems are much greater than conventional air-source heat
pump systems. A higher COP (coefficient of performance) can be achieved by a GSHP
because the source/sink earth temperature is relatively constant compared to air
Compressor
Expansion
Device
“Load-Side”
Cycle
Refrigerant
Cycle
Ground-Source
Heat Exchanger
Cycle
Cooler Fluid
Supply
Warmer Fluid
Return
Water-Refrigerant
Heat Exchanger
To
Cooling Load
Return from
Cooling Load
Heat
Pump
Reversing
Valve
Water-Refrigerant
or
Air-Refrigerant
Heat Exchanger
Compressor
Expansion
Device
“Load-Side”
Cycle
Refrigerant
Cycle
Ground-Source
Heat Exchanger
Cycle
Warmer Fluid
Supply
Cooler Fluid
Return
Water-Refrigerant
Heat Exchanger
To
Heating Load
Return from
Heating Load
Heat
Pump
Reversing
Valve
Water-Refrigerant
or
Air-Refrigerant
Heat Exchanger
3
temperatures. Additionally, heat is absorbed and rejected through water, which is a more
desirable heat transfer medium because of its relatively high heat capacity.
GSHP systems rely on the fact that, under normal geothermal gradients of about
0.5
o
F/100 ft (30
o
C/km) (Grant et al., 1982), the earth temperature is roughly constant in
a zone extending from about 20 ft (6.1 m) deep to about 150 ft (45.7 m) deep (Hart and
Couvillion, 1986). This constant temperature interval within the earth is the result of a
complex interaction of heat fluxes from above (the sun and the atmosphere) and from
below (the earth interior). As a result, the temperature of this interval within the earth is
approximately equal to the average annual air temperature (Hart and Couvillion, 1986).
Above this zone (less than about 20 feet (6.1 m) deep), the earth temperature is a damped
version of the air temperature at the earth’s surface. Below this zone (greater than about
150 ft (45.7 m) deep), the earth temperature begins to rise according to the natural
geothermal gradient.
ASHRAE (1995) groups GSHP systems into three categories based on the heat
source/sink used. A fourth category is added here for the sake of completeness. These
categories are: (1) ground-water heat pump (GWHP) systems, (2) ground-coupled heat
pump (GCHP) systems, (3) surface water heat pump (SWHP) systems, and (4) standing
column well (SCW) systems. Each of these is discussed in the following subsections.
4
1.1.1. Ground-Water Heat Pump Systems
GWHP systems, also referred to as open-loop systems, are the original type of
GSHP system. The first GWHP systems were installed in the late 1940s (Kavanaugh and
Rafferty, 1997). GWHP systems are not the focus of this thesis, so they will only be
briefly described here.
A schematic of a GWHP system is shown in Figure 1-2. In GWHP systems,
conventional water wells and well pumps are used to supply ground water to a heat pump
or directly to some application. Corrosion protection of the heat pump may be necessary
if ground water chemical quality is poor. The “used” ground water is typically discharged
to a suitable receptor, such as back to an aquifer, to the unsaturated zone (as shown in
Figure 1-2), to a surface-water body, or to a sewer. Design considerations for GWHP
systems are fairly-well established; well-drilling technologies and well-testing methods
have been well-known for decades. Design considerations include: ground-water
availability, ground-water chemical quality, and ground-water disposal method.
The main advantage of GWHP systems is their low cost, simplicity, and small
amount of ground area required relative to other GSHP and conventional systems.
Disadvantages include limited availability and poor chemical quality of ground water in
some regions. With growing environmental concerns over recent decades, many legal
issues have arisen over ground water withdrawal and re-injection in some localities.
5
Figure 1-2. A schematic of a ground-water heat pump system.
1.1.2. Ground-Coupled Heat Pump Systems
GCHP systems, also referred to as closed-loop ground-source heat pump systems,
were pioneered in the 1970s. Their main advantage over their water-well predecessors is
that they eliminate the problems associated with ground water quality and availability and
they generally require much less pumping energy than water well systems because there
is less elevation head to overcome. GCHP systems can be installed at any location where
drilling or earth trenching is feasible.
In GCHP systems, heat rejection/extraction is accomplished by circulating a heat
exchange fluid through a piping system buried in the earth. This fluid is either pure water
or an antifreeze solution and is typically circulated through high-density polyethylene
6
(HDPE) pipe installed in vertical boreholes or horizontal trenches as shown in Figure 1-3.
Thus, these systems are further subdivided into vertical GCHP systems and horizontal
GCHP systems.
1.1.2.1. Vertical Ground-Coupled Heat Pump Systems
Vertical borehole GCHP systems are the primary focus of this entire thesis.
Therefore, they are described in some detail here and their design challenges are
explained, laying the foundation for the motivation of this study.
In vertical borehole GCHP systems, ground heat exchanger configurations
typically consist of one to tens of boreholes each containing a U-shaped pipe through
which the heat exchange fluid is circulated. Some Swedish applications use boreholes
inclined from the vertical. A number of borehole to borehole plumbing arrangements are
possible. Typical U-tubes have a diameter in the range of ¾ in. (19 mm) to 1 ½ in. (38
mm) and each borehole is typically 100 ft (30.5 m) to 300 ft (91.4 m) deep with a
diameter ranging from 3 in. (76 mm) to 5 in. (127 mm). The borehole annulus is
generally backfilled with a material that prevents contamination of ground water.
7
(a)
(b)
Figure 1-3. A schematic of (a) a vertical borehole ground-coupled heat
pump system and (b) horizontal ground-coupled heat pump
system.
8
The design of vertical ground heat exchangers is complicated by the variety of
geological formations and properties that affect their thermal performance (ASHRAE,
1995). Proper subsurface characterization is not economically feasible for every project.
One of the fundamental tasks in the design of a reliable GCHP system is properly sizing
the ground-coupled heat exchanger length (i.e. depth of boreholes). Particularly for large
systems, an extensive effort is made to design the ground loop heat exchangers so that
they are not too large (resulting in too high of a first cost) or too small (resulting in the
building’s thermal load not being met).
In the early days of GCHP technology, the task of sizing the ground-loop heat
exchanger was accomplished using rules of thumb (i.e. 250 feet of bore length per ton of
heating or cooling capacity). These rules were slightly modified on a case-by-case basis
using some estimates of thermal conductivity of the formation or using previous design
experience, but additional costs of more detailed testing or calculations was judged to
outweigh the costs of a conservative design. This relatively simple approach proved to
be successful in most residential and other small applications, but in larger-scale
commercial and institutional applications, some ground-loop heat exchangers failed to
meet their design loads after the first few years of operation. Further, the practice of
greatly over-designing large GCHP systems was found to be unacceptable because the
first costs were simply not competitive with the first costs of conventional systems.
Consequently, intensive research regarding methods to optimize ground-loop heat
exchanger design has been ongoing for the last decade.
9
Simple approaches to sizing the ground-loop heat exchanger in larger-scale
applications are inadequate mainly because the heat transfer processes in the ground are
complicated by thermally interacting boreholes and hourly periodic heat
extraction/injection pulses. Annual heat rejection and heat extraction are usually not
equal and peak temperatures rise or fall over a number of years. As a result, ground-loop
heat exchanger designers need to consider hourly heating and cooling loads of the
building and need to perform some simulation of the ground-loop temperatures over the
life-cycle of the building. Recent research efforts have produced several methods and
computer software programs for this purpose. However, none of the programs consider
the effects of ground water flow on ground-loop heat exchanger performance; these
effects have not been well understood, perhaps because of the lack of relevant
investigations. This is the topic of Chapter 2 of this thesis.
Another challenge in the design of GCHP systems arises from the fact that most
commercial and institutional buildings, even in moderate climates, are generally cooling-
dominated and therefore reject more heat to the ground than they extract over the annual
cycle. This load imbalance may require the heat exchanger length to be significantly
greater than the length required if the annual loads were balanced. As a result, the GSHP
system may be eliminated from consideration early in the design phase of the project due
to excessive first cost. This has given rise to the concept of “supplemental heat rejecters”
or so-called “hybrid GSHP systems”.
10
Supplemental heat rejecters have been integrated into building designs to
effectively balance the ground loads and therefore reduce the necessary length of the
ground-loop heat exchanger. In some applications, the excess heat that would otherwise
build up in the ground may be used for domestic hot water heaters, car washes, and
pavement heating systems. In cases where the excess heat cannot be used beneficially,
conventional cooling towers or shallow ponds can provide a cost-effective means to
reduce heat exchanger length.
Design of these supplemental components adds to the challenge of designing the
overall hybrid GCHP system because of their highly transient nature. Heat rejection
systems are likely to operate more often during the night-time hours or when the building
is not in use. Therefore, it is essential that the hourly (or less) behavior of these systems
be examined during their design phase. These are the topics of Chapters 3 (shallow
ponds) and Chapter 4 (pavement heating systems) of this thesis.
1.1.2.2. Horizontal Ground-Coupled Heat Pump Systems
In horizontal GCHP systems, ground heat exchanger configurations typically
consist of a series of parallel pipe arrangements laid out in dug trenches or horizontal
boreholes about 3 ft (0.91 m) to 6 ft (1.83 m) deep. A number of piping arrangements are
possible. “Slinky” configurations (as shown in Figure 1-3 (b)) are popular and simple to
install in trenches and shallow excavations. In horizontal boreholes, straight pipe
configurations are installed. Typical pipes have a diameter in the range of ¾ in. (19 mm)
11
to 1 ½ in. (38 mm) and about 400 ft (121.9 m) to 600 ft (182.9 m) of pipe is installed per
ton of heating and cooling capacity.
The thermal characteristics of horizontal GCHP systems are similar to those of
vertical ones. The main difference is that horizontal ground-loop heat exchangers are
more affected by weather and air temperature fluctuations because of their proximity to
the earth’s surface. This may result in larger loop temperature fluctuations and therefore
lower heat pump efficiencies. Recent research activities have focussed on using these
systems as supplemental heat rejecters with vertical borehole GCHP systems. A specific
application (i.e. the use of a shallow pavement heating system) is the focus of Chapter 4
of this thesis.
Aside from the invention of the Slinky coil itself and the use of these systems as
supplemental heat rejecters, horizontal systems have received much less attention than
vertical systems with respect to recent research efforts. This may be due to the fact that
vertical systems tend to be preferred in larger applications since much less ground area is
required. Also, since horizontal systems are installed at shallow depths, geologic site
characterization is relatively easier because soils can be readily seen and sampled.
Further, over-conservative designs are not as cost prohibitive as with vertical borehole
designs because of the relatively low installation costs of the heat exchanger pipe.
12
1.1.3. Surface-Water Heat Pump Systems
The third category of GSHP systems is the surface-water heat pump (SWHP)
system. A specific application of SWHP systems (i.e. the use of a shallow pond as a
supplemental heat rejecter in vertical GCHP systems) is the focus of Chapter 3 of this
thesis.
A schematic of a SWHP system is shown in Figure 1-4. The surface-water heat
exchanger can be a closed-loop or open-loop type. Typical closed-loop configurations
are the Slinky coil type (as shown in Figure 1-4) or the loose bundle coil type. In the
closed-loop systems, heat rejection/extraction is accomplished by circulating a heat
exchange fluid through HDPE pipe positioned at an adequate depth within a lake, pond,
reservoir, or other suitable open channel. Typical pipe diameters range from ¾ in. (19
mm) to 1 ½ in. (38 mm) and a length of 100 feet (30.48 m) to 300 feet (91.44 m) per ton
of heating or cooling capacity is recommended by ASHRAE (1995b), depending on the
climate. In open-loop systems, water is extracted from the surface-water body through a
screened intake area at an adequate depth and is discharged to a suitable receptor.
Heat transfer mechanisms and the thermal characteristics of surface-water bodies
are quite different than those of soils and rocks. This subject will be further discussed in
Chapter 3 of this thesis. At the present time, design tools for surface-water heat pump
systems are in their developmental infancy (Kavanaugh and Rafferty, 1997). However,
many successful installations are currently in operation and some guidelines do exist. In
13
short, the loop design involves selection of sufficient length of coil for heat transfer,
specifying adequate diameter piping, specifying a sufficient numbers of parallel loops,
and locating the coil at a proper depth in a water body with adequate thermal capacity.
Figure 1-4. A schematic of a surface-water heat pump system.
1.1.4. Standing Column Well Systems
The fourth category of GSHP systems is known as a standing column well (SCW)
system. These systems are about as old as the ground-water heat pump systems, but are
recently receiving much attention. Since these are not the subject of this thesis, they are
only briefly discussed here.
14
A schematic of an SCW system is shown in Figure 1-5. This type of GSHP draws
water to a heat pump from a standing column of water in a deep well bore and returns the
water to the same well. These systems, primarily installed in hard rock areas, use
uncased boreholes with typical diameters of about 6 in. (15.24 cm) and depths up to 1500
feet (457.2 m). The uncased borehole allows the heat exchange fluid to be in direct
contact with the earth (unlike closed-loop heat exchangers) and allows ground water
infiltration over the entire length of the borehole. Properly sited and designed, SCW
systems have been shown to have significant installation cost savings over closed-loop
GCHP systems. Design guidelines for SCW systems are currently under development.
Figure 1-5. A schematic of a standing column well system.
15
1.2. Thesis Objectives and Scope
This study deals with the modeling of vertical closed-loop and hybrid, ground-
source heat pump systems. The challenges associated with the design of these systems
were discussed in the previous section. A considerable amount of research in the past
decade has been geared toward optimizing the performance of these types of systems and
this study is part of those efforts.
There are three primary objectives of this study. These are to:
(1) examine the effects of ground-water flow on closed-loop GSHP systems,
(2) develop a design and simulation tool for modeling the performance of a
shallow pond as a supplemental heat rejecter with closed-loop GSHP systems,
and
(3) develop a design and simulation tool for modeling the performance of a
hydronic pavement heating system as a supplemental heat rejecter with
closed-loop GSHP systems.
Chapter 2 of this thesis addresses the first objective. Given the time and cost
constraints of finding a suitable site with significant ground water flow, instrumenting
that site, and collecting and analyzing data at the site over a time period of years, a
computer modeling study was conducted as a preliminary assessment of the effects of
ground-water flow on closed-loop GSHP systems. Hydraulic and thermal properties of
soils and rocks were compiled and AQUA3D, a commercially-available numerical
16
ground-water flow and heat transport model developed by Vatnaskil Consulting
Engineers, Inc., Reykjavik, Iceland, was used to simulate the impact of ground-water
flow on the average heat exchange fluid temperature in single and multiple borehole
systems. The impact of ground-water flow on the estimation of soil/rock thermal
conductivity from in-situ test data was also examined.
Chapter 3 of this thesis addresses the second objective. The development and
validation of a design and simulation tool for modeling the performance of a shallow
pond as a supplemental heat rejecter with closed-loop GSHP systems is presented. The
model has been developed in the TRNSYS modeling environment and can therefore be
coupled to other GSHP system component models for short-time step (hourly or
minutely) system analyses. TRNSYS, developed at the University of Wisconsin-
Madison, is a widely-used modular transient thermal systems simulation program where
each system component is described mathematically by a FORTRAN subroutine. By a
simple language, the components are connected together in a manner analogous to piping,
ducting, and wiring in a physical system (Duffie and Beckman, 1991). An example
application of the pond model is also presented.
Chapter 4 of this thesis addresses the third objective. The development and
validation of a design and simulation tool for modeling the performance of a hydronic
pavement heating system as a supplemental heat rejecter with closed-loop GSHP systems
is presented. This model has also been developed in the TRNSYS modeling environment
17
for use in short-time step system analyses. An example application of the pavement
heating model is also presented.
Finally, Chapter 5 of this thesis summarizes the conclusions of the individual
studies.
1.3. The Overall Modeling Approach
Each of the objectives of this thesis deals with the development and/or application
of a model. Since the term “model” can be used loosely, some definitions and
approaches as applicable to this study are described here.
A “model” is a physical or a mathematical representation of an actual system.
This study deals only with mathematical representations of systems. American Society
for Testing and Materials (ASTM) defines a mathematical model as “mathematical
equations expressing the physical system behavior and including simplifying
assumptions”. Mathematical models are solved analytically or numerically using manual
or computer methods.
The overall modeling approach consists of six stages: (1) define the purpose of
the model, (2) develop a conceptual model of the system, (3) develop or define the
mathematical model of the system, (4) implement the solution method, (5) validate the
model, and (6) apply the model. Each of these is described in the following paragraphs.
18
The first stage in the modeling approach is to clearly define the purpose and
objectives of model. This helps to determine the level of detail and accuracy desired by
the model and helps in making decisions regarding the resources needed.
The second stage in the modeling approach is to develop a conceptual model of
the system. ASTM defines a conceptual model as “an interpretation or working
description of the characteristics and dynamics of the physical system”. The purpose of
the conceptual model is to describe the system by a set of assumptions and concepts that
can be evaluated mathematically.
The third stage in the modeling approach is to develop a mathematical model. In
this stage, the conceptual model is translated to mathematical equations that can be
solved for the desired unknowns. The solution method and limiting assumptions or
simplifications are also identified.
The fourth stage in the modeling approach is to implement the solution method to
solve the mathematical equations. With respect to this thesis, this stage involved using
computer methods with a combination of commercially-available software and
FORTRAN code developed specifically for this study.
19
The fifth stage in the modeling approach is to validate the model. With respect to
this thesis, this stage involved comparing model results, where applicable, to an
analytical solution or to experimental data.
The sixth stage in the modeling approach is to finally use the model to analyze the
performance and behavior of actual thermal systems.
20
2. A Preliminary Assessment of the Effects of Ground-Water Flow on
Closed-Loop Ground-Source Heat Pump Systems
2.1. Introduction
One of the fundamental tasks in the design of a reliable ground-coupled heat
pump system is properly sizing the ground-coupled heat exchanger length (i.e. depth of
boreholes). Recent research efforts have produced several methods and commercially-
available design software tools for this purpose (Ingersoll, 1954; Kavanaugh, 1984;
Eskilson, 1987; IGSHPA, 1991; Spitler et al., 1996; and Kavanaugh and Rafferty, 1997).
All of these design tools are based on principles of heat conduction and rely on some
estimate of the ground thermal conductivity and volumetric specific heat. These
parameters are perhaps the most critical to the system design, yet adequately determining
them is often the most difficult task in the design phase.
Methods of determining the thermal properties of the ground have also been the
subject of considerable recent research (Eklof and Gehlin, 1996; and Austin et al., 2000).
Current methods range from estimating values from published data to conducting
laboratory experiments on soil/rock samples to conducting single-borehole, in-situ field
tests. In general, thermal property values derived from in-situ field tests are most
representative because the values are site-specific and a larger volume of material is
evaluated under more realistic conditions than is possible in the laboratory. The typical
field procedure in in-situ tests is to measure the temperature response of a fluid flowing
through a ground heat exchanger in a single borehole. A schematic of the typical field
21
apparatus is shown in Figure 2-1. The fluid temperature at regular time intervals and heat
added to the fluid stream are recorded over the course of the test.
Figure 2-1. Schematic of typical apparatus for in-situ ground thermal
conductivity testing.
Determination of thermal conductivity from temperature-time data is an inverse
problem. Several analytical and numerical methods exist for interpreting the data set.
These methods will not be discussed in detail here, but include the “cylinder-source”
analytical solution (Carslaw and Jaeger, 1946), the “line-source” analytical solution
(Kelvin, 1882; Ingersoll, 1954), numerical solutions (Mei and Emerson, 1985; Muraya et
al., 1996; and Rottmayer et al., 1997), and numerical solutions with parameter estimation
Borehole
U-Tube Heat Exchanger
Ground
Surface
Pipe
Insulation
Water
Heaters
+
Watt
Trans-
ducer
Flow
Meter
Temperature
Sensors
Portable
Unit
Water
Circulating
Pumps
22
(Shonder and Beck, 1999, Austin et al., 2000). Each of these methods is based on
Fourier’s Law of heat conduction.
A further complication in the design of ground-coupled heat pump systems is the
presence of ground water. Where ground water is present, flow will occur in response to
hydraulic gradients and the physical process affecting heat transfer in the ground is
inherently a coupled one of heat diffusion (conduction) and heat advection by moving
ground water. In general, ground-water flow can be expected to be beneficial to the
thermal performance of closed-loop ground heat exchangers since it will have a
moderating effect on borehole temperatures in both heating and cooling modes.
Complications in the borehole field design process due to the presence of flowing
ground water arise from the fact that both current in-situ conductivity test data
interpretation methods and ground-loop heat exchanger design methods are based on
models that only consider heat conduction. Therefore, ground-water flow may impact the
design process in two ways: (1) thermal conductivities derived from in-situ tests may
appear artificially high and (2) borehole fields designed from artificially high thermal
conductivity values may be over- or under-designed. An unusually high thermal
conductivity value was determined from in-situ test data at a site in Minnesota where
significant ground-water flow was believed to occur (Remund, 1998).
The objectives of the work presented in this chapter have been to make a
preliminary examination of the effects of ground water flow on both in-situ ground
23
thermal conductivity measurements and long-term borehole field performance. This has
been attempted firstly by examining the range of hydrogeological conditions that might
be expected and estimating the order of magnitude of the corresponding ground-water
flows. A simple method of examining the importance of heat advection from ground-
water flow is then presented.
A finite-element numerical ground-water flow and heat transport model has been
used to simulate and observe the effects of ground-water flow on the average fluid
temperature in a single U-tube borehole in various geologic materials. The model was
used to simulate several in-situ ground thermal conductivity tests, and thermal
conductivities were derived from these data using a standard approach. For each test
case, the derived thermal conductivities, along with the thermal loads from an actual
building, were used to design a hypothetical multi-borehole field by employing
conventional design tools and procedures. For different sets of hydrogeological
conditions, a numerical model of the whole borehole field was used to simulate its long-
term performance. Conclusions are presented on the ability of conventional design
procedures to correctly predict the long-term performance of closed-loop ground heat
exchangers under different ground-water flow conditions.
24
2.2. Coupled Ground Water Flow and Heat Transport
2.2.1. Ground-Water Flow
Underground water occurs in two zones: the unsaturated zone and the saturated
zone. The term “ground water” refers to water in the saturated zone. The surface
separating the saturated zone from the unsaturated zone is known as the “water table”. At
the water table, water in soil or rock pore spaces is at atmospheric pressure. In the
saturated zone (below the water table), pores are fully saturated and water exists at
pressures greater than atmospheric. In the unsaturated zone, pores are only partially
saturated and the water exists under tension at pressures less than atmospheric. In this
paper, we deal only with water in the saturated zone.
Ground water is present nearly everywhere, but it is only available in usable
quantities in aquifers. An “aquifer” is defined by Driscoll (1986) as a formation, group
of formations, or part of a formation that contains sufficient saturated permeable material
to yield economical quantities of water to wells and springs. Aquifers are described as
being either confined or unconfined. Unconfined aquifers are bounded at their upper
surface by the water table. Confined aquifers are bounded between two layers of lower
permeablity materials. In practice, the boreholes of ground-loop heat exchangers may
partially penetrate several geologic layers.
The governing equation describing flow through porous media is Darcy’s Law:
25
dx
dh
K q − · (2-1)
where q is the specific discharge (volume flow rate per unit of cross-sectional area), K is
the hydraulic conductivity, and h is the hydraulic head. The specific discharge is related
to average linear ground water velocity, v, by:
n
q
v · (2-2)
where n is the porosity and is introduced to account for the difference between the unit
cross-sectional area and the area of the pore spaces through which the ground water flows
(Freeze and Cherry, 1979; Fetter, 1988).
By applying the law of conservation of mass to a control volume and by making
use of Darcy’s Law (Equation 2-1), an equation defining the hydraulic head distribution
can be derived. Transient ground-water flow with constant density can then be expressed
in Cartesian tensor notation as:
*
ij
R
h
K ·

,
_

¸
¸







i i
s
x x t
h
S (2-3)
Since ground water at 110
o
F (43.3
o
C) (an extreme temperature limit expected in GSHP
applications) has a specific gravity of approximately 0.991, the assumption of constant
density flow for low-temperature geothermal applications may be considered valid.
26
2.2.2. Heat Transport in Ground Water
Heat can be transported through a saturated porous medium by the following three
processes:
(1) heat transfer through the solid phase by conduction,
(2) heat transfer through the liquid phase by conduction, and
(3) heat transfer through the liquid phase by advection.
The governing equation describing mass or heat transport in groundwater is a partial
differential equation of the advection-dispersion type (Freeze and Cherry, 1979). By
applying the law of conservation of energy to a control volume, an equation for heat
transport in ground water can be found and can be expressed as:
*
ij
Q
T
D ·

,
_

¸
¸







+


i i i
i
x x x
T
v
t
T
nR (2-4)
where the velocity, v
i
is determined from the solution of Equation 2-3 and T is the
temperature of rock/water matrix. It is the second term in Equation 2-4 that represents
advection of heat by the ground water and couples Equations 2-3 and 2-4 together. If the
ground water velocity is zero, Equation 2-4 reduces to a form of Fourier’s Law of heat
conduction.
The diffusion coefficient tensor D
ij
is modeled here as an effective thermal
diffusivity given by:
27
l lc
k
D
eff *
ρ
· (2-5)
The effective thermal conductivity k
eff
is a volume-weighted average thermal conductivity
of the saturated rock matrix and can be expressed using the porosity as:
s eff
n)k (1 nk k − + ·
l
(2-6)
It is necessary to distinguish between the conductivity and thermal capacity of the water
and soil/rock in this way to account for the fact that heat is stored and conducted through
both the water and soil/rock, but heat is only advected by the water. Similarly, it is
necessary to define a retardation coefficient R accounting for retardation of the thermal
plume which results from differences in the liquid and solid volumetric heat capacities:
l lρ
ρ
c n
n)c (1 1
R
s s
− +
· (2-7)
2.2.3. Typical Hydraulic and Thermal Property Values for Soils and Rocks
In assessing the significance of ground-water flow to closed loop heat exchanger
performance, the question arises as to what locations have significant ground-water flow.
Darcy’s Law indicates that flow is dependent on both the local hydraulic gradient and the
hydraulic conductivity of the geologic material. Heat transfer is dependent on the flow
velocity and the thermal properties of the material. It is therefore useful, in making a
preliminary assessment of the significance of ground-water flow, to consider the range of
naturally-occurring soil and rock properties and possible values of hydraulic gradient.
28
Naturally-occurring ranges of values of hydraulic and thermal properties of soils
and rocks are summarized in Table 2-1. Values of hydraulic gradient are somewhat more
site-specific; the United States Environmental Protection Agency (1996) reports a typical
range of hydraulic gradient values of 0.0001 to 0.05.
Some specific examples of natural ground water velocities include: 1796 ft/yr
(547.5 m/yr) to 7185 ft/yr (2190 m/yr) under a hydraulic gradient of 0.002 to 0.012 in the
Snake River Group basalt, Idaho, USA (Lindholm and Vaccaro, 1988); 361 ft/yr (110
m/yr) in the High Plains sand and gravel aquifer, western central USA (Weeks and
Gutentag, 1988); and 1.3 x 10
-3
ft/yr (4.0 x 10
-4
m/yr) to 1.50 x 10
-2
ft/yr (4.6 x 10
-3
m/yr) in glacial clay soils in Southern Ontario, Canada (Stephenson et al., 1988). Local
pumping activities may further increase ground-water flow rates in aquifers.
The thermal properties of soils and rocks are functions of mineral content,
porosity, and degree of saturation. Of these, porosity may be considered the most
important property simply because of the origin and nature of soils and rocks. Rocks
originate under higher heat and pressure environments than soils and consequently
generally possess lower porosities. Lower porosities in rocks result in higher contact area
between grains and therefore higher thermal conductivities than soils, regardless of
mineral content. For saturated materials, increased porosity results in increased heat
capacities and therefore lower thermal diffusivities.

T
A
B
L
E

2
-
1
.
T
y
p
i
c
a
l

V
a
l
u
e
s

o
f

H
y
d
r
a
u
l
i
c

a
n
d

T
h
e
r
m
a
l

P
r
o
p
e
r
t
i
e
s

o
f

S
o
i
l
s

a
n
d

R
o
c
k
s
P
o
r
o
u
s

M
e
d
i
u
m
H
y
d
r
a
u
l
i
c

P
r
o
p
e
r
t
i
e
s
T
h
e
r
m
a
l

P
r
o
p
e
r
t
i
e
s
H
y
d
r
a
u
l
i
c

C
o
n
d
u
c
t
i
v
i
t
y
*
P
o
r
o
s
i
t
y
*
V
e
l
o
c
i
t
y
*
*
T
h
e
r
m
a
l

C
o
n
d
u
c
t
i
v
i
t
y
*
*
*
V
o
l
u
m
e
t
r
i
c

H
e
a
t

C
a
p
a
c
i
t
y*
*
*
(
K
)
(
n
)
(
v
)
(
k
)
(
ρ
s
c
s
)
f
t
/
s
(
-
-
)
f
t
/
y
r
B
t
u
/
h
r
-
f
t
-
o
F
B
t
u
/
f
t
3
-
o
F
(
m
/
s
)
(
m
/
y
r
)
(
W
/
m
-
o
C
)
(
J
/
m
3
-
o
C
)
R
a
n
g
e
G
e
o
m
e
t
r
i
c
R
a
n
g
e
A
r
i
t
h
m
e
t
i
c
A
r
i
t
h
m
e
t
i
c
R
a
n
g
e
A
r
i
t
h
m
e
t
i
c
A
v
e
r
a
g
e
A
v
e
r
a
g
e
R
a
n
g
e
A
v
e
r
a
g
e
A
v
e
r
a
g
e
S
o
i
l
s
G
r
a
v
e
l
9
.
8
4
E
-
0
4
-
9
.
8
4
E
-
0
2
9
.
8
4
E
-
0
3
0
.
2
4
-
0
.
3
8
0
.
3
1
1
.
0
0
E
+
0
4
0
.
4
0
-
0
.
5
2
0
.
4
6
-
-
-
-
-
2
.
0
9
E
+
0
1
3
.
0
0
E
-
0
4
-
3
.
0
0
E
-
0
2
3
.
0
0
E
-
0
3
3
.
0
5
E
+
0
3
(
0
.
7
0
)
-
(
0
.
9
0
)
(
0
.
8
0
)
-
-
-
-
-
(
1
.
4
0
E
+
0
6
)
S
a
n
d

(
c
o
a
r
s
e
)
3
.
0
E
-
0
6
-
2
.
0
E
-
0
2
2
.
4
E
-
0
4
0
.
3
1
-
0
.
4
6
0
.
3
8
5
1
.
9
8
E
+
0
2
0
.
4
0
-
0
.
5
2
0
.
4
6
-
-
-
-
-
2
.
0
9
E
+
0
1
(
9
.
0
E
-
0
7
)
-
(
6
.
0
E
-
0
3
)
(
7
.
3
E
-
0
5
)
(
6
.
0
1
E
+
0
1
)
(
0
.
7
0
)
-
(
0
.
9
0
)
(
0
.
8
0
)
-
-
-
-
-
(
1
.
4
0
E
+
0
6
)
S
a
n
d

(
f
i
n
e
)
6
.
6
E
-
0
7
-
6
.
6
E
-
0
4
2
.
1
E
-
0
5
0
.
2
6
-
0
.
5
3
0
.
4
0
1
.
6
6
E
+
0
1
0
.
4
0
-
0
.
5
2
0
.
4
6
-
-
-
-
-
2
.
0
9
E
+
0
1
(
2
.
0
E
-
0
7
)
-
(
2
.
0
E
-
0
4
)
(
6
.
3
E
-
0
6
)
(
5
.
0
5
E
+
0
0
)
(
0
.
7
0
)
-
(
0
.
9
0
)
(
0
.
8
0
)
-
-
-
-
-
(
1
.
4
0
E
+
0
6
)
S
i
l
t
3
.
3
E
-
0
9
-
6
.
6
E
-
0
5
4
.
6
E
-
0
7
0
.
3
4
-
0
.
6
1
0
.
4
7
5
3
.
0
8
E
-
0
1
0
.
6
9
-
1
.
3
9
1
.
0
4
3
.
5
8
E
+
0
1
-
4
.
9
2
E
+
0
1
4
.
2
5
E
+
0
1
(
1
.
0
E
-
0
9
)
-
(
2
.
0
E
-
0
5
)
(
1
.
4
E
-
0
7
)
(
9
.
4
0
E
-
0
2
)
(
1
.
2
0
)
-
(
2
.
4
0
)
(
1
.
8
0
)
(
2
.
4
0
E
+
0
6
)
-
(
3
.
3
0
E
+
0
6
)
(
2
.
8
5
E
+
0
6
)
C
l
a
y
3
.
3
E
-
1
1
-
1
.
5
E
-
0
8
7
.
1
E
-
1
0
0
.
3
4
-
0
.
6
0
0
.
4
7
4
.
7
8
E
-
0
4
0
.
4
9
-
0
.
6
4
0
.
5
6
4
.
4
7
E
+
0
1
-
5
.
3
7
E
+
0
1
4
.
9
2
E
+
0
1
(
1
.
0
E
-
1
1
)
-
(
4
.
7
E
-
0
9
)
(
2
.
2
E
-
1
0
)
(
1
.
4
6
E
-
0
4
)
(
0
.
8
5
)
-
(
1
.
1
0
)
(
0
.
9
8
)
(
3
.
0
0
E
+
0
6
)
-
(
3
.
6
0
E
+
0
6
)
(
3
.
3
E
+
0
6
)
R
o
c
k
s
L
i
m
e
s
t
o
n
e
,

D
o
l
o
m
i
t
e
3
.
3
E
-
0
9
-
2
.
0
E
-
0
5
2
.
5
E
-
0
7
0
-
0
.
2
0
0
.
1
0
8
.
0
2
E
-
0
1
0
.
8
7
-
1
.
9
1
1
.
3
9
3
.
1
7
E
+
0
2
-
8
.
2
0
E
+
0
1
1
.
9
9
E
+
0
2
(
1
.
0
E
-
0
9
)
-
(
6
.
0
E
-
0
6
)
(
7
.
7
E
-
0
8
)
(
2
.
4
4
E
-
0
1
)
(
1
.
5
0
)
-
(
3
.
3
0
)
(
2
.
4
0
)
(
2
.
1
3
E
+
0
7
)
-
(
5
.
5
0
E
+
0
6
)
(
1
.
3
4
E
+
0
7
)
K
a
r
s
t

L
i
m
e
s
t
o
n
e
3
.
3
E
-
0
6
-
3
.
3
E
-
0
2
3
.
3
E
-
0
4
0
.
0
5
-
0
.
5
0
0
.
2
7
5
3
.
7
6
E
+
0
2
1
.
4
4
-
2
.
4
8
1
.
9
6
3
.
1
7
E
+
0
2
-
8
.
2
0
E
+
0
1
1
.
9
9
E
+
0
2
(
1
.
0
E
-
0
6
)
-
(
1
.
0
E
-
0
2
)
(
1
.
0
E
-
0
4
)
(
1
.
1
5
E
+
0
2
)
(
2
.
5
0
)
-
(
4
.
3
0
)
(
3
.
4
0
)
(
2
.
1
3
E
+
0
7
)
-
(
5
.
5
0
E
+
0
6
)
(
1
.
3
4
E
+
0
7
)
S
a
n
d
s
t
o
n
e
9
.
8
E
-
1
0
-
2
.
0
E
-
0
5
1
.
4
E
-
0
7
0
.
0
5
-
0
.
3
0
0
.
1
8
2
.
5
1
E
-
0
1
1
.
3
3
-
3
.
7
6
2
.
5
4
3
.
1
7
E
+
0
1
-
7
.
4
6
E
+
0
1
5
.
3
1
E
+
0
1
(
3
.
0
E
-
1
0
)
-
(
6
.
0
E
-
0
6
)
(
4
.
2
E
-
0
8
)
(
7
.
6
5
E
-
0
2
)
(
2
.
3
0
)
-
(
6
.
5
0
)
(
4
.
4
0
)
(
2
.
1
3
E
+
0
6
)
-
(
5
.
0
0
E
+
0
6
)
(
3
.
5
6
E
+
0
6
)
S
h
a
l
e
3
.
3
E
-
1
3
-
6
.
6
E
-
0
9
4
.
6
E
-
1
1
0
-
0
.
1
0
0
.
0
5
2
5
2
.
7
9
E
-
0
4
0
.
8
7
-
2
.
0
2
1
.
4
4
3
.
5
4
E
+
0
1
-
8
.
2
0
E
+
0
1
5
.
8
7
E
+
0
1
(
1
.
0
E
-
1
3
)
-
(
2
.
0
E
-
0
9
)
(
1
.
4
E
-
1
1
)
(
8
.
5
0
E
-
0
5
)
(
1
.
5
0
)
-
(
3
.
5
0
)
(
2
.
5
0
)
(
2
.
3
8
E
+
0
6
)
-
(
5
.
5
0
E
+
0
6
)
(
3
.
9
4
E
+
0
6
)
F
r
a
c
t
u
r
e
d

I
g
n
e
o
u
s
2
.
6
E
-
0
8
-
9
.
8
E
-
0
4
5
.
1
E
-
0
6
0
-
0
.
1
0
0
.
0
5
3
.
2
1
E
+
0
1
1
.
4
7
-
3
.
8
3
2
.
6
5
-
-
-
-
-
3
.
2
8
E
+
0
1


a
n
d

M
e
t
a
m
o
r
p
h
i
c
(
8
.
0
E
-
0
9
)
-
(
3
.
0
E
-
0
4
)
(
1
.
5
E
-
0
6
)
(
9
.
7
8
E
+
0
0
)
(
2
.
5
0
)
-
(
6
.
6
0
)
(
4
.
5
8
)
-
-
-
-
-
(
2
.
2
0
E
+
0
6
)
U
n
f
r
a
c
t
u
r
e
d

I
g
n
e
o
u
s
9
.
8
E
-
1
4
-
6
.
6
E
-
1
0
8
.
0
E
-
1
2
0
-
0
.
0
5
0
.
0
2
5
1
.
0
1
E
-
0
4
1
.
4
7
-
3
.
8
3
2
.
6
5
-
-
-
-
-
3
.
2
8
E
+
0
1


a
n
d

M
e
t
a
m
o
r
p
h
i
c
(
3
.
0
E
-
1
3
)
-
(
2
.
0
E
-
1
0
)
(
2
.
4
E
-
1
2
)
(
3
.
0
9
E
-
0
5
)
(
2
.
5
0
)
-
(
6
.
6
0
)
(
4
.
5
8
)
-
-
-
-
-
(
2
.
2
0
E
+
0
6
)
N
o
t
e
s
:
T
h
e
r
m
a
l

c
o
n
d
u
c
t
i
v
i
t
y

v
a
l
u
e
s

a
r
e

t
a
k
e
n

t
o

r
e
p
r
e
s
e
n
t

t
h
a
t

o
f

m
a
t
e
r
i
a
l
s

i
n

t
h
e

d
r
y

c
o
n
d
i
t
i
o
n
.
*

h
y
d
r
a
u
l
i
c

c
o
n
d
u
c
t
i
v
i
t
y

a
n
d

p
o
r
o
s
i
t
y

d
a
t
a

f
r
o
m
D
o
m
e
n
i
c
o

a
n
d

S
c
h
w
a
r
t
z

(
1
9
9
0
)
.
*
*

v

i
s

t
h
e

a
v
e
r
a
g
e

l
i
n
e
a
r

g
r
o
u
n
d

w
a
t
e
r

v
e
l
o
c
i
t
y

b
a
s
e
d

o
n

a
n

a
s
s
u
m
e
d

g
r
a
d
i
e
n
t

o
f

0
.
0
1

f
t
/
f
t

(
m
/
m
)
.
*
*
*

t
h
e
r
m
a
l

p
r
o
p
e
r
t
y

d
a
t
a

f
r
o
m

H
e
l
l
s
t
r
o
m

(
1
9
9
1
)
.


F
o
r

s
e
d
i
m
e
n
t
a
r
y

r
o
c
k
s
,

H
e
l
l
s
t
r
o
m

l
i
s
t
s

o
n
l
y

c
s
.

I
n

t
h
e
s
e

c
a
s
e
s
,

a

d
e
n
s
i
t
y

o
f

2
5
0
0

k
g
/
m
3

i
s

a
s
s
u
m
e
d
.
30
The porosity of soils and rocks can also be an important controlling influence on
hydraulic conductivity (Freeze and Cherry, 1979). Materials with higher porosity
generally also have higher hydraulic conductivity. However, this correlation does not
hold for fine-grained soils (see Table 2-1.). Porosity and hydraulic conductivity of soils
and rocks can be increased by so-called “secondary porosity” which is attributed to
solution channels (i.e. in karst limestone) or to fracturing (i.e. in rocks and cohesive
soils).
2.2.4. Conduction Versus Convection in Geologic Materials
It has already been noted that it is the presence of advection that distinguishes the
heat transfer regime under ground-water flow conditions from that of heat conduction
alone. Some assessment of the significance of the flow can be made by considering the
order of magnitude of the advection of heat compared to conduction (diffusion).
A dimensionless parameter describing conduction versus convection is the Peclet
number (Pe). In this application, the Peclet number expresses the transport of heat by
bulk fluid motion to the heat transported by conduction. Domenico and Schwartz (1990)
define Pe for heat transport in ground water as:
eff
k
L q c
Pe
l l ρ
· (2-8)
The term L is defined as some characteristic length dependent on the situation.
According to Bear (1972), L can be chosen as any length dimension, so long as it is
consistent with other comparisons. In principle, advection becomes significant when Pe
31
is of order one. The exact value of Pe at which advection becomes significant is slightly
dependent on the choice of L.
The Peclet number has been used to quantify the relative importance of
mechanical (or advective) dispersion versus molecular diffusion in mass transport in
ground water. Many studies have been conducted and the data have been summarized by
Bear (1972). In short, when the characteristic length was chosen as mean grain size,
diffusion is the process controlling mass transport at Peclet numbers less than about 0.4.
At Peclet numbers in the range of 0.4 to 5, a transition occurs where mechanical
dispersion (or advective dispersion) and diffusion are of the same order of magnitude.
Above a Peclet number of about 5, mechanical dispersion (or advective dispersion)
dominates. No similar studies conducted for heat transport have been found.
An analysis of the Peclet number using the typical hydraulic and thermal values
of soils and rocks presented in Table 2-1 may be used to assess the role of ground water
flow in the design of closed-loop ground heat exchangers. The characteristic length
could conceivably be chosen as (1) a typical borehole spacing or (2) the length of the
borehole field in the direction of flow. The calculated Peclet numbers are listed in Table
2-2 using a typical borehole spacing of 14.8 ft (4.5 m) and assuming the fluid property
values of ρ
l
, c
l
, and k
l
as 62.4 lb/ft
3
(1000 kg/m
3
), 1.0 Btu/lb-
o
F (4180 J/kg-
o
C), and
0.347 Btu/hr-ft-
o
F (0.60 W/m-
o
C).
32
A review of the data presented in Table 2-2 reveals that heat advection by ground
water flow is significant process contributing to heat transfer in coarse-grained soils
(sands and gravels) and in rocks exhibiting secondary porosities (fracturing and solution
channels). When the characteristic length is defined as the borehole spacing, Peclet
numbers exceeding 1 exist only for sands, gravels, and karst limestones. It is possible
however, that even when the Peclet number is of order one or higher, the effects of the
ground-water flow on the temperature response may not be seen within the normal time
scale of an in-situ thermal conductivity test. This is one of the reasons for conducting
numerical borehole field simulations for the duration of several years.
TABLE 2-2.
Peclet Numbers Corresponding to Typical Values of Hydraulic and Thermal
Properties of Soils and Rocks
Porous Medium Peclet Number
where L = a typical
borehole spacing of
14.8 ft (4.5 m)
[--]
Soils
Gravel 5.72E+02
Sand (coarse) 1.34E+01
Sand (fine) 1.15E+00
Silt 1.28E-02
Clay 3.24E-05
Rocks
Limestone, Dolomite 5.92E-03
Karst Limestone 5.28E+00
Sandstone 1.77E-03
Shale 1.05E-06
Fractured Igneous 6.32E-02
and Metamorphic
Unfractured Igneous 1.00E-07
and Metamorphic
33
2.2.5. Numerical Ground-Water Flow and Heat Transport Models
In assessing the effects of ground-water flow on U-tube heat exchanger
performance, one is mainly interested in the temperature of the heat exchange fluid.
Therefore, modeling of the U-tubes in some detail is important in this problem. The heat
exchange fluid temperature is affected by the transient building thermal loads in addition
to the heat transfer in the porous medium around the borehole. Consequently, this
problem is characterized by an irregular model domain with time-varying boundary
conditions and is best handled by a numerical model.
Numerous commercially-available and public domain numerical software codes
exist for modeling mass and/or heat transport in ground water. Of these, the following
8 were selected for a more detailed review for potential application to this project:
• 3DFEMFAT (3-Dimensional Finite Element Method Flow and Transport) by G.
Yeh, Pennsylvania State University. This code was developed to simulate mass
transport in variably-saturated porous media. Density-dependent flow can also be
simulated.
• AQUA3D by Vatnaskil Consulting Engineers, Reykjavik, Iceland. This code is also a
three-dimensional, finite-element code. It was developed mainly for simulation of
mass-transport problems but allows easy adaptation of boundary conditions to model
heat transport without density-dependent ground-water flow.
34
• FEFLOW (Finite Element FLOW) by WASY Institute for Water Resources
Planning and Systems Research, Ltd., Berlin, Germany. This code is also a three-
dimensional, finite-element code. It is capable of simulating both mass and heat
transport in variable-density ground-water flow systems.
• Flowpath II by Waterloo Hydrologic, Inc. (WHI), Waterloo, Ontario. This code is a
two-dimensional finite difference code. It was developed originally for simulation of
ground-water flow problems only; contaminant-transport simulation capabilities,
mainly in the horizontal plane, have been recently added.
• HST3D (Heat and Solute Transport in 3 Dimensions) by USGS, Denver, Colorado.
This code is a three-dimensional finite-difference code. It is capable of simulating
mass and heat transport in variable-density ground-water flow systems. It was
developed mainly for simulating problems involving waste injection into aquifers.
• MT3D
96
(Mass Transport in 3 Dimensions) by S.S. Papadopulos & Associates, Inc.,
Bethesda, Maryland. This code is also a three-dimensional finite-difference code. It
solves the mass transport equation only and requires a solution to the ground-water
flow equation from another code. It was developed for simulating contaminant-
transport problems in ground water.
35
• SUTRA (Saturated-Unsaturated TRAnsport) by United States Geological Survey
(USGS), Denver, Colorado. This code is a two-dimensional finite-element code. It is
capable of simulating mass or energy transport in variably-saturated, variable-density
ground-water flow systems. It was mainly developed as a cross-sectional model for
simulating salt-water intrusion into fresh-water aquifers.
• SWIFT (Sandia Waste Isolation Flow and Transport) by HSI GeoTrans, Sterling,
Virginia. This code is a three-dimensional finite-difference code. It is capable of
simulating mass and heat transport in variable-density ground-water flow systems in
porous or fractured media. It was developed mainly for simulating problems
involving deep-well radioactive waste injection into geologic repositories.
In the code selection process, particular attention was paid to the following items: (1)
the type of boundary conditions handled by the code, (2) the solution scheme employed
by the code, (3) verification of the code, and (4) cost. Each of these points is described
in more detail below.
The type of boundary conditions handled by the code was perhaps the most important
consideration in the code selection process. For simulation of periodic heat extraction or
heat rejection to the ground, the selected code needed to be capable of handling time-
varying, heat flux boundary conditions. Further, since a relatively large number of time-
varying data were to be used as input, the selected code needed to be capable of reading
time-varying conditions from an external file.
36
The software codes that were reviewed for this project employ either finite-difference
methods (FDM) of finite-element methods (FEM) to solve the partial differential
equations describing heat/mass transport in ground water (Equations 2-3 and 2-4). The
solution scheme was considered important for two main reasons. First, FEM offers an
advantage over FDM in the ability to represent complex or irregular geometries (i.e.
circular U-tubes in a rectangular domain). Second, there has been controversy in the
literature over advantages of FEM over FDM in solving the advection-dispersion
equation (Equation 2-4). In general, experience has shown the FEM to be generally
superior to FDM in solution stability (Wang and Anderson (1982) and Mercer and Faust
(1986)). Consequently, codes employing an FEM solution scheme were preferred.
Documented verification of the code was an important consideration since it often
requires years to find and fix bugs in these types of software programs. All of the 8
codes listed above have been originally developed in the 1980s and many validation
examples exist.
The cost of the code was also a main consideration in the selection process since the
project had an allocated budget for software.
37
2.3. The Numerical Model
2.3.1. Model Description
The computer code selected for this study was AQUA3D. It is a commercially-
available software package originally developed in 1983 by Vatnaskil Consulting
Engineers of Reykjavik, Iceland. The partial differential Equations 2-3 and 2-4 are
discretized spatially by a Galerkin finite element method using triangular elements with
linear weighting functions (Vatnaskil, 1998). The temporal term of the equations is dealt
with by first order backward differencing in time. AQUA3D does not allow for the
explicit representation of the heat transport equation, but provides a general form of the
mass transport equation (Equation 2-4). Temperatures were in fact calculated by suitable
choice of the coefficients of the mass transport equation and corresponding adaptation of
the boundary conditions.
The finite element ground-water flow and mass/heat transport model was used in
this study as the primary means of assessing the effects of ground-water flow on closed-
loop heat exchangers. Use of a numerical model allows a wide range of conditions to be
examined and is the only practical means of modeling a whole borehole field. In each
test case, a uni-directional flow field was imposed over the whole numerical domain. As
the flow was assumed to be fully-saturated and within homogeneous geologic material, it
was only necessary to use a two-dimensional model.
38
2.3.2. The Finite Element Mesh
Finite element meshes for a single borehole geometry, and for a complete
borehole field geometry have been constructed using triangular elements. Nodal spacing
was kept relatively fine around the pipe walls where the steepest temperature gradients
were expected. The mesh for the single borehole geometry was constructed within a
square domain and consisted of 465 nodes, as shown in Figure 2-2.
Figure 2-2. Finite element mesh representing a single borehole.
3.46 in. (8.8 cm)
14.4 ft (4.4 m)
39
A mesh for a four-by-four configuration borehole field was constructed by using
the single borehole mesh (Figure 2-2) as the basis for the mesh at each borehole and by
expanding the mesh in the direction of ground-water flow as shown in Figure 2-3. This
mesh consisted of 4532 nodes.
Figure 2-3. Finite element mesh representing a 16 borehole field.
2.3.3. Boundary Conditions
Two sets of boundary conditions are required: one set for the flow model and one
set for the transport model.
1640 ft (500 m)
3
2
8

f
t

(
1
0
0

m
)
57.7 ft (17.6 m)
40
In the flow model, first-type or fixed-value (Dirichlet) boundary conditions were
set on the left-hand and right-hand boundaries in order to impose a fixed hydraulic
gradient across the domain. Second-type or fixed-gradient (Neumann) boundary
conditions were set on the upper and lower boundaries of the flow domain and were
specified as zero flux. This work assumes that no ground-water recharge takes place
across the water table within the model domain. In the transport model, Dirichlet
boundary conditions were set on all four sides of the model domain. These conditions
represent fixed background or far-field temperatures.
In order to impose the ground thermal loads as boundary conditions at the U-tube
pipe walls, some adaptation of the usual boundary condition was required. This arises
from the use of the mass transport equation to model heat transport. First, a zero flux
condition for the mass (heat) transport equation was applied at each of the sixteen nodes
forming each pipe wall. The required heat flux is imposed using a source term in the
ground-water flow equation at these nodes (representing injection of the heating/cooling
water). The flow injected, V
*
, was set negligibly small (3.53 x 10
-19
ft
3
/s [1.0 x 10
-20
m
3
/s]) so as not to disturb the ground-water flow field. The temperature of this injection
flow, T
w
was set to achieve the required heat input (the ground thermal load), so that,
* V c
* q
T
w
l l ρ
≈ (2-9)
The values of ρ
l
and c
l
are taken as constants of 62.4 lb/ft
3
(1000 kg/m
3
) and 1.0 Btu/lb-
o
F (4180 J/kg-
o
C). The average temperature of the heat exchange fluid in each borehole
is taken as the average of the nodal temperatures of the 32 nodes defining the U-tube pipe
41
in each borehole. Where single borehole cases were simulated, the heat input per pipe
node, q
*
, was set at a fixed value representative of in-situ thermal conductivity test
conditions. Where the whole borehole field was modeled, q
*
was determined from the
time-varying building loads.
2.3.4. Validation of the Numerical Model
In order to check the accuracy of the AQUA3D model and the implementation of
the boundary conditions, an appropriate analytical solution was sought. Numerous
analytical solutions have been developed for the advection-dispersion equation (Equation
2-4). However, these are mostly specific to pollutant-transport problems (T is replaced
by solute concentration in Equation 2-4) involving point or line sources with uniform
concentration in time. Fetter (1988) and Bear (1972) summarize solutions for boundary
and initial conditions describing situations that are commonly found in nature. The
literature gives little to no attention to analytical solutions describing the explicit
transport of heat in ground water.
The most appropriate analytical solution found was that described by Eskilson
(1987) for steady-state heat extraction from a borehole in a ground-water flow field.
However, Eskilson’s solution contains some approximations.
Eskilson (1987) describes the steady-state temperature at the borehole wall (T
sw,b
)
as:
42
¹
¹
¹
;
¹
¹
¹
¹
'
¹
1
]
1

¸

,
_

¸
¸

,
_

¸
¸
+ ·
l
H
P
r
H
k H
q
T T
w
b s
om b sw
2
ln
2
1
*
,
π
(2-10)
The logarithmic term gives the steady-state resistance and
,
_

¸
¸
l
H
P
w
is a correction term
accounting for ground-water flow which is approximated as:
( ) 3 ln
2
1
2
ln − + ·
,
_

¸
¸
γ
l l
H H
P
w
(2-11)
for l defined as:
q c
k
s
l l
l
ρ
2
· where l >>r
b
(2-12)
Comparisons of the steady-state temperature at the wall of a pipe were made
using both the AQUA3D and the Eskilson (1987) solution. The following data were used
as inputs: a pipe diameter of 0.787 in. (2 cm) (i.e. a single leg of a typical U-tube), a pipe
depth of 250 ft (76.2 m), soil thermal property values of sand from Table 2-1, water
properties as described above, a far-field temperature of 63
o
F (17.2
o
C), and a heat flux of
8530 Btu/hr (2500 W). The temperature predicted by AQUA3D at steady-state time and
Eskilson’s (1987) solution were 101.4
o
F (38.54
o
C) and 99.59
o
F (37.55
o
C), respectively.
In terms of temperature increase above the far-field temperature, the percent difference in
error between the two solutions is 4.9%. This error was considered to be acceptable.
43
2.4. Results And Discussion
2.4.1. Single-Borehole Simulations
The numerical model was initially used to determine average borehole
temperatures for a range of soil and rock types over a two-year simulation time. The
objective was to examine trends in heat exchanger performance with increasing Peclet
number. The hydraulic and thermal property inputs are those listed in Table 2-1 and a
hydraulic gradient of 0.01 was assumed. A constant heat flux of 8530 Btu/hr (2500 W)
was applied on a U-tube in a 250 ft (76.2 m) deep borehole. The initial temperature and
first-type boundary conditions were set at 63
o
F (17.2
o
C). The model domain is that
shown in Figure 2-2. A simulation time of two years with a time step of 5 days was used
for these simulations. For comparison purposes, simulations were made for each case but
with zero ground-water flow.
Plots of average borehole fluid temperature versus time for three example
geologic materials are shown in Figure 2-4. A review of these plots reveals that a
“typical” ground water flow rate in a coarse sand dramatically lowers the average
borehole fluid temperature when compared to the zero-flow case. After a one-year
period, the average fluid temperature in the borehole is approximately 15
o
F (8.3
o
C) lower
than the average fluid temperature in the borehole where no ground-water flow was
simulated, and appears to have reached a steady state. A small reduction in peak
temperature is shown for the case of fine sand. However, “typical” ground water flow
44

Figure 2-4. Average borehole fluid temperature vs. time for (a) coarse sand,
(b) fine sand, and (c) shale.
(a) coarse sand
60
70
80
90
100
110
120
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Time (years)
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e

F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
15.6
21.1
26.7
32.2
37.8
43.3
48.9
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e

F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
No Ground Water Flow
Ground Water Flow Rate
= 196.8 ft/yr (60 m/yr)
(b) fine sand
60
70
80
90
100
110
120
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Time (years)
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e

F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
15.6
21.1
26.7
32.2
37.8
43.3
48.9
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e

F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
No Ground Water Flow
Ground Water Flow Rate
= 16.6 ft/yr (5.05 m/yr)
(c) shale
60
70
80
90
100
110
120
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Time (years)
A
v
g
.

B
o
r
e
h
o
l
e

F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
15.6
21.1
26.7
32.2
37.8
43.3
48.9
A
v
g
.

B
o
r
e
h
o
l
e

F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
No Ground Water Flow
Ground Water Flow Rate =
2.8E-4 ft/yr (8.5E-5 m/yr)
45
rates in rocks such as granites, limestones, dolomites, and shales were found to have a
negligible effect on the average borehole fluid temperature.
The trends shown in these results are in agreement with the previous Peclet
number analysis. At Peclet numbers of order one or higher, advection of heat by flowing
ground water is a significant process contributing to heat transfer in the ground. At
Peclet numbers of order less than one, conduction is the dominant heat transfer process
and enhancement to the heat exchanger performance is negligible.
2.4.2. Simulated In-Situ Thermal Conductivity Tests
The second objective of the single-borehole simulations was to determine the
effects of ground-water flow (in a material where ground-water flow is expected to be
significant) on the interpretation of data from in-situ ground thermal conductivity tests.
The previous results showed the effects of ground-water flow to be most significant in the
cases of gravel and coarse sand. Accordingly, the simulated in-situ thermal conductivity
test calculations have been based on coarse sand properties.
As previously discussed, in-situ thermal conductivity tests involve the application
of a steady heat flux to a test borehole along with the measurement of the temperature
response of the circulating water. These data are used either with an analytical model or
with a numerical model and parameter estimation technique to arrive at a value of
thermal conductivity of the soil/ rock formation. Here, the borehole temperature response
46
is calculated for a range of ground-water flows using the AQUA3D model. These data
have been analyzed in exactly the same way as if the data had been measured in-situ.
Hence, ‘effective’ thermal conductivities have been estimated for different flow
conditions.
In-situ test conditions were modeled by applying a constant heat flux of 8530
Btu/hr (2500 W) on a U-tube in a 250 ft (76.2 m) deep borehole. The simulation time
periods were 50 hours and one week, corresponding to typical durations of in-situ ground
thermal conductivity tests. The model time step was 2.5 minutes. The initial temperature
and first-type boundary conditions were set at 63
o
F (17.2
o
C) and the model domain is that
shown in Figure 2-2. Model input hydraulic and thermal property values are those listed
in Table 2-1 for a coarse sand, except the ground water flow velocity was varied from a
“typical” value of 196.8 ft/yr (60 m/yr) to a more extreme value of 1968.5 ft/yr (600
m/yr) by adjusting the hydraulic conductivity value. Twelve cases were simulated as
listed in Table 2-3.
Resulting temperature responses for the 12 cases are plotted in Figure 2-5. A
review of Figure 2-5 shows that ground water flow in a coarse sand significantly impacts
the average borehole fluid temperature over the time scales of an in-situ ground thermal
conductivity test. Two noteworthy conclusions can be drawn from these simulations: (1)
as ground water velocity increases, the time to reach steady-state conditions decreases
and (2) as ground water velocity increases, the steady-state temperature decreases. Also,
the deviation from the zero-flow condition can be seen to be dependent on the duration of
47
the test; temperatures are further reduced with increasing duration. Hence, the duration
of the test could be expected to have an influence on the estimated thermal conductivity
derived from an in-situ test.
TABLE 2-3
Case Summary of Simulated In-Situ Ground Thermal Conductivity Tests
The effective thermal conductivity values predicted by the Austin et al. (2000)
model are plotted against the corresponding ground-water flow velocity for each of the
two in-situ test simulation times (50 hours and 1 week) in Figure 2-6. The actual values
are listed by case number in Table 2-4. A review of these results shows that as ground
water flow velocity increases, the predicted effective thermal conductivity values from a
conduction-based model are significantly different, depending on the duration of the
simulated test. These values are “effective” values since they include the effects of
ground water advection. However, at this stage of the design process, it is not clear if the
Case Simulation Time Period Ground Water Flow Velocity
1 50 hours No Ground Water Flow
2 50 hours 196.8 ft/yr (60 m/yr)
3 50 hours 393.7 ft/yr (120 m/yr)
4 50 hours 787.4 ft/yr (240 m/yr)
5 50 hours 1574.8 ft/yr (480 m/yr)
6 50 hours 1968.5 ft/yr (600 m/yr)
7 1 week No Ground Water Flow
8 1 week 196.8 ft/yr (60 m/yr)
9 1 week 393.7 ft/yr (120 m/yr)
10 1 week 787.4 ft/yr (240 m/yr)
11 1 week 1574.8 ft/yr (480 m/yr)
12 1 week 1968.5 ft/yr (600 m/yr)
48
50-hour data set or the 1-week data set produces more representative values, or if either
data set produces representative values at all.
Figure 2-5. Average borehole fluid temperatures for the 12 simulated in-situ
ground thermal conductivity test cases in a coarse sand with ground
water velocities ranging from 0 to 1968 ft/yr (600 m/yr) for (a) 50
hours and (b) one week.
(a)
60
65
70
75
80
85
90
0 10 20 30 40 50
Time (hrs)
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e
F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
15.6
18.3
21.1
23.9
26.7
29.4
32.2
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e
F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Case 1
Case 2
Case 3
Case 4 Case 5
Case 6
(b)
60
65
70
75
80
85
90
95
0 20 40 60 80 100 120 140 160
Time (hrs)
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e
F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
15.6
18.3
21.1
23.9
26.7
29.4
32.2
35.0
A
v
e
r
a
g
e

B
o
r
e
h
o
l
e
F
l
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Case 7
Case 8
Case 9
Case 10
Case 12
Case 11
49
Figure 2-6. Predicted effective ground thermal conductivity values versus
ground-water flow velocity for 50-hour and 1-week simulated in-
situ thermal conductivity tests.
2.4.3. Borehole Field Simulations
In order to investigate the effects of ground-water flow on borehole field
performance and system design procedures further, the predicted ground thermal
conductivities were used to design a borehole field for a test building. The test building
was an actual building located in north-central Oklahoma. This building is a single-story
office building with 8 thermal zones and has a predominant demand for cooling. The
hourly building loads were determined for one year using building energy simulation
software (BLAST, 1986). The building loads were then converted to ground loads under
0
2
4
6
8
10
12
14
16
0.0 328.1 656.2 984.2 1312.3 1640.4 1968.5
Ground Water Velocity (ft/yr)
P
r
e
d
i
c
t
e
d

k

V
a
l
u
e

(
B
t
u
/
h
r
-
f
t
-
o
F
)
0.0
3.5
6.9
10.4
13.8
17.3
20.8
24.2
27.7
0 100 200 300 400 500 600
Ground Water Velocity (m/yr)
P
r
e
d
i
c
t
e
d

k

V
a
l
u
e

(
W
/
m
-
o
C
)
50 hour simulated in-situ test
1 week simulated in-situ test
50
the assumption that all heat pumps in the system have a constant coefficient of
performance of 4.0. The ground loads for this building are shown in Figure 2-7.
TABLE 2-4
Summary of Borehole Field Design Parameters for Each Test Case
Case Simulation Ground-Water Ground Thermal Design Borehole
Number Duration Flow Rate Conductivity Depth
Predicted by Predicted by
Numerical Model of Design Software of
Austin et al. (2000) Spitler et al. (1996)
(hours) ft/yr Btu/hr-ft-
o
F ft
(m/yr) (W/m-
o
C) (m)
1 50 0 0.643 239.98
(1.11) (73.15)
2 50 196.85 0.650 238.56
(60.00) (1.12) (72.71)
3 50 393.70 0.731 224.10
(120.00) (1.26) (68.31)
4 50 787.40 1.146 171.56
(240.00) (1.98) (52.29)
5 50 1574.80 3.657 87.24
(480.00) (6.33) (26.59)
6 50 1968.50 6.074 61.58
(600.00) (10.51) (18.77)
7 168 0 0.625 243.86
(1.08) (74.33)
8 168 196.85 0.691 230.86
(60.00) (1.20) (70.37)
9 168 393.70 0.962 191.58
(120.00) (1.66) (58.39)
10 168 787.40 2.250 115.91
(240.00) (3.89) (35.33)
11 168 1574.80 8.229 48.02
(480.00) (14.24) (14.64)
12 168 1968.50 15.107 26.90
(600.00) (26.14) (8.20)
51
Figure 2-7. Hourly ground loads for the test building. Heating load is shown
negative, representing heat extracted from the ground; cooling load
is shown positive, representing heat rejected to the ground.
Borehole field designs were produced for each of the twelve test cases. This was
done with GLHEPRO, a commercially-available ground-loop heat exchanger design
software tool developed by Spitler et al. (1996). A 16 borehole field (four-by four
boreholes in a square pattern) was deemed adequate for the test building ground loads
(Figure 2-7). The monthly loads and peak hourly loads are generally input in the design
software. For this study, no peak hourly loads were specified for the sake of the
computational time required for the subsequent borehole field simulations (see discussion
below). Peak design entering fluid temperatures to the heat pump were specified at 90
o
F
(32.2
o
C) maximum and 35
o
F (1.7
o
C) minimum. The borehole depths were sized for 20-
years of operation.
-200
-150
-100
-50
0
50
100
150
200
250
0 2190 4380 6570 8760
Time (hours)
G
r
o
u
n
d

L
o
a
d
s

(
k
B
t
u
/
h
r
)
-59
-44
-29
-15
0
15
29
44
59
G
r
o
u
n
d

L
o
a
d
s

(
k
W
)
52
For each test case, the corresponding effective thermal conductivity shown in
Figure 2-6 and Table 2-4 was input into the ground-loop heat exchanger design software
(GLHEPRO). The borehole depths predicted by GLHEPRO are plotted against the
corresponding ground water flow velocity for each of the two in-situ test simulation times
(50 hours and 1 week) in Figure 2-8. The values are listed by case number in Table 2-4.
Figure 2-8. Design borehole depths versus ground-water flow velocity for 50-
hour and 1-week simulated in-situ thermal conductivity tests.
AQUA3D was further used to simulate the long-term performance of each
borehole field designed from the simulated in-situ ground thermal conductivity test cases.
The model domain was that previously described for the multi-borehole field simulations
shown in Figure 2-3. The simulation time for all cases was 10 years and the model time
step was 5 days. The simulated heat flux at the internal boundary nodes defining the U-
tube pipes was a time-varying source corresponding to the monthly ground loads for the
0
50
100
150
200
250
0.0 328.1 656.2 984.3 1312.3 1640.4 1968.5
Ground Water Velocity (ft/yr)
D
e
s
i
g
n

B
o
r
e
h
o
l
e

D
e
p
t
h

(
f
t
)
0.0
15.2
30.5
45.7
61.0
76.2
0 100 200 300 400 500 600
Ground Water Velocity (m/yr)
D
e
s
i
g
n

B
o
r
e
h
o
l
e

D
e
p
t
h

(
m
)
50 hour simulated in-situ test
1 week simulated in-situ test
53
test building. Hydraulic and thermal property inputs for each borehole field case number
is the same as the corresponding single-borehole case number, except for the borehole
depths, which are those listed in Table 2-4. Each 10-year simulation required
approximately 60 hours of computation time on a personal computer with a 233
megahertz pentium II processor.
Annual maximum and minimum peak temperatures are plotted for each case in
Figure 2-9. Examination of the cases with no ground-water flow (cases 1 and 7) shows
annual rises in peak temperature typical of cooling-dominated buildings. After the
second year, all of the cases with ground-water flow show no changes in minimum and
maximum temperatures from year to year.
Some notable differences can be seen between the borehole field designs based on
50-hour test data compared to one week test data. This is shown by cases 5 and 6 which
represent thermal conductivity values determined from a 50-hour test at ground water
flow velocities of 1574.8 ft/yr (480 m/yr) and 1968.5 ft/yr (600 m/yr), respectively, and
by cases 11 and 12 which are for the same flow rates but based on thermal conductivity
values determined from one-week test data. The thermal conductivity values determined
in cases 11 and 12 are unrealistically high and consequently the design borehole depths
are too shallow; the result is that the maximum peak temperature of the simulated
borehole field in both cases exceeds the maximum design temperature during the first
year. This implies that for in-situ test cases where the average borehole fluid temperature
reaches steady-state in a relatively short time (as demonstrated by case 4/10, case 5/11,
54
and case 6/12 in Figure 2-5), increasing the duration of the in-situ test produces decreased
confidence in the accuracy of the effective thermal conductivity value determined from
the test.
Figure 2-9. Annual maximum (a) and (c) and minimum (b) and (d) average
borehole fluid temperatures for the 16 borehole field simulations.
(a)
70
80
90
100
110
0 1 2 3 4 5 6 7 8 9 10
Time (years)
M
a
x
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
21.1
26.7
32.2
37.8
43.3
M
a
x
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Case 1 Case 2 Case 3 Case 4 Case 5 Case 6
Maximum Design Temperature
(b)
25
35
45
55
65
0 1 2 3 4 5 6 7 8 9 10
Time (years)
M
i
n
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
-3.9
1.7
7.2
12.8
18.3
M
i
n
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Case 1 Case 2 Case 3 Case 4 Case 5 Case 6
Minimum Design Temperature
(c)
70
80
90
100
110
0 1 2 3 4 5 6 7 8 9 10
Time (years)
M
a
x
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
21.1
26.7
32.2
37.8
43.3
M
a
x
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Case 7 Case8 Case 9 Case 10 Case 11 Case 12
Maximum Design Temperature
(d)
25
35
45
55
65
0 1 2 3 4 5 6 7 8 9 10
Time (years)
M
i
n
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
-3.9
1.7
7.2
12.8
18.3
M
i
n
i
m
u
m

T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Case 7 Case 8 Case 9 Case 10 Case 11 Case 12
Minimum Design Temperature
55
Except for cases 11 and 12, the annual maximum and minimum temperatures fell
within the design conditions. Having followed conventional design procedures, it is
interesting to note from Figure 2-9 (a) that it is the cases where the ground-water flow is
moderate (cases 2, 3, and 4) that are most over-designed. These cases have maximum
peak temperatures of about 74
o
F (23.3
o
C), some 16
o
F (8.9
o
C) below the maximum design
temperature. Considerable drilling cost savings could be seen in cases like this where
shallower borehole depths could have been adequate. It is at higher flow rates (cases 5
and 6) that the peak temperature was closest to the original design condition after ten
years. This illustrates the non-linearity introduced into the design problem by the
presence of advection. It also illustrates the difficulty in adapting conventional design
methods to accurately size closed-loop ground heat exchangers in cases with significant
ground-water flow.
The temperature field predicted by the numerical model for case 8 is shown in
Figure 2-10 in the form of a series of contour plots over the 10 year simulation period.
The data are plotted for the end of September, when the average annual ground
temperatures are the greatest (i.e. at the end of the cooling season).
56
Figure 2-10. Temperature contours for Case 8 for the end of September of years 1, 2, 5,
and 10.
Temperature Key
(
o
C)
end of Sept. Year 1
end of Sept. Year 2
end of Sept. Year 5
end of Sept. Year 10
ground water flow rate = 60 m/yr (197 ft/yr)
500 m
100 m
57
A further feature that is shown in the predicted temperature field (Figure 2-10) is
the development of a peak in the temperatures immediately downstream of the borehole
field after year 2. This arises from the advection of the heat rejected to the ground at the
boreholes during the previous year. In the contours plotted for year 10, thermal plumes
from previous years can be distinguished.
2.5. Concluding Remarks and Recommendations for Future Work
Using a compilation of “typical” hydraulic and thermal properties of soils and
rocks, a preliminary analysis of the effects of ground-water flow on the design and
performance of closed-loop ground-coupled heat pump systems has been made. A
simple but useful method of assessing the relative importance of heat conduction in the
ground versus heat advection by moving ground water is demonstrated through the use of
the dimensionless Peclet number.
A finite-element numerical ground-water flow and heat transport model was used
to simulate and observe the effects of ground-water flow on the heat transfer from a
single U-tube closed-loop ground heat exchanger in various geologic materials. From
these simulations and from a Peclet number analysis, it appears that it is only in geologic
material with high hydraulic conductivities, such as coarse-grained soils (sands and
gravels) and in rocks exhibiting secondary porosities such fractures and solution
channels, that ground-water flow could be expected to have a significant effect on closed-
loop heat exchanger performance.
58
The effect of ground-water flow on in-situ thermal conductivity test results has
been examined by numerically simulating test conditions around a single borehole under
different flow conditions. These data were analyzed as if it came from real in-situ
sources to arrive at effective thermal conductivity values. As expected, in all cases of
ground-water flow, these values were artificially high. Results from one week test data
have been shown to be no more reliable than data from 50-hour tests.
The finite-element numerical ground water flow and heat transport model was
also used to simulate the 10-year performance of borehole fields designed from
application of conventional design procedures using the derived thermal conductivity
data. For coarse-grained sands, the presence of moderate ground-water flows had the
effect of removing the year-by-year increase in ground temperature found in systems
where there is no ground-water flow. The borehole fields designed using conventional
methods and the derived effective thermal conductivities were generally over-designed.
However, in some cases at very high ground-water flow rates, temperatures were found
to rise above the design criteria.
From this preliminary assessment of the effects of ground-water flow, it appears
difficult to adapt results from current design and in-situ measurement methods to fully
account for ground-water flow conditions. Over the last decade, considerable progress
has been made in developing both in-situ test methods and design procedures for
borehole field design for situations where there is no ground-water flow. Research would
59
be required in a number of areas before the same progress could be made to deal with the
situations of ground-water flow. These include:
• Identification of suitable numerical and/or analytical models that include ground-
water flow and could be used to analyze in-situ test data.
• Experimental investigation of potential in-situ test data analysis methods at sites with
significant ground-water flow.
• Identification of suitable design methods, or adaptations to existing methods, that
could be used for closed-loop ground heat exchanger design.
• Development of design guidelines and software tools that could be used by practicing
engineers for in-situ testing and system design tasks in situations of significant
ground-water flow.
60
3. A Model for Simulating the Performance of a Shallow Pond as a
Supplemental Heat Rejecter with Closed-Loop Ground-Source Heat
Pump Systems
3.1. Introduction
Commercial buildings and institutions are generally cooling-dominated, and
therefore reject more heat than they extract over the annual cycle. In order to adequately
dissipate the imbalanced annual loads, the required ground-loop heat exchanger lengths
are significantly greater than the required length if the annual loads were balanced.
Consequently, under these circumstances, ground-source heat pump systems may be
eliminated from consideration during the feasibility study phase of the HVAC design
process because of excessive first cost.
To effectively balance the ground loads and reduce the necessary size of the
ground loop heat exchanger, supplemental components can be integrated into the ground-
loop heat exchanger design. GSHP systems that incorporate some type of supplemental
heat rejecter are commonly referred to as hybrid GSHP systems. In some applications,
the excess heat that would otherwise build up in the ground may be used for domestic hot
water heaters, car washes, and pavement heating systems. In cases where the excess heat
cannot be used beneficially, shallow ponds can provide a cost-effective means to balance
the thermal loading to the ground and reduce heat exchanger length.
The objective of the work presented in this chapter has been to develop a design
and simulation tool for modeling the performance of a shallow pond that can be usefully
61
and cost-effectively integrated into a ground-source heat pump system as a supplemental
heat rejecter. The pond model has been developed in the TRNSYS modeling
environment (SEL, 1997) and can be coupled to other GSHP system component models
for short-time step (hourly or minutely) system analyses. The model has been validated
by comparing simulation results to experimental data. As an example of the model’s
applicability, GSHP system simulation results are presented for a commercial building
located in Tulsa, Oklahoma with a hypothetical closed-loop GSHP system with and
without a shallow pond supplemental heat rejecter.
3.2. Heat Transfer In Ponds
3.2.1. General Overview
Pertinent concepts of heat transfer in ponds and lakes have been summarized by
many sources. Dake and Harleman (1969) conducted studies of thermal stratification in
lakes and addressed the overall thermal energy distribution in lakes. ASHRAE (1995),
ASHRAE (1995b), and Kavanaugh and Rafferty (1997) describe heat transfer in lakes in
relation to their use as heat sources and sinks.
Solar energy is identified as the main heating mechanism for ponds and lakes.
The main cooling mechanism is evaporation. Thermal radiation can also account for a
significant amount of cooling during night hours. Convective heating or cooling to the
atmosphere is less significant. Natural convection of water due to buoyancy effects is the
primary mechanism for heat transfer within a surface water body. Conduction heat
62
transfer to the ground is generally a relatively insignificant process, except in cases where
the water surface is frozen.
Shallow ponds are generally thermally unstratified. Natural stratification of
deeper ponds and lakes is due to buoyancy forces and to the fact that water is at its
greatest density at 39.2
o
F (4
o
C). Therefore, over the annual cycle, water in deeper ponds
will completely over-turn. Thermal stratification in ponds is also dictated by inflow and
outflow rates or ground water seepage rates. If inflow and outflow rates are high enough,
the pond will not stratify. Consequently, thermal stratification occurs only in ponds and
lakes that are relatively deep, generally greater than 20 - 30 ft (6.1 - 9.1 m), with low
inflow rates. The relevant heat transfer mechanisms occurring within shallow ponds are
illustrated in Figure 3-1.
3.2.2. Existing Pond and Lake Models
Several mathematical and computer models have been developed for simulation
of lakes used as heat sinks/sources and for solar ponds.
Raphael (1962) developed a numerical model for determining the temperature of
surface water bodies as heat sinks for power plants. Thermal stratification of the water
body was not considered. Input data to the model included weather data and inflow and
outflow data for the water body. Raphael reported that the model successfully predicted
the temperature changes in a river used as a heat sink for a power plant.
63
Figure 3-1. Heat transfer mechanisms in shallow ponds.
Jobson (1973) developed a mathematical model for water bodies used as heat
sinks for power plants. Thermal stratification of the water body was not considered. The
results of that work showed that the heat transfer at the water/air interface is highly
dependent on the natural water temperature and the wind speed..
Cantrell and Wepfer (1984) developed a numerical model for evaluating the
potential of shallow ponds for dissipating heat from buildings. The model takes weather
data and building cooling load data as inputs and computes the steady-state pond
temperature using an energy balance method. Thermal stratification of the pond was not
considered. The model showed that a 3-acre (12,141 m
2
), 10-feet (3.048 m) deep pond in
Subsurface
Conduction
Wind
Incident
Solar Radiation
Convection
E v a p o r a t I o n
Thermal Radiation
Coil Heat Rejection/Absorption
Fluid from heat pump
Fluid return to
heat pump
Subsurface Conduction
Groundwater Discharge
64
Cleveland, Ohio could reject 1000 tons (3516 kW) of thermal energy with a maximum
increase in pond temperature of about 5
o
F (2.78
o
C) over a daily cycle.
Rubin et al. (1984) developed a model for solar ponds. The purpose of a solar
pond is to concentrate heat energy from the sun at the pond bottom. This is accomplished
by suppressing natural convection within the pond induced by bottom heating, usually by
adding a brine layer at the pond bottom. As a result, solar ponds have three distinct zones
as described by Newell (1984): (1) a top layer which is stagnated by some method and
acts as a transparent layer of insulation, (2) a middle layer which is usually allowed to be
mixed by natural convection, and (3) a lower layer where solar energy is collected. The
model of Rubin et al. (1984) applied an implicit finite difference scheme to solve a one-
dimensional heat balance equation on a solar pond. Large-scale convective currents in
the pond were assumed to be negligible while small-scale convective currents were
handled by allowing the coefficient of heat diffusion to vary through the pond depth.
Solar radiation was modeled as an exponentially decaying function through the pond
depth. The model successfully predicted seasonal variations in solar pond temperatures.
Srinivasan and Guha (1987) developed a model similar to the model of Rubin et
al. (1984) for solar ponds. The Srinivasan and Guha (1987) model consisted of three
coupled differential equations, each describing a thermal zone within the solar pond.
Solar radiation in each zone is computed as a function of depth. The model also
successfully predicted seasonal variations in solar pond temperatures with various heat
extraction rates.
65
Pezent and Kavanaugh (1990) developed a model for lakes used as heat sources
or sinks with water-source heat pumps. The model essentially combined the models of
Srinivasan and Guha (1987) to handle stratified cases and of Raphael (1962) to handle
unstratified cases. As such, thermal stratification of a lake could be handled in the
summer months, when lakes are generally most stratified, and neglected in the winter
months, when lakes are generally unstratified. The model is driven by monthly average
bin weather data and handles both heat extraction and heat rejection. With no heat
extraction or rejection, the model favorably predicted a lake temperature profile in
Alabama. The temperatures within the upper zone of the lake (the epilimnion) and the
lower zone of the lake (the hypolimnion) were predicted to within 4
o
F (2.22
o
C) and
approximately 1
o
F (0.55
o
C), respectively. However, the model had some difficulty in
matching the intermediate zone (the thermocline), perhaps due to the fact that this zone
possesses moving boundaries (unlike the boundaries of a solar pond which are more
distinct). As concluded by Pezent and Kavanaugh (1990), a numerical method is
necessary to more accurately predict the thermocline profile.
The model presented in this paper is based on the assumption that thermal
gradients in shallow ponds are negligible, especially during times of heat rejection. This
model is developed in the TRNSYS modeling environment and can be coupled to other
component models for larger system simulations. Further, this model allows the pond
performance to be simulated on hourly or minutely time scales.
66
3.3. Experimental Methods
3.3.1. Pond Description and Data Collection
Pond construction and data collection for this study have been conducted by
researchers affiliated with the Division of Engineering Technology at the Oklahoma State
University.
Two ponds were constructed on a test site at the north end of the campus. The
ponds are rectangular with a plan area of 40 ft (12.19 m) by 3 ft (0.91 m). Each pond was
constructed with vertical sidewalls with one of the ponds being 2 feet (0.61 m) deep and
the other being 3.5 ft (1.07 m) deep. The walls and the bottom of each pond were
constructed of reinforced concrete, approximately 8 in. (20.3 cm) thick.
Heat was rejected to each pond by circulating heated water through a “slinky”
heat exchanger (a pipe coiled in a circular fashion such that each loop overlaps the
adjacent loop) installed in each pond. Each slinky pipe was made of high-density
polyethylene plastic and is 500 feet (152.40 m) long with a nominal diameter of ¾ in.
(0.019 m). The pipe was coiled such that the resulting slinky heat exchanger was 40 ft
(12.19 m) long with a diameter of 3 ft (0.91 m) and a 10-in. (0.254 m) pitch (the
separation distance between the apex of each successive loop). In the 2-ft (0.61-m) deep
pond, the slinky heat exchanger was positioned horizontally within the pond at a depth of
approximately 10 in. (0.254 m). In the 3.5-ft (1.07-m) deep pond, the slinky heat
67
exchanger was positioned vertically within the pond along the center-line of the long axis
of the pond.
The temperature of the pond water was measured by thermistors positioned at
four locations within the pond: (1) near the pond surface at the center of the slinky, (2)
below the slinky at its center, (3) near the pond surface at the end opposite from the
supply end, and (4) below the slinky at the end of the pond opposite from the supply end.
Slinky supply and return water temperatures were measured by thermistors embedded in
the slinky header. Each system also included a flow meter, a water heating element, and
a watt transducer. All sensor information was recorded by the data acquisition system at
time intervals of 6 minutes.
The tests were controlled to maintain a set supply water temperature by heating
the supply water if the temperature fell below a set point. Two set point temperatures
were used in this study: 90
o
F (32.2
o
C) in the summer season and 75
o
F (23.9
o
C) in the
winter season.
3.3.2. Weather Instrumentation and Data Collection
Weather data for this study were obtained from the Oklahoma Mesonet
(mesoscale network), which is a weather station network consisting of weather
monitoring sites scattered throughout Oklahoma. Depending on the weather parameter,
data are recorded at time intervals ranging from 3 seconds to 30 seconds and averaged
over 5-minute observation intervals.
68
Weather data at 15-minute intervals for the Stillwater monitoring station were
acquired for the time periods of interest for this study. The Stillwater station is located
approximately 1 mile from the test pond site. Data for the following parameters were
obtained: wind speed, wind direction, air temperature, relative humidity, and solar
radiation. Further details of the weather station network may be found in Elliott et al.
(1994).
3.4. Model Development
3.4.1. Governing Equations
The governing equation of the model is an overall energy balance on the pond
using the lumped capacitance (or lumped parameter) approach:
dt
dT
c V q q
p out in
ρ · − (3-1)
where q
in
is the heat transfer to the pond, q
out
is the heat transfer from the pond, V is the
pond volume, ρ is the density of the pond water, c
p
is the specific heat capacity of the
pond water, and
dt
dT
is the rate of change of temperature of the pond water. This
approach assumes that temperature gradients within the water body are negligible.
Considering the heat transfer mechanisms shown in Figure 3-1, Equation 3-1 can be
expressed to describe the rate of change in average pond temperature as:
69
p
fluid n evaporatio r groundwate ground convection thermal solar
c V
q q q q q q q
dt
dT
ρ
+ + + + + +
· (3-2)
where, q
solar
is the solar radiation heat gain to the pond, q
thermal
is the thermal radiation
heat transfer at the pond surface, q
convection
is the convection heat transfer at the pond
surface, q
ground
is the heat transfer to/from the ground in contact with the pond, q
groundwater
is the heat transfer due to groundwater inflow or outflow, q
evaporation
is the heat/mass
transfer due to evaporation at the pond surface, and q
fluid
is the total heat transfer to/from
the heat exchange fluid flowing in all spools or coils in the pond. Each of the heat
transfer terms used in the above equation is defined below.
3.4.1.1. Solar Radiation Heat Gain
Solar radiation heat gain (q
solar
) is the net solar radiation absorbed by the pond. It
is assumed that all solar radiation incident on the pond surface becomes heat gain except
for the portion reflected at the surface.
To determine the reflected component of solar radiation, the angle of incidence
(θ) of the sun’s rays is first computed at each time step from correlations given by
Spencer (1971), Duffie and Beckman (1991), and ASHRAE (1997). The angle of
refraction (θ
r
) of the sun’s rays at the pond surface is given by Snell’s Law. The
reflectance (ρ’) is computed from:
τ τ ρ − ·
a
' (3-3)
70
where τ

is the transmittance of solar radiation by the pond surface and the subscript ‘a’
refers to the absorbed component. These are computed after Duffie and Beckman (1991)
as:
r
d
a
e
θ
µ
τ
cos
' −
· (3-4)
and
r
d
e
r
r
r
r
θ
µ
τ
cos
'
||
||
1
1
1
1
2
1


,
_

¸
¸
+

+
+

· (3-5)
where µ’ is the extinction coefficient for water, d is the pond depth, r
||
represents the
parallel component of unpolarized radiation, and r

represents the perpendicular
component of unpolarized radiation which are computed after Duffie and Beckman
(1991) as:
) ( tan
) ( tan
2
2
||
θ θ
θ θ
+

·
r
r
r (3-6)
) ( sin
) ( sin
2
2
θ θ
θ θ
+

·

r
r
r (3-7)
Finally, the amount of solar radiation absorbed by the pond (q
solar
) is expressed as:
pond solar
A I q ) ' 1 ( ρ − · (3-8)
where I is the solar radiation flux incident on the pond surface (here, the total reflectance
is approximated by the beam reflectance) and A
pond
is the area of the pond surface. The
71
model also accepts solar radiation in the form of beam (I
b
) and diffuse (I
d
) components, in
which case I is computed from:
d b
I I I + · θ cos (3-9)
3.4.1.2. Thermal Radiation Heat Transfer
This heat transfer mechanism accounts for heat transfer at the pond surface due to
thermal or long-wave radiation. This model uses a linearized radiation coefficient (h
r
)
defined as:
3
2
4

,
_

¸
¸ +
·
sky pond
r
T T
h εσ (3-10)
where ε is the emissivity coefficient of the pond water, σ is the Stefan-Boltzmann
constant, T
pond
is the pond temperature in absolute units, and T
sky
is the sky temperature in
absolute units. T
sky
is computed from relationship given by Bliss (1961) which relates
sky temperature to the dew point temperature (T
dp
) and the dry bulb temperature (T
db
):
4
1
250
8 . 0

,
_

¸
¸
+ ·
dp
db sky
T
T T (3-11)
where all temperatures are in absolute units. The model uses the TRNSYS psychrometric
subroutine to compute T
dp
if it is unknown. T
dp
is computed from either of the following
pairs of state properties: (1) dry bulb temperature (T
db
) and wet bulb temperature (T
wb
) or
(2) T
db
and relative humidity. The thermal radiation heat transfer (q
thermal
) is then
computed by:
72
) (
pond sky pond r thermal
T T A h q − · (3-12)
3.4.1.3. Convection Heat Transfer at the Pond Surface
This mechanism accounts for heat transfer at the pond surface due to free and
forced convection. Several empirical formulations exist for determining the convection
coefficient for different geometries. For a pond surface, correlations for a horizontal flat
plate are the most applicable.
In free convection heat transfer, the Nusselt Number (Nu) is often correlated to
the Rayleigh Number (Ra), which describes the relative magnitude of the buoyancy and
viscous forces in the fluid:
να
β
3
) (
Ra
L T g ∆
· (3-13)
where g is the acceleration due to gravity, α is the thermal diffusivity of air, β is the
volumetric thermal expansion coefficient of air, ν is the kinematic viscosity of air, ∆T
refers to the temperature difference between the pond and the air, and L is the
characteristic length. The thermal properties α, β, and ν are all evaluated at the film
temperature which can be approximated as the average of the air and pond temperatures.
In the model, the thermal properties of air are computed at each time step using
correlations given by Irvine and Liley (1984). For horizontal flat plates, the characteristic
length (L) can be defined as the ratio of the area (A) to the perimeter (P) (Incropera and
DeWitt, 1996):
73
P
A
L · (3-14)
In external free convection flows over a horizontal flat plate, the critical Rayleigh
Number is about 10
7
. Therefore, two empirical relations for the Nusselt number are used
in the model as described by Incropera and DeWitt (1996) for free convection from the
upper surface of a heated plate or the lower surface of a cooled plate:
4
1
Ra 54 . 0 Nu · (10
4
< Ra < 10
7
– laminar flow) (3-15a)
and
3
1
Ra 15 . 0 Nu · (10
7
> Ra > 10
11
– turbulent flow) (3-15b)
The convection coefficient (h
c
) for free convection can then be determined from:
L
k
h
c
Nu
· (3-16)
where k is the thermal conductivity of air evaluated at the film temperature as with the
other thermal properties described above and L is the characteristic length described by
Equation 3-14.
In forced convection heat transfer, Nu is a function of the Reynolds (Re) and
Prandtl (Pr) Numbers. The Reynolds number is described as:
74
ν
vL
· Re (3-17)
where v is the wind speed, ν is the kinematic viscosity of air, and the characteristic length
(L) is now defined by the length dimension over which the wind blows and is determined
from the pond orientation (degrees from north) and wind direction. The Prandtl Number
is defined as:
k
c
p
µ
· Pr (3-18)
where c
p
is the specific heat capacity of air, µ is the dynamic viscosity of air, and k is the
thermal conductivity of air, all evaluated at the air film temperature.
For external forced convection over a flat plate (i.e. the pond surface), the critical
Reynolds Number is approximately 10
5
(Incropera and DeWitt, 1996). Therefore, two
empirical relations for the Nusselt number are used in the model as described by
Incropera and DeWitt (1996) for forced convection over a flat plate:
3
1
2
1
Pr Re 664 . 0 Nu · (laminar flow regime) (3-19a)
and
3
1
5
4
Pr Re 037 . 0 Nu · (mixed and turbulent flow) (3-19b)
The convection coefficient (h
c
) for forced convection can then be determined by Equation
3-16 with the characteristic length value described by Equation 3-14 for a horizontal flat
plate.
75
Finally, the convection heat transfer at the pond surface (q
convection
) is computed
by:
) (
pond air pond c convection
T T A h q − · (3-20)
where T
air
is the ambient air temperature and h
c
is taken as the maximum of the free
convection coefficient and the forced convection coefficient. This practice of choosing
the larger of the free and forced convection coefficients is recommended by Duffie and
Beckman (1991) and McAdams (1954) and is used in the absence of additional
experimental evidence regarding combined free and forced convection.
3.4.1.4. Heat Transfer to the Ground
This heat transfer mechanism accounts for heat conduction to/from the soil or
rock in contact with the sides and the bottom of the pond. This mechanism of heat
transfer is highly site-specific and complex and depends on many factors such as
soil/rock thermal properties, climatic factors, pond geometry, and thermal loading
history. In this model, a semi-empirical approach developed by Hull et al. (1984) was
chosen to determine heat losses/gains from the bottom and sides of the pond. Hull et al.
(1984) used a three-dimensional numerical code to compute steady-state ground heat
losses from solar ponds of varying sizes, geometries, and sidewall insulation types.
Hull et al. (1984) expresses ground heat losses from any pond as a function of the
pond area, pond perimeter, the ground thermal conductivity (k
ground
), and the distance
76
from the pond bottom to a constant temperature sink. For practical purposes, the constant
temperature sink can be taken as the groundwater table (Kishore and Joshi, 1984). For a
rectangular pond with vertical side walls, a heat transfer coefficient for ground heat
transfer (U
ground
) can be computed from:

,
_

¸
¸
+

,
_

¸
¸

·
pond
pond ground
pond r groundwate
ground
ground
A
P k
d d
k
U 37 . 1 999 . 0 (3-21)
where k
ground
is the thermal conductivity of the ground, d
groundwater
is the depth to the water
table or the constant source/sink from the ground surface, d
pond
is the pond depth, and
P
pond
is the pond perimeter. The conduction heat transfer between the ground and the
pond is then given by:
) (
pond r groundwate pond ground ground
T T A U q − · (3-22)
It is recognized that the above conduction heat transfer model is a relatively
simple representation of the true transient behavior of heat transfer in the ground.
However, ground heat conduction is a relatively minor process affecting the overall heat
transfer within the pond as compared to other processes.
3.4.1.5. Heat Transfer Due to Ground Water Seepage
This heat transfer mechanism accounts for inflows and outflows of ground water
to the pond. Although ground water contributions may not be expected in shallow heat
rejecter ponds, this heat transfer mechanism can be used to account for other inflows and
outflows, such as make-up water or rain water.
77
The volumetric groundwater flow rate (Q) is computed by Darcy’s Law:
) ) ( (
pond r groundwate pond pond
A d d P Ki Q + − · (3-23)
where K is the hydraulic conductivity of the soil/rock surrounding the pond and i is the
hydraulic gradient. The heat transfer contribution from ground water (q
groundwater
) is then
given by:
) (
pond r groundwate p r groundwate
T T c Q q − · ρ (3-24)
where ρ and c
p
represent the density and specific heat capacity of ground water. These
properties of ground water are computed from relationships given in the Handbook of
Chemistry and Physics (CRC, 1980).
3.1.4.6. Heat Transfer Due to Evaporation
This heat transfer mechanism is the most important one contributing to pond
cooling. This model uses the j-factor analogy to compute the mass transfer of
evaporating water (
w
m′ ′ & ) at the pond surface:
) (
surf air d w
w w h m − · ′ ′ & (3-25)
where h
d
is the mass transfer coefficient, w
air
is the humidity ratio of the ambient air, and
w
surf
represents the humidity ratio of saturated air at the pond surface. The w
surf
term is
78
computed using the TRNSYS psychrometric subroutine by setting T
db
and T
wb
equal to
the pond temperature. The w
air
term may also be computed using the TRNSYS
psychrometric subroutine depending on what two state properties of the air are known.
The model accepts the following pairs of state properties for the calculation of w
air
if it is
unknown: (1) T
db
and T
wb
, (2) T
db
and relative humidity or (3) T
db
and T
dp
. The mass
transfer coefficient (h
d
) is defined using the Chilton-Colburn analogy as:
3
2
Le
p
c
d
c
h
h · (3-26)
where h
c
is the convection coefficient defined previously, c
p
is the specific heat capacity
of the air evaluated at the pond-air film temperature, and Le is the Lewis number. Le is
computed as:
AB
D
α
· Le (3-27)
where α is the thermal diffusivity of the air evaluated at the pond-air film temperature
and D
AB
represents the binary diffusion coefficient. The thermal properties (α and c
p
) of
air are computed at each time step using correlations given by Irvine and Liley (1984).
D
AB
is computed after Mills (1995) who references Marrero and Mason (1972):
air
AB
P
T x
D
072 . 2 10
10 87 . 1

· (280K < T < 450K) (3-28)
79
where T refers to the pond-air film temperature in absolute units and P
air
is the
atmospheric pressure in atmospheres.
The heat transfer due to evaporation (q
evaporation
) is then computed by:
w pond fg n evaporatio
m A h q ′ ′ · & (3-29)
where h
fg
is the latent heat of vaporization and is computed at each time step from the
relationship given by Irvine and Liley (1984).
3.4.1.7. Heat Transfer Due to the Heat Exchange Fluid
Heat transfer due to the heat exchange fluid represents the pond thermal load.
This model has been developed to account for water or antifreeze as the heat exchange
fluid. The thermal properties of the fluid are computed at each time step from
correlations given in the Handbook of Chemistry and Physics (CRC, 1980) for water and
from correlations given by Wadivkar (1997) for an antifreeze solution. The thermal
properties are computed at the average fluid temperature (T
fluid
). This temperature is
computed as the average of the inlet and outlet temperatures at the given time step. Since
the outlet temperature at any current time step is not known, the previous converged
value is used as an initial guess and calculation of T
fluid
is iterative. Solution of the pond
temperature is also an iterative procedure as discussed below.
The heat transfer due to the heat exchange fluid (q
fluid
) is computed by:
80
) )( (
circuit pond fluid pipe fluid
N T T UA q − · (3-30)
where UA
pipe
is the overall heat transfer coefficient for the pipe expressed in terms of
inside pipe area and N
circuit
refers to the number of flow circuits (i.e. the number of
spools) installed in the pond. Thus, Equation 3-30 is based on the assumption that one
spool is one flow circuit and that the flow rate is divided evenly between the circuits in a
parallel arrangement. The term UA
pipe
is expressed in terms of the inside pipe area as:
t
spool i
pipe
R
L r
UA
Σ
·
π 2
(3-31)
where r
i
is the inner pipe radius, L
spool
is the length of one spool or circuit, and ΣR
t
represents the composite thermal resistance which is defined by the following resistance
network:
ff R R R R
o pipe i t
+ + + · Σ (3-32)
where R
i
is the thermal resistance due to fluid flow through the pipe, R
pipe
is the pipe
thermal resistance, R
o
is the thermal resistance at external pipe surface, and ff represents a
fouling factor at both the inner and outer pipe walls. The resistance terms are defined as
follows (in terms of inner pipe radius):
i
i
h
R
1
· , (3-33)
81

,
_

¸
¸
·
i
o
pipe
i
pipe
r
r
k
r
R ln (3-34)
and

,
_

¸
¸
·
o o
i
o
h r
r
R
1
(3-35)
where h
i
is the convection coefficient due to fluid flow through the pipe, k
pipe
is the
thermal conductivity of the pipe material, h
o
is the convection coefficient at the outer
surface of the pipe, and r
i
and r
o
are the inner and outer radii of the pipe, respectively.
The above convection coefficients are determined using correlations for the
Nusselt number in flow through a horizontal cylinder, since no specific correlations exist
for a slinky coil. A constant heat flux at the pipe surface is assumed.
In convection heat transfer due to internal flow, Nu is a function of the Reynolds
and Prandtl numbers. Determination of Re is described in Equation 3-17. For this case,
v represents the mass flow rate of the heat exchange fluid, ν represents the kinematic
viscosity of the heat exchange fluid, and the characteristic length (L) is the inner pipe
diameter. A Reynolds number of 2000 is assumed to be critical. For laminar, fully-
developed flow in the pipe (Re<2000), the Nusselt Number is a constant equal to 4.36
(Incropera and DeWitt, 1996, Equation 8-53). For turbulent flow, the Dittus-Boelter
relation is used to compute the Nusselt Number:
x
i
Pr Re 023 . 0 Nu
5
4
· (3-36)
82
where Pr is defined by Equation 3-18 for µ, c
p
, and k representing the thermal properties
of the heat exchange fluid evaluated at its average temperature. The value of the
exponent x in Equation 3-36 is dependent upon whether the entering fluid is being heated
or cooled; x is equal to 0.3 if the entering fluid is greater than the pond temperature and x
is equal to 0.4 if the entering fluid is less than the pond temperature. The inside pipe
convection coefficient can then be determined by using Equation 3-16 where Nu is Nu
i
, k
is the thermal conductivity of the heat transfer fluid, and L is the characteristic length
which is defined in this case as the inner diameter of the pipe (Incropera and DeWitt,
1996).
In free convection at the external pipe surface, the Nusselt Number is a function
of the Rayleigh Number. Ra is computed using Equation 3-13 where α, β, and
ν represent the thermal diffusivity of water, the volumetric thermal expansion coefficient
of water, and the kinematic viscosity of water, respectively, all evaluated at the pipe film
temperature, which is approximated as the average of the pipe surface and pond
temperatures at the given time step. The term ∆T refers to the temperature difference
between the pipe surface and the pond temperatures. Nu for free convection from a
horizontal cylinder is defined as (Churchill and Chu (1975) ):
2
27
8
16
9
6
1
Pr) / 559 . 0 ( 1
Ra 387 . 0
60 . 0 Nu

,
_

¸
¸

,
_

¸
¸
+
+ ·
o
(3-37)
83
where Pr is defined by Equation 3-18 for µ, c
p
, and k representing the thermal properties
of the pond water evaluated at the pipe film temperature. Now the external pipe
convection coefficient can be determined by using Equation 3-16 where Nu is Nu
o
, k is
the thermal conductivity of the pond water evaluated at the pipe film temperature, and L
is the characteristic length which is defined now as the outer pipe diameter.
The outlet fluid temperature (T
out
) is computed from an overall energy balance on
the pipe:
p
circuit
fluid out
c m
q
T T
& 2
− · (3-38)
where m& is the mass flow rate of the heat exchange fluid per flow circuit, c
p
is the specific
heat capacity of the heat exchange fluid, and q
circuit
is the heat rejected/extracted by one
flow circuit. This outlet temperature is used to compute the average fluid temperature at
the next iteration as described above.
3.4.1.8. Solving the Overall Energy Balance Equation
The differential equation describing the overall energy balance on the pond
(Equation 3-2) is rearranged in the form:
2 1
x T x
dt
dT
+ · (3-39)
84
where T represents the pond temperature, x
1
contains all terms of Equation 3-2 that
multiply T, and x
2
contains all terms of Equation 3-2 that are independent of T.
Equation 3-39 is a linear first-order ordinary differential equation which is solved
at each time step using the exponential function as an integrating factor. The solution is
given by the TRNSYS differential equation solver subroutine as:
2
1
2
1
1
x
x
e
x
x
T T
t x
t t t

,
_

¸
¸
+ ·

∆ −
(3-40)
where T
t
is the average pond temperature at the new time step and T
t-∆t
is the average
pond temperature at the previous time step.
Many of the quantities in the heat transfer equations described above require that
the average pond temperature at the current time step be known. Thus, the actual pond
temperature is arrived at iteratively. A convergence criterion for the pond temperature of
1.8x10
-5o
F (1x10
-5o
C) is used.
3.4.2. Computer Implementation
Thc component configuration for the pond model is shown in Figure 3-2. A
companion model was also developed which manipulates any weather data needed for the
pond model. The weather component model makes use of the TRNSYS psychrometric
subroutine to compute moist air properties given two known state properties. The two
85
state properties are dry bulb temperature and one of wet bulb temperature, relative
humidity, or dew point temperature. The weather component model also computes the
sky temperature, the solar radiation on a horizontal surface, and the solar incidence angle.
A computer algorithm is shown in Figure 3-3 in the form of a flow chart.
POND MODEL PARAMETERS:
1. initial pond temperature 2. pond orientation from north
3. pond length 4. pond width
5. pond depth 6. emissivity coefficient
7. extinction coefficient for water 8. number of spools or coils
9. spool length 10. pipe outer diameter
11. pipe thermal conductivity 12. pipe wall thickness
13. fluid type (water or antifreeze) 14. antifreeze concentration if used
15. ground thermal conductivity 16. fouling factor
17. ground water or far field temperature 18. soil hydraulic conductivity
19. hydraulic gradient 20. depth to water table
Figure 3-2. Pond model component configuration.
air
temperature
humidity
ratio
sky
temperature
wind
speed
wind
direction
solar
radiation
solar angle
of incidence
inlet
fluid
temperature
inlet mass
flow rate
pond
temperature
outlet fluid
temperature
mass flow
rate
heat
rejected/absorbed
86
Figure 3-3. Pond model computer algorithm.
Start
Get model inputs
for the TRNSYS time step
Get model
parameters
1
st
call in simulation?
Weather data from
component model
Fluid temperature
and flow rate from
upstream component
Set pond temperature
equal to final value
at previous time step
Compute the average temperature of the
heat exchange fluid
1
st
call in time step?
Set the outputs
T
pond
converged?
Compute heat transfer
by heat exchange fluid
Compute heat transfer quantities
END
Yes
No
Yes
No
Yes
No
Call TRNSYS differential equation solver
to compute the average pond temperature
Compute new outlet fluid temperature
87
3.5. Results and Discussion
3.5.1. Model Comparison to Experimental Results with No Heat Rejection
The first step in the model verification process was to compare the model pond
temperatures to measured pond temperatures during times when no heat was being
rejected to the ponds. This comparison allowed a validity check of the simulation of the
several environmental heat transfer mechanisms occurring within the ponds. Simulated
and actual pond average hourly temperatures are shown in Figure 3-4 for an 8-day period
in July 1998 when no heat was rejected to the ponds. Therefore, in these cases, the model
is driven by weather data input only.
A review of the plots in Figure 3-4 shows that the model temperatures compare
favorably to the measured pond temperatures. The simulated temperatures are within 3
o
F
(1.67
o
C) of the observed temperatures throughout the test period. The difference
between the average simulated pond temperature and the average observed pond
temperature for the entire test period is 1.93
o
F (1.07
o
C) for the 2-feet deep pond and
1.55
o
F (0.86
o
C) for the 3.5-feet deep pond. Shallow ground water was not encountered at
the site and therefore ground water contributions were not considered. Fouling of the
heat exchanger pipe was also not considered.
88
Figure 3-4. Comparison of observed and simulated average pond temperatures
with no heat rejection in the (a) 2-feet (0.61 m ) deep pond and (b)
3.5-feet (1.07 m) deep pond.
(a)
70
75
80
85
90
95
100
105
110
21-Jul 22-Jul 23-Jul 24-Jul 25-Jul 26-Jul 27-Jul 28-Jul 29-Jul 30-Jul
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
21.1
23.9
26.7
29.4
32.2
35.0
37.8
40.6
43.3
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Air Actual Pond Model Pond
(b)
70
75
80
85
90
95
100
105
110
21-Jul 22-Jul 23-Jul 24-Jul 25-Jul 26-Jul 27-Jul 28-Jul 29-Jul 30-Jul
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
21.1
23.9
26.7
29.4
32.2
35.0
37.8
40.6
43.3
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Air Actual Pond Model Pond
89
3.5.2. Model Comparison to Experimental Results with Heat Rejection
Heat rejection to the ponds was simulated over a 25-day period from November
12 to December 7, 1998. Input data to the model consisted of weather data as described
previously in addition to measured slinky heat exchanger supply water temperatures and
flow rates on 6-minutely time intervals. The model performance was evaluated by
comparing (1) the simulated to the observed return temperature of the heat exchange fluid
and (2) the simulated cumulative heat rejected to the ponds to the measured water
heating element and pump power input. These comparisons are shown in Figures 3-5 and
3-6 respectively. As with the previous comparisons, ground water contributions and
fouling of the heat exchanger pipe were not considered.
A review of the temperature plots in Figure 3-5 shows that model fluid return
temperatures compare favorably to the observed fluid return temperatures. The average
observed and modeled fluid return temperatures over the test period in the 2-feet (0.61-
meter) deep pond were 70.5
o
F (21.4
o
C) and 70.2
o
F (21.2
o
C), respectively, and in the 3.5-
feet (1.07-meter) deep pond were 69.2
o
F (20.7
o
C) and 70.4
o
F (21.3
o
C), respectively. The
deeper pond has slightly larger differences between modeled and observed fluid return
temperatures. The error is small, however, and is probably acceptable for purposes of
simulating hybrid GSHP systems; even a 2
o
F (1.11
o
C) error in return fluid temperature
from the pond will cause only a slight difference in modeled heat pump performance.
90
Figure 3-5. Comparison of observed and simulated heat exchange fluid return
temperatures for the (a) 2-feet (0.61-m ) deep pond and (b) 3.5-feet
(1.07-m) deep pond.
(a)
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
16.7
17.2
17.8
18.3
18.9
19.4
20.0
20.6
21.1
21.7
22.2
22.8
23.3
23.9
24.4
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Actual Return Model Return
(b)
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
16.7
17.2
17.8
18.3
18.9
19.4
20.0
20.6
21.1
21.7
22.2
22.8
23.3
23.9
24.4
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Actual Return Model Return
91
Figure 3-6. Comparison of observed and simulated heat rejected to the (a) 2-feet
(0.61-m ) deep pond and (b) 3.5-feet (1.07-m) deep pond.
A review of the plots in Figure 3-6 shows that the model cumulative heat rejected
compares well to the measured heating element and pump power input. At the end of the
25-day test period, the percent difference between the cumulative simulated heat rejected
and the cumulative measured heat rejected is –2.95 % for the 2-feet deep (0.61-meter)
(a)
0
200
400
600
800
1000
1200
1400
1600
1800
12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec
C
u
m
m
.

H
e
a
t

R
e
j
e
c
t
e
d

(
k
W
-
h
r
)
Heating Element + Pump Power Model Heat Rejected
(b)
0
200
400
600
800
1000
1200
1400
1600
1800
12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec
C
u
m
m
.

H
e
a
t

R
e
j
e
c
t
e
d

(
k
W
-
h
r
)
Heating Element + Pump Power Model Heat Rejected
92
pond and –5.20 % for the 3.5-feet (1.07-meter) deep pond. These discrepancies may be
due partly to heat losses from the pond supply/return pipes to the ground and to the
atmosphere in the equipment building. A model and experimental uncertainty analysis is
presented in Appendix A.
3.5.3. Model Application
To illustrate the applicability of the model as well as the viability of using shallow
ponds as supplemental heat rejecters in GSHP systems, a model of a hypothetical GSHP
system was constructed in the TRNSYS modeling environment. A simplified system
schematic is shown in Figure 3-7. Each of the component models is described briefly
below.
Figure 3-7. System schematic for the example model of a GSHP system with a
shallow pond supplemental heat rejecter.
Building
Loads
Ground Loop
Heat Exchanger
Pond Loop
Heat Exchanger
Heat
Pumps
Controller
T
T
93
The building is not modeled explicitly in this application. The hourly building
thermal loads are pre-computed using a proprietary building energy analysis program and
are read from a file and passed to the heat pump subroutines. The building is an actual
four-story, 45,000-ft
2
(4181-m
2
) office building located in Tulsa, Oklahoma and is highly
cooling dominated. The building thermal loads are shown in Figure 3-8.
A simple water-to-air heat pump model was developed for this and other GSHP
system simulations. Inputs to the model include sensible and latent building loads,
entering fluid temperature, and fluid mass flow rate. The model uses quadratic curve-fit
equations to manufacturer’s catalog data to compute the heat of rejection in cooling
mode, heat of absorption in heating mode, and the heat pump power consumption.
Outputs provided by the model include exiting fluid temperature, power consumption,
and fluid mass flow rate. In this application, one heat pump component model handles
the heating load and a second heat pump component model handles the cooling load.
The ground-loop heat exchanger model used in this application is that described
by Yavuzturk and Spitler (1999) which is based partly on the work of Eskilson (1987)
who developed “long time-step” (monthly) response factors for vertical ground-coupled
U-tube heat exchangers. The model of Yavuzturk and Spitler (1999) extends the work of
Eskilson (1987) to hourly or less (“short-time step”) time intervals. The development of
the short time-step response factors are based on an analytically validated, transient two-
dimensional implicit finite volume model (Yavuzturk et al., 1999) that simulates the heat
transfer over a vertical U-tube ground heat exchanger. In this application, the modeled
borehole field consisted of one hundred 250-feet (76.2-m) deep boreholes arranged in a
94
10 by 10 square pattern. The total system flow was rate 270 gpm (61.36 m
3
/hr).
Representative thermal properties of sedimentary rock were chosen.
Figure 3-8. Building thermal loads for the example building in Tulsa, OK.
Cooling loads are shown as positive values, indicating heat to be
rejected to the GSHP system; heating loads are shown as negative
values, indicating heat to be extracted from the GSHP system.
Models for ancillary components such as pumps, t-pieces, flow diverters, and the
differential controller are described by SEL (1997). The control strategy used to activate
the circulating pump to the pond was chosen somewhat arbitrarily by using the
temperature difference between the pond and the exiting fluid temperature from the heat
pumps. When this temperature difference exceeds 9
o
F (5
o
C), the circulating pump to the
pond is energized and heat will be rejected to the pond. During these times of heat
rejection to the pond, flow is diverted to the pond such that each heat exchanger coil in
the pond receives 4 gpm (0.909 m
3
/hr) of water. The properties of each heat exchanger
-200
0
200
400
600
800
1000
1200
1400
1
-
J
a
n
1
-
F
e
b
1
-
M
a
r
1
-
A
p
r
1
-
M
a
y
1
-
J
u
n
1
-
J
u
l
1
-
A
u
g
1
-
S
e
p
1
-
O
c
t
1
-
N
o
v
1
-
D
e
c
Time
B
u
i
l
d
i
n
g

L
o
a
d

(
k
B
t
u
/
h
r
)
-59
0
59
117
176
234
293
352
410
B
u
i
l
d
i
n
g

L
o
a
d

(
k
W
)
Cooling Load
Heating Load
95
coil in the example model are the same as those described in the experimental procedure.
Hourly input weather data for the pond model were taken from a typical meteorological
year (TMY) record for Tulsa, Oklahoma.
The model was run for two cases for a duration of 3 years with a time step of one
hour. The first case was the GSHP system with no pond and the second case was the
GSHP system with the pond. Hourly heat pump entering water temperatures are shown
in Figure 3-9 for both cases.
Figure 3-9. Entering heat pump water temperatures for the example GSHP
system simulation with no pond and with a 2-feet (0.6096-m) deep,
6000-ft
2
(557.4 m
2
) pond.
A review of the data presented in Figure 3-9 shows the advantages of using a
pond supplemental heat rejecter. Assuming that a maximum heat pump entering water
temperature of 100
o
F (37.78
o
C) is desirable, the system without the pond would fail
during the second year of operation. In fact, based on the results of a ground-loop heat
30
40
50
60
70
80
90
100
110
120
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (years)
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
10.0
15.6
21.1
26.7
32.2
37.8
43.3
48.9
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
no
pond
with
pond
96
exchanger sizing program (Spitler et al., 1996), the boreholes of a 10 by 10 square pattern
would need to be approximately 400 feet (121.9 m) deep to accommodate the cooling-
dominated loads of this building for 20 years of operation. Such a system would likely
be eliminated from consideration early on in the design phase because of excessive first
cost.
Using the TRNSYS model as a design tool, the size of the pond supplemental heat
rejecter was determined under the assumption that the 10 by 10 borehole field could not
be feasibly deeper than 250 feet (76.2 m). The heat pump entering water temperatures
for the GSHP system with the pond shown in Figure 3-9 were produced by simulating a
2-feet (0.61-m) deep, 6000 ft
2
557 (m
2
) pond with 50 slinky heat exchanger coils. A
summary of pond performance is given in Table 3-1. By adding the pond supplemental
heat rejecter in this example, the depth of the borehole field could be decreased by
approximately 35%. A more detailed system simulation could involve system life-cycle
operating cost analyses, control strategy variations, and design variable optimization.
Table 3-1. Summary of Pond Performance for Example GSHP
System Simulation
Year Hours Average Pond Heat Pump Maximum Heat Rejected
ON Temperature Entering Fluid
Temperature
(
o
F) (
o
C) (
o
F) (
o
C) (kBtu) (MJ)
1 3937 74.79 23.77 99.95 37.75 1,618,224 1,706,903
2 4873 76.37 24.65 100.29 37.94 2,160,080 2,278,452
3 5324 77.52 25.29 100.18 37.88 2,498,961 2,635,904
97
3.6. Concluding Remarks and Recommendations for Future Work
A design and simulation tool for modeling the performance of a shallow pond as a
supplemental heat rejecter in ground-source heat pump systems has been developed. The
model has been developed in the TRNSYS modeling environment (SEL, 1997) and can
be coupled to other GSHP system component models for short-time step (hourly or
minutely) system analyses. The model has been validated by comparing simulation
results to experimental data.
The model accounts for several natural heat transfer mechanisms within a surface
water body plus convective heat transfer due to a closed-loop heat exchanger coil. The
heat transfer fluid is assumed to be carried by a series of pipes in the form of bundle
spools or “slinky” coils. Environmental heat transfer mechanisms that are simulated by
the model include solar radiation heat gain, heat and mass transfer due to evaporation,
convection heat transfer to the atmosphere, thermal or long-wave radiation heat transfer,
conduction heat transfer to the surrounding soil or fill material, and ground water
discharge contributions. The solution scheme involves a lumped-capacitance approach
and the resulting first-order differential equation describing the overall energy balance on
the pond is solved numerically. Some outputs provided by the model include average
pond temperature, exiting fluid temperature, and heat rejected to the pond.
An example application has been presented to demonstrate the use of the model as
well as the viability of the use of shallow ponds as supplemental heat rejecters in GSHP
98
systems. Through this example, it is shown that the size of ground-loop heat exchangers
can be significantly decreased by incorporating a shallow pond into the GSHP system.
The potential exists for significantly increasing the performance of shallow ponds
used as supplemental heat rejecters in GSHP systems. Further research is suggested in
the following areas:
• Optimization of the design procedure and control strategy. Hybrid ground source
heat pump systems have many degrees of freedom; there are tradeoffs between the
reduction in size of the ground loop heat exchanger, the size of the pond, and the
control strategy. To more fully understand this, additional research using the
simulation techniques developed in this paper is needed. This research would also
take into account the economic costs and benefits, which we have not investigated.
• Additional validation of the model, using data from a working system, would be
useful.
• Extension of the model to cover deep ponds for situations where an existing pond or
lake is available.
• The use of spray fountains and other aeration devices in the pond to enhance pond
cooling.
• The impact of pipe configuration within the pond on the overall system performance.
99
4. A Model for Simulating the Performance of a Pavement Heating
System as a Supplemental Heat Rejecter with Closed-Loop Ground-
Source Heat Pump Systems
4.1. Introduction
The reasons for using supplemental heat rejecters in vertical borehole GSHP
systems have been described in Section 3.1. Several combinations of so-called “hybrid
GSHP systems” are possible. Chapter 3 has dealt with using a shallow pond as
supplemental heat rejecter. This chapter deals with using a hydronic pavement heating
system as a supplemental heat rejecter. With additional heating equipment where
applicable, these types of systems can also provide a useful and cost-effective method for
pavement de-icing.
The objective of the work presented in this chapter has been to develop a design
and simulation tool for modeling the performance of a hydronic pavement heating system
that can be usefully and cost-effectively integrated into a ground-source heat pump
system as a supplemental heat rejecter. The pavement heating model has been developed
in the TRNSYS modeling environment (SEL, 1997) and can therefore be coupled to other
GSHP system component models for short-time step (hourly or minutely) system
analyses. The model has been validated by comparing simulation results to experimental
data. As an example of the model’s applicability, GSHP system simulation results are
presented for a commercial building located in Tulsa, Oklahoma with a hypothetical
closed-loop GSHP system with and without a pavement supplemental heat rejecter.
100
4.2. Heat Transfer in Pavement Slabs
Hydronically-heated pavement systems are commonly one of two types of
configurations: (1) “serpentine” configuration (Figure 4-1) or (2) “slinky” configuration
(Figure 4-2).
The serpentine configuration is that commonly used in snow-melting systems.
The pipes are embedded in the pavement material and are placed on even centers and
connected with U-shaped tubing. In the slinky configuration, a pipe is coiled in a circular
fashion such that each loop overlaps the adjacent loop. The slinky is typically installed in
fill material near the base of the pavement slab.
Pertinent concepts of heat transfer in pavement slabs have been addressed for
snow melting applications by many sources including Adlam (1950), Chapman (1952),
Kilkis (1994), ASHRAE (1995), and Ramsey et al. (1999). Heat transfer mechanisms
acting upon the pavement slab are shown schematically in Figure 4-3. Heat transfer
within the slab itself is by conduction. Internal sources of heat are due to convection
from flow of the heat transfer fluid through the pipes. Heat fluxes at the pavement
surface are due to a number of environmental interactions and include convection, solar
radiation, thermal (long-wave) radiation, sensible heat transfer from precipitation, and
latent heat transfer from melting snow and evaporating water. On the bottom and sides of
the slab, heat fluxes are due to conduction to the ground and may or may not be
101
significant as compared to the top surface heat fluxes. If the slab bottom is exposed, as in
the case of a bridge or parking deck, heat transfer from the bottom surface is by
convection and radiation to the surroundings.
4.3. Experimental Methods
4.3.1. Test Slab Description and Data Collection
Two hydronically-heated concrete slabs have been constructed on a test site at the
Oklahoma State University and used for this study. Each is discussed in the following
subsections.
4.3.1.1. Bridge Deck Test Section
The first test slab was constructed resembling a concrete bridge deck.
Construction details are given by Liao and Hogue (1996). The test slab is rectangular
with a plan area of 10 ft (3.05 m) by 3 ft (0.91 m) and a thickness of 8 in. (0.2032 m).
The slab has been constructed on a steel frame and insulated on all four sides to minimize
edge losses. Heat was rejected to the slab by circulating a heated fluid through a pipe
system installed in a serpentine configuration. The pipes are made of polybutylene with a
nominal diameter of ¾-inch (0.01905-m) and were embedded at a depth of 2.5 in.
(0.0635 m) from the slab surface on 6.5-inch (0.1651-m) centers.
102
Figure 4-1. Serpentine pipe configuration in a hydronically-heated pavement
slab in (a) plan view and (b) cross-sectional view.
Figure 4-2. Slinky pipe configuration in a hydronically-heated pavement slab in
(a) plan view and (b) cross-sectional view along the slinky center-
line.
2 - 3 in.
typ.
Pavement Top
Pavement Bottom
6 in. typ.
(b)
Pipes
Pavement Material
Fill Material
Pavement Surface
Pipes
(b) 8-12 in. typ.
Supply
Supply
Return
Return
(a)
“slinky” pipes
X-sect
cut
Return
Supply
(a)
Pipes
X-sect
cut
103
Figure 4-3. Heat transfer mechanisms in hydronically-heated pavement slabs
with (a) no bottom exposure to the atmosphere and (b) bottom
exposure to the atmosphere.
Data used in this portion of the study were collected by Wadivkar (1997). An
antifreeze heat exchange fluid was circulated through the bridge deck coupled to a
closed-loop ground-source heat pump system over several night-time periods during the
Incident
Solar Radiation
Thermal Radiation
Wind
Convection
Sensible heat + heat of fusion
to melt snow
Heat of evaporation
of rain and melted snow
Fill
Pavement Convection due to internal pipe flow
Conduction
Conduction
(a)
Incident
Solar Radiation
Thermal Radiation
Wind
Convection
Sensible heat + heat of fusion
to melt snow
Heat of evaporation
of rain and melted snow
Bridge/Parking Deck Pavement Convection due to internal pipe flow
Conduction
Conduction
Wind
Convection
(b)
Thermal Radiation
104
winter of 1996. The bridge deck fluid supply and return temperatures, the fluid flow rate,
and the bridge surface temperature were recorded at 5-minutely intervals. The bridge
surface temperature was measured using 28 thermocouples embedded in the concrete
surface and the bridge fluid supply and return temperatures were measured using
thermocouples embedded in the pipes. Icing conditions on the concrete surface were also
noted during the Wadivkar (1997) study.
4.3.1.2. Parking Lot Test Section
The construction of this second test slab and data collection for this study have
been conducted by researchers affiliated with the Division of Engineering Technology at
the Oklahoma State University.
This test slab was constructed resembling a parking lot or concrete sidewalk. This
test slab is rectangular with a plan area of 40 ft (12.19 m) by 4 ft (1.22 m) and a thickness
of 6 in. (0.1524 m). The concrete slab was underlain by 6 in. (0.1524 m) of sand fill
material. Heat was rejected to the slab by circulating heated water through a slinky heat
exchanger installed at the concrete/sand fill interface. The slinky pipe is made of HDPE
plastic and is 500 feet (152.40 m) long with a nominal diameter of ¾ in. (0.01905 m).
The pipe was coiled such that the resulting slinky heat exchanger is 40 ft (12.19 m) long
with a diameter of 3 ft (0.91 m) and a 10-in. (0.254 m) pitch (the separation distance
between the apex of each successive loop).
The temperature of the concrete surface was measured by two thermistors
embedded in the concrete at a depth of approximately ¼ in. (0.0064 m) from the surface.
105
One thermistor was placed near the slinky center between two pipes of the slinky and the
other was placed near the end of the slinky above a pipe. Slinky supply and return water
temperatures were measured by thermistors embedded in the slinky header. The
remainder of the heat rejection system included a flow meter, a water heating element,
and a watt transducer. All sensor information was recorded at time intervals of 6
minutes.
The tests are controlled to maintain a set water temperature by heating the water if
the temperature falls below a set point. Two set point temperatures were used in this
study: 90
o
F (32.2
o
C) in the summer season and 75
o
F (23.9
o
C) in the winter season.
4.3.2. Weather Instrumentation and Data Collection
Weather data for this study were obtained from the Oklahoma Mesonet
(mesoscale network), which is a weather station network consisting of weather
monitoring sites scattered throughout Oklahoma. Depending on the weather parameter,
data are recorded at time intervals ranging from 3 seconds to 30 seconds and averaged
over 5-minute observation intervals.
Weather data at 15-minute intervals for the Stillwater monitoring station were
acquired for the time periods of interest for this study. The Stillwater station is located
approximately 1 mile from the test site. Data for the following parameters were obtained:
wind speed, wind direction, air temperature, relative humidity, and solar radiation.
Further details of the weather station network may be found in Elliott et al. (1994).
106
4.4. Model Development
4.4.1. Governing Equations
Transient heat transfer in the pavement slab is represented in the model in two-
dimensional (2-D) cross-section using the cartesian coordinate system. The 2-D
approach is reasonable if the cross-section is taken through the mid-section of the slab
along the major direction of fluid flow in the pipe system. The transient 2-D heat
conduction equation can be expressed as:
t
T
z
T
x
T


·


+


α
1
2
2
2
2
(4-1)
This equation has been discretized using an explicit finite difference method. The typical
geometry and notation of the finite difference cells in the x-z cartesian coordinate plane
are shown in Figure 4-4.
The nodal equations are formulated using a node-centered, energy balance
approach. The resulting general form of the explicit finite difference equation is:

,
_

¸
¸


·
∆ −
·
∆ −

t
T T
c V A q
t t
n m
t
n m
p
i
t t
i
) (
) , ( ) , (
4
1
) (
" ρ (4-2)
where,
) ( t t
i
q
∆ −
′ ′ is the heat flux across the cell face i at the previous time step, A is the cell
face area (assuming a unit depth), V is the cell volume (assuming a unit depth), ρ is the
107
average density of the cell material, c
p
is the average specific heat capacity of the cell
material, T
(m,n)
t
is the nodal temperature at the current time step, T
(m,n)
t-∆t
is the nodal
temperature at the previous time step, and ∆t is the time step. The heat flux, q ′ ′ , for
conduction into node (m,n) during a given time step is given by Fourier’s Law in discrete
form as:
l
) (
"
) , (
) , (
n m i
n m i
T T
k q

·

(4-3)
where subscript i denotes a neighboring node (from 1 to 4), k is the average thermal
conductivity of the material between nodes i and (m,n), and l is the distance between
nodes i and (m,n).
Figure 4-4. Finite difference cell geometry and notation.
x
z
(m, n)
(m, n + 1) (m - 1, n + 1)
(m - 1, n) (m + 1, n)
(m + 1, n + 1)
(m - 1, n - 1) (m + 1, n - 1)
q
4
(m, n - 1)
q
1
q
2
q
3
∆x
∆z
108
The size of the time step is limited by the need to maintain stability. In this fully
explicit method, the stability criterion for two-dimensional problems is given by:
4
1
2


l
t α
(4-4)
where α is the thermal diffusivity of the material within the cell and l is the nodal
spacing.
4.4.2. The Finite Difference Grid
The finite difference grid used in the model is shown in Figure 4-5. A uniform
square nodal spacing equal to the pipe radius has been used. Because of symmetry and
small temperature differences between adjacent pipe, and neglecting edge effects, the
model domain was reduced to a width equivalent to one-half of the pipe spacing as
shown in Figure 4-5. In the z direction, the domain corresponds to the top of the slab and
bottom of the slab or the base of the underlying fill material. In the x direction, the
domain corresponds to a distance from the center-line of a pipe to half the distance to the
adjacent pipe.
4.4.3. Boundary Conditions
The boundaries of the model domain are treated as flux-type (Neumann)
conditions as shown in Figure 4-5. The temperature at each boundary node is given by
109
the energy balance equation (Equation 4-2), where
i
q ′ ′ represents the appropriate external
flux and conduction flux from adjacent nodes.
Figure 4-5. The model domain showing the finite-difference grid and boundary
conditions. Shaded squares show example control volumes for
different types of grid node geometries.
Lines of symmetry on the left- and right-hand boundaries are, by definition, zero-
flux conditions. Heat flux at the pipe surface nodes represents convection from the heat
transfer fluid. Heat flux at each top surface node (
) 1 , (m
q ′ ′ ) is given by:
latent snow rain sensible snow rain convection thermal solar m
q q q q q q
− −
+ + + + ·
, , ) 1 , (
" " " " " " (4-5)
1/2 pipe spacing
∆z
∆x
Adiabatic or Convective Boundary- Pavement Bottom Surface
A
d
i
a
b
a
t
i
c

B
o
u
n
d
a
r
y

-

l
i
n
e

o
f

s
y
m
m
e
t
r
y
A
d
i
a
b
a
t
i
c

B
o
u
n
d
a
r
y
A
d
i
a
b
a
t
i
c

B
o
u
n
d
a
r
y
Heat Flux Boundary - Pavement Top Surface
(solar heat gain, convection, thermal radiation, sensible heat from
precipitation, heat of fusion from snow melt, heat of evaporation of precipitation)
Heat Flux Boundary -
Pipe Wall
(fluid convection from
internal pipe flow)
Pipe
radius
x
z
110
where
solar
q ′ ′ is the solar radiation heat flux,
thermal
q ′ ′ is the thermal radiation heat flux,
convection
q ′ ′ is the convection heat flux,
sensible snow rain
q

′ ′
,
is the sensible heat flux from falling
rain and snow, and
latent snow rain
q

′ ′
,
is the latent heat flux from melting snow and
evaporating/condensing water. The bottom surface is treated either as an insulated
surface or as a surface exposed to convective plus radiative conditions. Each of the heat
flux terms is further described below.
4.4.3.1. Solar Radiation Heat Flux
Solar radiation heat flux (
solar
q ′ ′ ) is the net solar radiation absorbed by the
pavement surface and is given by:
I q
solar
α · " (4-6)
where I is the solar radiation incident on the pavement surface and α is the absorptivity
coefficient for the pavement material. The absorptivity coefficient is corrected for solar
incidence angle (θ) dependence using an empirical correlation given by Duffie and
Beckman (1991). The model also accepts solar radiation in the form of beam (I
b
) and
diffuse (I
d
) components, in which case I is computed from:
d b
I I I + · θ cos (4-7)
The angle of incidence (θ) of the sun’s rays is computed at each time step from
correlations given by Spencer (1971), Duffie and Beckman (1991), and ASHRAE (1997).
111
4.4.3.2. Thermal Radiation Heat Flux
This heat transfer mechanism accounts for heat flux at the pavement top surface
and bottom surface (if exposed) due to thermal or long-wave radiation. This model uses
a linearized radiation coefficient (h
r
) defined as:
3
2 ) , (
2
4

,
_

¸
¸ +
·
T T
h
n m
r
εσ (4-8)
where ε is the emissivity coefficient of the pavement material, σ is the Stefan-Boltzmann
constant, T
(m,n)
is the surface node temperature in absolute units, and T
2
represents the sky
temperature in absolute units. T
sky
is computed from the relationship given by Bliss
(1961). If the bottom of the slab is exposed, Equation 4-8 is also used to determine h
r
for
the bottom surface, where T
2
represents the ground temperature in absolute units, which
is approximated as the air temperature. The thermal radiation heat flux at each node
(
thermal
q ′ ′ ) is then computed by:
) ( "
) , ( 2 n m r thermal
T T h q − · (4-9)
4.4.3.3. Convection Heat Flux at the Pavement Surfaces
This mechanism accounts for heat transfer at the pavement top and bottom
surfaces due to free and forced convection. The convection coefficient (h
c
) is a function
of the Nusselt Number (Nu). Several empirical formulations exist for determining the
convection coefficient for different geometries. For a pavement surface, correlations for
112
a horizontal flat plate are the most applicable and therefore the convection coefficient (h
c
)
is determined as described in Section 3.4.1.3.
In free convection heat transfer, Nu is a function of the Rayleigh Number (Ra).
Ra is determined as described by Equation 3-13 where ∆T refers to the temperature
difference between the pavement surface at node (m,n) and the air, L is the characteristic
length described for horizontal flat plates as the ratio of the area to the perimeter
(Incropera and DeWitt, 1996), and the thermal properties α, β, and ν are evaluated at the
film temperature, which can be approximated as the average of the air and the pavement
surface temperature at node (m,n). In external free convection flows over a horizontal
flat plate, the critical Rayleigh Number is about 10
7
. Therefore, two empirical relations
for the Nusselt number are used in this model as described by Incropera and DeWitt
(1996) for free convection from the upper surface of a heated plate or the lower surface of
a cooled plate (Equations 3-15a and 3-15b). The convection coefficient (h
c
) for free
convection can then be determined from Equation 3-16 where k is the thermal
conductivity of air evaluated at the film temperature and L is the characteristic length
described above.
In forced convection heat transfer, Nu is usually correlated to the Reynolds (Re)
and Prandtl (Pr) Numbers. For external forced convection over a flat plate (i.e. the
pavement surface), the critical Reynolds Number is approximately 10
5
. Therefore, two
empirical relations for the Nusselt number are used in the model as described by
Incropera and DeWitt (1996) for forced convection over a flat plate (Equations 3-19a and
113
3-19b). The convection coefficient (h
c
) for forced convection can then be determined by
Equation 3-16 with the characteristic length value described as the ratio of the length
(parallel to the wind direction) to the perimeter.
Finally, the convection heat flux at each pavement surface node (
convection
q ′ ′ ) is
computed by:
) ( "
) , ( n m air c convection
T T h q − · (4-10)
where T
air
is the dry-bulb air temperature and h
c
is taken as the maximum of the free
convection coefficient and the forced convection coefficient. This practice of choosing
the larger of the free and forced convection coefficients is recommended by Duffie and
Beckman (1991) and McAdams (1954) and is used in the absence of additional
experimental evidence regarding combined free and forced convection.
4.4.3.4. Heat Flux Due to Rain and Snow
This heat transfer mechanism includes both sensible and latent effects. This
model uses a simple mass/energy balance on water at the pavement surface to account for
heat and mass transfer. The thermal properties of water are computed from correlations
given in the Handbook of Chemistry and Physics (CRC, 1980).
The sensible heat flux due to falling rain or snow at each pavement surface node
(
snow rain
q
,
′ ′ ) is given by:
114
) ( " "
) 1 , ( , , m air p snow rain snow rain
T T c m q − · & (4-11)
where
snow rain
m
,
" & is the rainfall or snowfall rate in water equivalent mass per unit time per
unit area and c
p
is the specific heat capacity of water at the air temperature.
Latent heat transfer is considered only if the air temperature or the slab surface
temperature is above 33
o
F (0.55
o
C). Accumulation of rain is not considered; rainfall is
assumed to drain instantaneously from the pavement surface, forming a thin film from
which evaporation occurs.
This model uses the j-factor analogy to compute the mass flux of evaporating
water at each pavement surface node (
w
m" & ):
) ( "
) 1 , (m air d w
w w h m − · & (4-12)
where h
d
is the mass transfer coefficient, w
air
is the humidity ratio of the ambient air, and
w
(m,1)
represents the humidity ratio of saturated air at the surface node. The mass transfer
coefficient (h
d
) is defined using the Chilton-Colburn analogy defined previously by
Equation 3-26 where h
c
is the convection coefficient defined above, c
p
is the specific heat
capacity of the air evaluated at the pavement node - air film temperature, and Le is the
Lewis number described by Equation 3-27 where α and D
AB
are each evaluated at the
pavement node - air film temperature. The heat flux due to evaporation (
n evaporatio
q ′ ′ ) is
then given by:
115
w fg n evaporatio
m h q " " & · (4-13)
where h
fg
is the latent heat of vaporization and is computed from the relationship given by
Irvine and Liley (1984).
The heat flux due to melting snow and ice is determined using a mass balance on
freezing precipitation that has accumulated at the pavement surface. The accumulation of
ice at the beginning of each time step is determined from the sum of the rainfall rate and
the snowfall rate if the air temperature or the slab surface temperature is below 33
o
F
(0.55
o
C). Although snow is a porous medium, it is treated in the model as an equivalent
ice layer. Sublimation of ice is not considered. The mass flux of water due to melting ice
(
icemelt
m" & ) at the pavement surface is then given by:

if
ice conduction n evaporatio sensible snow rain convection thermal solar
icemelt
h
q q q q q q
m
, ,
" " " " " "
"
+ + + + +
·

& (4-14)
where
ice conduction
q
,
′ ′ is the conduction heat flux from the pavement surface into the ice layer
and h
if
is the latent heat of fusion of water. The other heat flux terms have been defined
previously. If the ice thickness is greater than zero, the heat flux into each pavement
surface node (
) 1 , (m
q ′ ′ ) is given by:
if icemelt m
h m q " "
) 1 , (
& − · (4-15)
The thickness of the ice layer at the end of the time step (
new ice,
l ) is computed by:
116
t
m m
ice
icemelt w
old ice new ice

,
_

¸
¸ +
− ·
ρ
" "
, ,
& &
l l (4-16)
4.4.3.5. Heat Transfer Due to the Heat Exchange Fluid
Heat transfer due to the heat exchange fluid is represented by heat flux at the pipe
surface nodes. This model has been developed to account for water or antifreeze as the
heat exchange fluid. The thermal properties of the fluid are computed at each time step
from correlations given in the Handbook of Chemistry and Physics (CRC, 1980) for
water and from correlations given by Wadivkar (1997) for an antifreeze solution. The
thermal properties are computed at the average fluid temperature (T
fluid
). This
temperature is computed as the average of the inlet and outlet temperatures at the given
time step. Since the outlet temperature at any current time step is not known, the
previous converged value is used as an initial guess and calculation of T
fluid
is iterative.
The heat flux due to the heat exchange fluid (
fluid
q ′ ′ ) is computed per flow circuit
by:
) ( "
) , ( n m fluid pipe fluid
T T U q − · (4-17)
where U
pipe
is the overall heat transfer coefficient for the pipe and is expressed as:
pipe pipe
pipe
k h
U
l
+
·
1
1
(4-18)
117
where h
pipe
is the convection coefficient due to fluid flow through the pipe, k
pipe
is the
thermal conductivity of the pipe material, and
pipe
l is the wall thickness of the pipe.
The convection coefficient due to fluid flow in the pipe is determined using
correlations for the Nusselt Number in flow through a horizontal cylinder. A constant
heat flux at the pipe surface is assumed. For laminar flow in the pipe (Re<2000), the
Nusselt Number is a constant equal to 4.36 (Incropera and DeWitt, 1996, Equation 8-53).
For turbulent flow, the Dittus-Boelter relation is used to compute the Nusselt Number as
described by Equation 3-36. The value of the exponent x in Equation 3-36 is dependent
upon whether the entering fluid is being heated or cooled; x is equal to 0.3 if the entering
fluid is greater than the slab temperature and x is equal to 0.4 if the entering fluid is less
than the slab temperature. The convection coefficient (h
fluid
) is given by Equation 3-16
where k is the thermal conductivity of the heat transfer fluid and the characteristic length
(L) is defined as the inner diameter of the pipe.
The outlet fluid temperature (T
out
) is computed from an overall energy balance on
the pipe flow circuit:
p
pipe fluid
fluid out
c m
A q
T T
& 2
"
− · (4-19)
where A is the inside surface area of the pipe per flow circuit, m& is the mass flow rate of
the heat exchange fluid per flow circuit, and c
p
is the specific heat capacity of the heat
exchange fluid. This outlet temperature is used to compute the average fluid temperature
at the next iteration as described above.
118
The total heat transfer due to the fluid flow (q
transfer,fluid
) is given by:
circuit out in p fluid transfer
N T T c m q ) (
,
− · & (4-20)
where N
circuit
is the number of flow circuits.
4.4.4. Computer Implementation
The component configuration for the pavement heating model is shown in Figure
4-6. A companion model was also developed which manipulates any weather data
needed for the pavement heating model. The weather component model makes use of the
TRNSYS psychrometric subroutine to compute moist air properties given two known
state properties. The two state properties are dry bulb temperature and one of wet bulb
temperature, relative humidity, or dew point temperature. The weather component model
also computes the sky temperature, the solar radiation on a horizontal surface, and the
solar incidence angle. A computer algorithm is shown in Figure 4-7 in the form of a flow
chart.
119
PAVEMENT HEATING MODEL PARAMETERS:
1. slab length 2. slab width
3. slab orientation from north 4. thickness of slab + fill
5. pipe spacing 6. pipe diameter
7. pipe depth below surface 8. depth to material 1-2 interface
9. thermal conductivity, layer 1 10. thermal conductivity, layer 2
11. emissivity coefficient 12. absorptivity coefficient
13. volumetric heat capacity, layer 1 14. volumetric heat capacity, layer 2
15. thermal conductivity, pipe material 16. pipe wall thickness
17. flag for fluid type (water or antifreeze) 18. antifreeze concentration if used
19. number of flow circuits 20. pipe length per flow circuit
21. time step for finite difference method 22. flag for bottom boundary cond.
Figure 4-6. Pavement heating model component configuration.
air
temperature
humidity
ratio
sky
temperature
wind
speed
wind
direction
solar
radiation
inlet
fluid
temperature
inlet mass
flow rate
slab surface
temperature
outlet fluid
temperature
mass flow
rate
heat
rejected/absorbed
solar angle
of incidence
snowfall
rate
rainfall
rate
120
Figure 4-7. Pavement heating model computer algorithm.
Start
Get model inputs
for the TRNSYS time step
Get model
parameters
1
st
call in simulation?
Weather data from
component model
Fluid temperature
and flow rate from
upstream component
Set
nodal temperatures
to initial values
Compute grid
parameters
Compute fluid thermal properties
Compute average fluid temperature
1
st
call in simulation?
Set the outputs
Compute outlet fluid temperature
Avg. fluid temperature
converged?
Compute average pavement surface temperature
Compute heat transferred by fluid
Perform heat balance on nodal network
END
Yes
No
Yes
No
Yes
No
121
4.5. Results and Discussion
4.5.1. Model Comparison to Experimental Results with No Heat Rejection
The first step in the model verification process was to compare the model
pavement slab temperatures to measured slab temperatures during times when no heat
was being rejected to the slab. This comparison allowed a validity check of the
simulation of the several environmental heat transfer mechanisms occurring within the
pavement. Simulated and actual pavement average hourly temperatures are shown for
the parking lot test section in Figure 4-8 for an 8-day period in July 1998 when no heat
was rejected to the pavement. Therefore, in this case, the model is driven by weather
data input only.
A review of the plot in Figure 4-8 shows that the model slab surface temperatures
compare favorably to the measured slab surface temperatures. The simulated peak daily
and nightly temperatures are generally lower than the measured values. These
discrepancies are generally within 3
o
F (1.67
o
C) .
4.5.2. Model Comparison to Experimental Results with Heat Rejection
Heat rejection to the bridge-deck test section was simulated for the nights of
March 7-8, 1996 and March 24-25, 1996. During the night of March 7-8, 1996, a thin
layer of ice had formed on the bridge deck prior to the heat rejection test. Heat rejection
to the parking lot test section was simulated over a 36-day period from November 12 to
122
December 18, 1998. Input data to the model consisted of weather data as described
previously in addition to measured slab heat exchanger supply water temperatures and
flow rates on 5- minutely time intervals for the bridge deck field tests and 6-minutely
time intervals for the parking lot field tests.
Figure 4-8. Comparison of observed and simulated slab surface temperatures for
the parking lot test section with no heat rejection to the slab.
Since different parameters had been measured during the bridge deck field tests
and the parking lot field tests, the model performance was evaluated accordingly. For
comparison of the model results to the bridge deck field test data, the model performance
was evaluated by comparing (1) the simulated to the observed bridge deck surface
temperature and (2) the simulated to the observed return temperature of the heat
exchange fluid. For comparison of the model results to the parking lot field test data, the
70
75
80
85
90
95
100
105
110
115
120
125
130
22-Jul 23-Jul 24-Jul 25-Jul 26-Jul 27-Jul 28-Jul 29-Jul 30-Jul
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
21.1
23.9
26.7
29.4
32.2
35.0
37.8
40.6
43.3
46.1
48.9
51.7
54.4
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Air Actual Slab Model Slab
123
model performance was evaluated by comparing (1) the simulated to the observed return
temperature of the heat exchange fluid and (2) the simulated cumulative heat rejected to
the pavement slab to the measured water heating element and pump power input. These
comparisons are shown in Figures 4-9 and 4-10 respectively.
A review of the plots in Figure 4-9 shows that model predicted the bridge deck
surface temperature and fluid return temperature quite well. For the night of March 7-8,
1996, the average observed and modeled surface temperatures were 27.9
o
F (-2.27
o
C) and
27.2
o
F (-2.67
o
C), respectively, and the average observed and modeled fluid return
temperatures were 84.8
o
F (29.33
o
C) and 85.1
o
F (29.50
o
C), respectively. For the night of
March 24-25, 1996, the average observed and modeled surface temperatures were 36.9
o
F
(2.72
o
C) and 36.8
o
F (2.67
o
C), respectively, and the average observed and modeled fluid
return temperatures were 85.0
o
F (29.44
o
C) and 85.7
o
F (29.83
o
C), respectively. One main
purpose for the bridge deck simulations was to verify the model’s applicability to system
performance over relatively short time periods. Another important purpose of these
simulations was to verify the model’s applicability to the serpentine pipe arrangement.
The quantity of data was limited regarding snow and ice conditions and further
experiments need to be conducted to fully evaluate the model’s capability to account for
snow and ice events.
124
Figure 4-9. Comparison of observed and simulated slab surface temperatures and
heat exchange fluid return temperatures for the bridge deck test
section for the nights of (a) March 7-8, 1996 and (b) March 24-25,
1996.
(a)
0
10
20
30
40
50
60
70
80
90
100
0 1 2 3 4 5 6 7 8 9 10 11
Elapsed Time (hours) from 11:30 PM Mar. 7, 1996
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
-12.2
-6.7
-1.1
4.4
10.0
15.6
21.1
26.7
32.2
37.8
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Model Avg. Bridge Surface Temperature Experimental Bridge Surface Temperature
Model Fluid Return Temperature Experimental Fluid Return Temperature
(b)
10
20
30
40
50
60
70
80
90
100
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Elapsed Time (hours) from 8:30 PM Mar. 24, 1996
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
-12.2
-6.7
-1.1
4.4
10.0
15.6
21.1
26.7
32.2
37.8
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Model Avg. Bridge Surface Temperature Experimental Bridge Surface Temperature
Model Fluid Return Temperature Experimental Fluid Return Temperature
125
Figure 4-10. Comparison of observed and simulated (a) heat exchange fluid
return temperatures and (b) heat rejected to the parking lot test
section.
(a)
66
68
70
72
74
76
78
80
12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec 12-Dec 17-Dec
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
18.9
20.0
21.1
22.2
23.3
24.4
25.6
26.7
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Actual Return Model Return
(b)
0
200
400
600
800
1000
1200
1400
12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec 12-Dec 17-Dec
C
u
m
m
u
l
a
t
i
v
e

H
e
a
t

R
e
j
e
c
t
e
d

(
k
W
-
h
r
)
Heating Element + Pump Power Model Heat Rejected
126
A review of the plots in Figure 4-10 shows that model also predicted the
fluid return temperature and heat rejected to the parking lot test section favorably. The
average observed and modeled fluid return temperatures over the test period were 73.1
o
F
(22.8
o
C) and 73.4
o
F (23.0
o
C), respectively. At the end of the 36-day test period, the
percent difference between the cumulative simulated heat rejected and the cumulative
measured heat rejected is –5.01 %. These discrepancies may be due partly to heat losses
from the supply/return pipes to the ground and to the atmosphere in the equipment
building. A model and experimental uncertainty analysis is presented in Appendix A.
Further explanation is required regarding the parking lot test section simulations.
Explicit modeling of a slinky pipe is difficult because there are no true lines of symmetry
along its cross-section. As shown in Figure 4-2b, the pipe spacings are not uniform.
Another complicating factor is quantifying the thermal interferences induced by overlap
of the pipe coils. To circumvent these difficulties, a heuristic approach was used to
determine an effective pipe spacing along the center-line of the slinky coil. For a 10-inch
pitch, an effective pipe spacing of 8.4 in. (0.21 m) was determined by taking the
arithmetic average of the pipe spacing at the points where the pipe coils overlap.
However, this approach may not be as appropriate for other slinky configurations.
4.5.3. Model Application
To illustrate the applicability of the model as well as the viability of using
pavement heating systems as supplemental heat rejecters in GSHP systems, a model of a
hypothetical GSHP system was constructed in the TRNSYS modeling environment. A
127
simplified system schematic is shown in Figure 4-11. For this example, a pavement
heating system similar to the parking lot test section described above was used. That is, a
slinky coil installed at the interface of a 6-inch (0.1524 m) concrete slab and underlying
sand fill material. Each of the component models is described briefly below.
Figure 4-11. System schematic for the example model of a GSHP system with a
pavement heating system supplemental heat rejecter.
The building is not modeled explicitly in this application. The hourly building
thermal loads are pre-computed using a proprietary building energy analysis program and
are read from a file and passed to the heat pump subroutines. The building, the same one
described in Section 3.5.3, is an actual four-story, 45,000-ft
2
(4181-m
2
) office building
located in Tulsa, Oklahoma and is highly cooling dominated. The building thermal loads
are shown in Figure 3-8.
Building
Loads
Ground Loop
Heat Exchanger
Pavement Heating
System
Heat
Pumps
Controller
T
T
128
A simple water-to-air heat pump model was developed for this and other GSHP
system simulations. Inputs to the model include sensible and latent building loads,
entering fluid temperature, and fluid mass flow rate. The model uses quadratic curve-fit
equations to manufacturer’s catalog data to compute the heat of rejection in cooling
mode, heat of absorption in heating mode, and the heat pump power consumption.
Outputs provided by the model include exiting fluid temperature, power consumption,
and fluid mass flow rate. In this application, one heat pump component model handles
the heating load and a second heat pump component model handles the cooling load.
The ground-loop heat exchanger model used in this application is that described
by Yavuzturk and Spitler (1999) which is based partly on the work of Eskilson (1987)
who developed “long time-step” (monthly) response factors for vertical ground-coupled
U-tube heat exchangers. The model of Yavuzturk and Spitler (1999) extends the work of
Eskilson (1987) to hourly or less (“short-time step”) time intervals. The development of
the short time-step response factors are based on an analytically validated, transient two-
dimensional implicit finite volume model (Yavuzturk et al., 1999) that simulates the heat
transfer over a vertical U-tube ground heat exchanger. In this application, the modeled
borehole field consisted of one hundred 250-feet (76.2-m) deep boreholes arranged in a
10 by 10 square pattern. A total system flow rate of 270 gpm (61.36 m
3
/hr) was assumed.
Representative thermal properties of sedimentary rock were chosen.
Ancillary components such as pumps, t-pieces, flow diverters, and the differential
controller are described by SEL (1997). The control strategy used to activate the
circulating pump to the pavement was chosen somewhat arbitrarily by using the
129
temperature difference between the pavement surface and the exiting fluid temperature
from the heat pumps. When this temperature difference exceeds 9
o
F (5
o
C), the
circulating pump to the pavement is energized and heat will be rejected to the pavement.
During these times, all flow is diverted to the pavement system. The properties of each
heat exchanger coil in the example model are the same as those described in the parking
lot test section experiments. Hourly input weather data for the model were taken from a
typical meteorological year (TMY) record for Tulsa, Oklahoma.
The model was run for two cases for a duration of 3 years with a time step of one
hour. The first case was the GSHP system with no pavement and the second case was the
GSHP system with the pavement. Hourly heat pump entering water temperatures are
shown in Figure 4-12 for both cases.
Figure 4-12. Entering heat pump water temperatures for the example GSHP
system simulation with no pavement heating and with a 24,000 ft
2
(2230 m
2
) parking lot with pavement heating.
30
40
50
60
70
80
90
100
110
120
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (years)
T
e
m
p
e
r
a
t
u
r
e

(
o
F
)
10.0
15.6
21.1
26.7
32.2
37.8
43.3
48.9
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
no
pavement
with
pavement
130
A review of the data presented in Figure 4-12 shows the advantages of using a
pavement heating system as a supplemental heat rejecter. Assuming that a maximum
heat pump entering water temperature of 100
o
F (37.78
o
C) is desirable, the system without
the pavement system would fail during the second year of operation. In fact, based on the
results of a ground-loop heat exchanger sizing program (Spitler et al., 1996), the
boreholes of a 10 by 10 square pattern would need to be approximately 400 feet (121.9
m) deep to accommodate the cooling dominated loads of this building for 20 years of
operation. Such a system would be eliminated from consideration early on in the design
phase because of excessive first cost.
Using the TRNSYS model as a design tool, the size of the pavement supplemental
heat rejecter was determined under the assumption that the 10 by 10 borehole field could
not be feasibly deeper than 250 feet (76.2 m). The heat pump entering water
temperatures for the GSHP system with the pavement heating system shown in Figure 4-
12 were produced by simulating a 24,000 ft
2
(2230 m
2
) with 200 slinky heat exchanger
coils. This size of a parking lot would have a capacity of 75 cars (in parking lot design,
an area per car of 320 ft
2
(29.73 m
2
) will allow for access through lots (Lindeburg, 1992)
). A summary of the pavement system performance is given in Table 4-1. By adding the
pavement supplemental heat rejecter in this example, the depth of the borehole field
could be decreased by approximately 35%. A more detailed system analysis could
involve system life-cycle operating cost analyses, control strategy variations, and design
variable optimization.
131
Table 4-1. Summary of Pavement Heating System Performance for
Example GSHP System Simulation
Year Hours Average Surface Heat Pump Maximum Heat Rejected
ON Temperature Entering Fluid
Temperature
(
o
F) (
o
C) (
o
F) (
o
C) (kBtu) (MJ)
1 6245 63.65 17.58 97.71 36.51 3,418,368 3,605,695
2 6448 63.96 17.76 99.41 37.45 3,670,277 3,871,409
3 6603 64.15 17.86 100.79 38.22 3,837,759 4,048,068
4.6. Concluding Remarks and Recommendations for Future Work
A design and simulation tool for modeling the performance of a pavement heating
system as a supplemental heat rejecter in ground-source heat pump systems has been
developed. The model has been developed in the TRNSYS modeling environment (SEL,
1997) and can be coupled to other GSHP system component models for short-time step
(hourly or minutely) system analyses. The model has been validated by comparing
simulation results to experimental data.
The model accounts for several environmental heat transfer mechanisms plus
convective heat transfer due to a closed-loop heat exchanger coil. The heat transfer fluid
is assumed to be carried by a series of pipes in a “serpentine” configuration or a “slinky”
configuration. Environmental heat transfer mechanisms that are simulated by the model
include solar radiation heat gain, convection heat transfer to the atmosphere, thermal or
long-wave radiation heat transfer, and sensible and latent heat and mass transfer due to
rain and snow. The model uses an explicit finite-difference method to solve the transient
two-dimensional heat conduction equation. Some outputs provided by the model include
132
the pavement surface temperature, the exiting fluid temperature, and the amount of heat
rejected to the pavement slab.
An example application has been presented to demonstrate the use of the model as
well as the viability of the use of pavement areas as supplemental heat rejecters in GSHP
systems. Through this example, it is shown that ground-loop heat exchanger sizes can be
significantly decreased by incorporating a pavement heating systems into the GSHP
system.
The potential exists for significantly increasing the performance of pavement
heating systems used as supplemental heat rejecters in GSHP systems. Further research
is suggested in the following areas:
• Optimization of the design procedure and control strategy as described for the
shallow pond supplemental heat rejecter in Section 3.6.
• Additional validation of the model, using data from a working system, would be
useful.
• Additional validation of the model, using data collected under a wider range of
weather conditions (i.e. rain, snow, and ice conditions), would be useful.
• The impact of the pipe configuration on the overall system performance, particularly
with regard to accounting for slinky pipe spacing in models.
• Application of the model to other uses such as modeling the performance of
horizontal ground-loop heat exchanger systems and snow melting systems.
133
5. Summary and Conclusions
Ground-source heat pump (GSHP) systems have received considerable attention
in recent decades as an alternative energy source for residential and commercial space
heating and cooling applications. GSHP systems offer proven advantages over
conventional heating and cooling systems, specifically with respect to efficiency,
maintenance costs, and overall operating costs. Depending on the configuration, a GSHP
system may either use the earth, ground water, or surface water as a heat source and/or
sink. Hybrid GSHP systems use combinations of these or a combination of a GSHP
system with conventional equipment (i.e. a cooling tower).
This study has dealt with the modeling of vertical closed-loop and hybrid, ground-
source heat pump systems. The challenges associated with the design of these systems
originate from the fact that they present a unique type of heat transfer problem. First,
there are inherent inabilities to make direct observations in the subsurface environment
with respect to both space and time. Second, heat transfer within the subsurface
environment can be highly transient. Consequently, a considerable amount of research in
the past decade has been geared toward optimizing the design and performance of GSHP
systems and this study is part of those efforts.
The objectives of this study were threefold: (1) to examine the effects of ground-
water flow on closed-loop GSHP systems, (2) to develop a design and simulation tool for
modeling the performance of a shallow pond as a supplemental heat rejecter with closed-
loop GSHP systems, and (3) to develop a design and simulation tool for modeling the
134
performance of a hydronic pavement heating system as a supplemental heat rejecter with
closed-loop GSHP systems.
Chapter 2 of this thesis has presented a preliminary assessment of the effects of
ground-water flow on closed-loop ground-source heat pump systems. A compilation of
“typical” hydraulic and thermal properties of soils and rocks was used in the study.
A simple but useful method of assessing the relative importance of heat
conduction in the ground versus heat advection by moving ground water was
demonstrated through the use of the dimensionless Peclet number.
A finite-element numerical ground-water flow and heat transport model was used
to simulate and observe the effects of ground-water flow on the heat transfer from a
single U-tube closed-loop ground heat exchanger in various geologic materials. From
those simulations, it appears that it is only in geologic material with high hydraulic
conductivities, such as coarse-grained soils (sands and gravels) and in rocks exhibiting
secondary porosities such fractures and solution channels, that ground-water flow could
be expected to have a significant effect on closed-loop heat exchanger performance.
The finite-element numerical ground-water flow and heat transport model was
also used to examine the effect of ground-water flow on in-situ thermal conductivity test
results. This was done by numerically simulating test conditions around a single
borehole under different flow conditions. As expected, in all cases of ground-water flow,
135
these values were artificially high. Results from one week test data have been shown to
be no more reliable than data from 50-hour tests when significant ground-water flow is
present.
The finite-element numerical ground water flow and heat transport model was
also used to simulate the 10-year performance of borehole fields designed from
application of conventional design procedures using the derived thermal conductivity
data. Even the presence of moderate ground-water flows had the effect of removing the
year-by-year increase in ground temperature found in systems where there is no ground-
water flow. The borehole fields designed using conventional methods and the derived
effective thermal conductivities were generally over-designed. However, in some cases
at very high ground-water flow rates, temperatures were found to rise above the design
criteria.
From this preliminary assessment of the effects of ground-water flow, it appears
difficult to adapt results from current design and in-situ measurement methods to fully
account for ground-water flow conditions. Over the last decade, considerable progress
has been made in developing both in-situ test methods and design procedures for
borehole field design for situations where there is no ground-water flow. Research would
be required in a number of areas before the same progress could be made to deal with the
situations of ground-water flow. These include:
• Identification of suitable numerical and/or analytical models that include ground-
water flow and could be used to analyze in-situ test data.
136
• Experimental investigation of potential in-situ test data analysis methods at sites with
significant ground-water flow.
• Identification of suitable design methods, or adaptations to existing methods, that
could be used for closed-loop ground heat exchanger design.
• Development of design guidelines and software tools that could be used by practicing
engineers for in-situ testing and system design tasks in situations of significant
ground-water flow.
Chapter 3 of this thesis has described the development and validation of a model
for simulating the performance of a shallow pond as a supplemental heat rejecter with
closed-loop ground-source heat pump systems. The model has been developed in the
TRNSYS modeling environment and can be coupled to other GSHP system component
models for short-time step (hourly or minutely) system analyses. The model has been
validated by comparing simulation results to experimental data.
The model accounts for several natural heat transfer mechanisms within a surface
water body plus convective heat transfer due to a closed-loop heat exchanger coil. The
heat transfer fluid is assumed to be carried by a series of pipes in the form of bundle
spools or “slinky” coils. Environmental heat transfer mechanisms that are simulated by
the model include solar radiation heat gain, heat and mass transfer due to evaporation,
convection heat transfer to the atmosphere, thermal or long-wave radiation heat transfer,
conduction heat transfer to the surrounding soil or fill material, and ground water
discharge contributions. The solution scheme involves a lumped-capacitance approach
137
and the resulting first-order differential equation describing the overall energy balance on
the pond is solved numerically. Some outputs provided by the model include average
pond temperature, exiting fluid temperature, and heat rejected to the pond.
An example application has been presented to demonstrate the use of the model as
well as the viability of the use of shallow ponds as supplemental heat rejecters in GSHP
systems. Through this example, it is shown that the size of ground-loop heat exchangers
can be significantly decreased by incorporating a shallow pond into the GSHP system.
The potential exists for significantly increasing the performance of shallow ponds
used as supplemental heat rejecters in GSHP systems. Further research is suggested in
the following areas:
• Optimization of the design procedure and control strategy. Hybrid ground source
heat pump systems have many degrees of freedom; there are tradeoffs between the
reduction in size of the ground loop heat exchanger, the size of the pond, and the
control strategy. To more fully understand this, additional research using the
simulation techniques developed in this paper is needed. This research would also
take into account the economic costs and benefits, which we have not investigated.
• Additional validation of the model, using data from a working system, would be
useful.
• Extension of the model to cover deep ponds for situations where an existing pond or
lake is available.
138
• The use of spray fountains and other aeration devices in the pond to enhance pond
cooling.
• The impact of pipe configuration within the pond on the overall system performance.
Chapter 4 of this thesis has described the development and validation of a model
for simulating the performance of a pavement heating system as a supplemental heat
rejecter with closed-loop ground-source heat pump systems. The model has been
developed in the TRNSYS modeling environment and can be coupled to other GSHP
system component models for short-time step (hourly or minutely) system analyses. The
model has been validated by comparing simulation results to experimental data.
The model accounts for several environmental heat transfer mechanisms plus
convective heat transfer due to a closed-loop heat exchanger coil. The heat transfer fluid
is assumed to be carried by a series of pipes in a “serpentine” configuration or a “slinky”
configuration. Environmental heat transfer mechanisms that are simulated by the model
include solar radiation heat gain, convection heat transfer to the atmosphere, thermal or
long-wave radiation heat transfer, and sensible and latent heat and mass transfer due to
rain and snow. The model uses the finite-difference method to solve the transient two-
dimensional heat conduction equation. Some outputs provided by the model include the
pavement surface temperature, the exiting fluid temperature, and the amount of heat
rejected to the pavement slab.
139
An example application has been presented to demonstrate the use of the model as
well as the viability of the use of pavement areas as supplemental heat rejecters in GSHP
systems. Through this example, it is shown that ground-loop heat exchanger sizes can be
significantly decreased by incorporating a pavement heating systems into the GSHP
system.
The potential exists for significantly increasing the performance of pavement
heating systems used as supplemental heat rejecters in GSHP systems. Further research
is suggested in the following areas:
• Optimization of the design procedure and control strategy as described for the
shallow pond supplemental heat rejecter.
• Additional validation of the model, using data from a working system, would be
useful.
• Additional validation of the model, using data collected under a wider range of
weather conditions (i.e. rain, snow, and ice conditions), would be useful.
• The impact of the pipe configuration on the overall system performance, particularly
with regard to accounting for slinky pipe spacing in models.
• Application of the model to other uses such as modeling the performance of
horizontal ground-loop heat exchanger systems and snow melting systems.
140
REFERENCES
Adlam, T.N. 1950. Snow Melting. The Industrial Press, New York, NY.
ASHRAE, 1995. ASHRAE Handbook, HVAC Applications. American Society of
Heating, Refrigeration and Air-Conditioning Engineers, Inc., Atlanta, GA.
ASHRAE, 1995b. Commercial/Institutional Ground-Source Heat Pump
Engineering Manual. American Society of Heating, Refrigeration and Air-Conditioning
Engineers, Inc., Atlanta, GA.
ASHRAE, 1997. ASHRAE Handbook, Fundamantals. American Society of Heating,
Refrigeration and Air-Conditioning Engineers, Inc., Atlanta, GA.
Austin, W.A. III, C. Yavuzturk, and J. D. Spitler. 2000. Development of an In-Situ
System for Measuring Ground Thermal Properties. Submitted for publication to
ASHRAE Transactions.
Bear, J. 1972. Dynamics of Fluids in Porous Media. New York: Dover Publications,
Inc.
BLAST 1986. BLAST (Building Loads and System Thermodynamics). Urbana-
Champaign: University of Illinois, BLAST Support Office.
Bliss, R.W., 1961. Atmospheric Radiation Near the Surface of the Ground. Solar
Energy, 5(103).
Cantrell, J.M. and W.J. Wepfer, 1984. Shallow Ponds for Dissipation of Building Heat:
A Case Study. ASHRAE Transactions 90(1): 239-246.
Carslaw, H. S. and J. C. Jaeger. 1946. Conduction of Heat in Solids. Oxford, U.K.:
Claremore Press.
Chapman, W.P. 1952. Design of Snow Melting Systems. Heating and Ventilating.
April, pp. 96-102.
Chemical Rubber Company (CRC), 1980. Handbook of Chemistry and Physics, 61
st
Edition. CRC Press, Cleveland, OH.
Churchill, S.W. and H.H.S. Chu. 1975. Correlating Equations for Laminar and Turbulent
Free Convection from a Horizontal Cylinder. International Journal of Heat and
Mass Transfer, 18: 1049-1053.
Dake, J.M.K. and D.R.F. Harleman, 1969. Thermal Stratification in Lakes: Analytical
and Laboratory Studies. Water Resources Research, 5(2): 484-495.
141
Domenico, P.A. and F. W. Schwartz. 1990. Physical and Chemical Hydrogeology. New
York, Chicester, Brisbane, Toronto, Singapore: John Wiley & Sons.
Driscoll, F. G. 1986. Groundwater and Wells. St. Paul Minnesota: Johnson Filtration
Systems, Inc.
Duffie, J.A. and W.A. Beckman, 1991. Solar Engineering of Thermal Processes, 2
nd
Edition. John Wiley and Sons.
Elliot, R.L., F.V. Brock, M.L. Stone, and S.L. Sharp. 1994. Configuration Decisions for
an Automated Weather Station Network. Applied Engineering in Agriculture,
10(1): 45-51.
Eklof, C. and S. Gehlin. 1996. TED – A Mobile Equipment for Thermal Response Test.
Master’s Thesis 1996:198E. Lulea University of Technology, Sweden.
Eskilson, P. 1987. Thermal Analysis of Heat Extraction Boreholes. Doctoral Thesis,
University of Lund, Department of Mathematical Physics. Lund, Sweden.
Fetter, C. W. 1988. Applied Hydrogeology. 2
nd
Edition, Columbus, Ohio: Merrill
Publishing Co.
Freeze, R. A. and J. A. Cherry. 1979. Groundwater. Englewood Cliffs, New Jersey:
Prentice Hall, Inc.
Gehlin, S. and B. Nordell. 1998. Thermal Response Tests of Boreholes – Results from
In Situ Measurements. Transaction of The Geothermal Project at Richard
Stockton College Conference. March 16-17, 1998.
Grant, M.A., I.G. Donaldson, and P.F. Bixley, 1982. Geothermal Reservoir Engineering.
Academic Press.
Hart, D.P. and R. Couvillion, 1986. Earth-Coupled Heat Transfer. National Water Well
Association, Dublin, OH.
Hellstrom, G. 1991. Ground Heat Storage. Thermal Analyses of Duct Storage Systems.
Lund, Sweden: University of Lund, Department of Mathematical Physics.
Holman, J.P., 1972. Heat Transfer, 3
rd
Edition. McGraw-Hill Book Company.
Hull, J.R., K.V. Liu, W.T. Sha, J. Kamal, and C.E. Nielsen, 1984. Dependence of
Ground Heat Losses Upon Solar Pond Size and Perimeter Insulation – Calculated
and Experimental Results. Solar Energy, 33(1): 25-33.
IGSHPA. 1991. (Bose, J. E., Editor) Design and Installations Standards. Stillwater,
Oklahoma: International Ground Source Heat Pump Association.
142
Incropera, F.P. and D.P. DeWitt, 1996. Introduction to Heat Transfer, 3
rd
Edition. John
Wiley & Sons.
Ingersoll, L. R., O. J. Zobel, and A. C. Ingersoll. 1954. Heat Conduction with
Engineering, Geological, and Other Applications. New York: McGraw-Hill.
Irvine, T.F. Jr. and P.E. Liley, 1984. Steam and Gas Tables with Computer Equations.
Academic Press, Inc.
Jobson, H., 1962. The Dissipation of Excess Heat from Water Systems. Journal of the
Power Division, Proceedings of the American Society of Civil Engineers.
99(P01), May, pp. 89-103.
Kavanaugh, S. P. 1984. Simulation and experimental verification of vertical ground-
coupled heat pump systems. Ph.D. dissertation. Stillwater, Oklahoma: Oklahoma
State University.
Kavanaugh, S.P. and K. Rafferty, 1997. Ground-Source Heat Pumps. Design of
Geothermal Systems for Commercial and Institutional Buildings. American
Society of Heating, Refrigeration and Air-Conditioning Engineers, Inc., Atlanta,
GA.
Kelvin, Sir W. Thomson. 1882. Mathematical and Physical Papers. II, p.41 ff.
Kilkis, I.B. 1994. Design of Embedded Snow Melting Systems: Part 1, Heat
Requirements – An Overall Assessment and Recommendations. ASHRAE
Transactions, 100(1):pp. 423-433.
Kishore, V.V.N. and V. Joshi, 1984. A Practical Collector Efficiency Equation for
Nonconvecting Solar Ponds. Solar Energy. 33(5): 391-395.
Liao, C.J. and T.D. Hogue. 1996. Thermal Stress Predictions for Geothermally Heated
Bridge Decks. Prepared for the State of Oklahoma Department of Transportation.
Lindeburg, M.R. 1992. Civil Engineering Reference Manual, 6
th
Edition. Professional
Publications, Inc., Belmont, CA.
Lindholm, G. F. and J. J. Vaccaro. 1988. (Back, W.B., J.S. Rosenshein, and P.R. Seaber,
Editors) Region 2, Columbia Lava Plateau (Chapter 5). In: The Geology of North
America, Vol. O-2, Hydrogeology. Boulder, Colorado: The Geological Society
of America, Inc..
Marrero, T.R. and E.A.Mason, 1972. Gaseous Diffusion Coefficients. Journal of
Physical and Chemical Reference Data. 1: 3-118.
McAdams, W.H., 1954. Heat Transmission, 3
rd
Edition. McGraw-Hill Book Company.
143
Mercer, J.W. and C.R. Faust. 1986. Ground-Water Modeling. National Water Well
Association.
Mei, V. C. and C. J. Emerson. 1985. New Approach for Analysis of Ground-Coil
Design for Applied Heat Pump Systems. ASHRAE Transactions. 91(2B):1216-
1224.
Mills, A.F., 1995. Basic Heat and Mass Transfer. Prentice Hall.
Muraya, N. K., D.L. O’Neal, and W.M. Heffington, 1996. Thermal Interference of
Adjacent Legs in a Vertical U-Tube Heat Exchanger for a Ground-Coupled Heat
Pump. ASHRAE Transactions 102(2):12-21.
Newell, T.A., 1984. Solar Ponds – Alternate Energy Systems. Military Engineer,
Journal of the Society of American Military Engineers, No. 492, March-April
1984: 110-113.
Pezent, M.C. and S.P. Kavanaugh, 1990. Development and Verification of a Thermal
Model of Lakes Used With Water Source Heat Pumps. ASHRAE Transactions
96(1): 574-582.
Ramsey, J.W., M.J. Hewett, T.H. Kuehn, and S.D. Petersen. 1999. Updated Design
Guidelines for Snow Melting Systems. ASHRAE Transactions, 105(1).
Raphael, J.M., 1962. Prediction of Temperatures in Rivers. Journal of the Power
Division, Proceedings of the American Society of Civil Engineers. 88(P02), July,
pp. 157-181.
Remund, C. 1998. Personal Communication. Northern Geothermal Support Center,
South Dakota State University, Brookings, SD.
Rottmayer, S. P., W. A. Beckman, and J. W. Mitchell. 1997. Simulation of a Single
Vertical U-Tube Ground Heat Exchanger in an Infinite Medium. ASHRAE
Transactions, 103(2):651-658.
Rubin, H., B.A. Benedict, and S. Bachu, 1984. Modeling the Performance of a Solar
Pond as a Source of Thermal Energy. Solar Energy, 32(6): 771-778.
Shonder, J. A. and J. V. Beck. 1999. Determining Effective Soil Formation Thermal
Properties From Field Data Using a Parameter Estimation Technique. ASHRAE
Transactions, 105(1).
Solar Energy Laboratory (SEL), University of Wisconsin-Madison. 1997. TRNSYS, A
Transient Systems Simulation Program, User’s Manual, Version 14.2.
144
Spencer, J.W., 1971. Fourier Series Representation of the Position of the Sun. Search,
2(5).
Spitler, J. D., C. Marshall, R. Delahoussaye, and M. Manicham. 1996. Users Guide of
GLHEPRO, School of Mechanical and Aerospace Engineering, Oklahoma State
University, Stillwater, OK.
Srinivasan, J. and A. Guha, 1987. The Effect of Reflectivity on The Performance
of a Solar Pond. Solar Energy. 39(4): 361-367.
Stephenson, D.A., A. H. Fleming, and D. M. Mickelson. 1988. (Back, W. B., J. S.
Rosenshein, and P. R. Seaber, Editors) Glacial Deposits (Chapter 35). In: The
Geology of North America, Vol. O-2, Hydrogeology. Boulder, Colorado: The
Geological Society of America, Inc..
United States Environmental Protection Agency. 1996. BIOSCREEN. Natural
Attenuation Decision Support System, User’s Manual. Cincinnati, Ohio:
National Risk Management Research Laboratory, Office of Research and
Development.
Vatnaskil Consulting Engineers. 1998. AQUA3D Groundwater Flow and Contaminant
Transport Model. Reykjavik, Iceland: Vatnaskil Consulting Engineers.
Wadivkar, O., 1997. An Experimental and Numerical Study of the Thermal
Properties of a Bridge Deck De-Icing System. Unpublished Masters Thesis, Oklahoma
State University, Stillwater, OK.
Wang, H.F. and M.P. Anderson. 1982. Introduction to Groundwater Modeling –
Finite Difference and Finite Element Methods. W.H. Freeman and Company, New York.
Weeks, J. B. and E. D. Gutentag. 1988. (Back, W. B., J. S. Rosenshein, and P. R.
Seaber, Editors) Region 17, High Plains (Chapter 20). In: The Geology of North
America, Vol. O-2, Hydrogeology. Boulder, Colorado: The Geological Society
of America, Inc..
Yavuzturk, C. and J.D. Spitler. 1999. A Short Time Step Response Factor Model for
Vertical Ground Loop Heat Exchangers. ASHRAE Transactions, 105(2).
Yavuzturk, C., J. D. Spitler, and S. J. Rees. 1999. A Transient two-dimensional Finite
Volume Model for the Simulation of Vertical U-Tube Ground Heat Exchangers.
ASHRAE Transactions, 105(2).
145
APPENDIX A
Model and Experimental Uncertainty Analysis
146
This appendix presents an analysis of parameter uncertainty and experimental
uncertainty regarding the validation and application of the pond and pavement models.
Model Uncertainty
This section presents a sensitivity analysis to quantify the uncertainty in the final
model results caused by uncertainties in the estimates of input and parameter values used
in the models. The results of this analysis are shown in Tables A-1 and A-2 for the pond
and the pavement models, respectively, and are discussed below.
The pond and pavement model uncertainty is quantified using influence
coefficients as described by Spitler et al. (1989). An influence coefficient is the partial
derivative of a simulation result with respect to a parameter which, for the purposes of
this study, is approximated as:
Parameter
Result
t Coefficien Influence


· (A-1)
where Result is the cumulative heat rejected and Parameter is a variable with some
uncertainty. The term Parameter should not be confused with TRNSYS model
parameters. Here, Parameter refers to any variable with a significant uncertainty,
including TRNSYS model parameters as well as TRNSYS model inputs such as weather
factors, fluid supply temperature, and fluid flow rate.
147
TABLE A-1
Pond Model Parameter Uncertainty Analysis
Modeled cumulative heat rejected from Nov.12, 1998 to Dec. 7, 1998: Horizontal Slinky 1486 kW-hr
Vertical Slinky 1508 kW-hr
Perturbed ∆(result) Uncertainty Model
Model ∆(parameter) in Parameter
Result Result Uncertainty
(kW-hr) (kW-hr) (%)
k
pipe
(W/m-
o
C)
Horizontal Slinky 0.391 (2) 0.06 (3) 0.5 1554 623.85 37.43 2.52%
Vertical Slinky 0.391 (2) 0.06 (3) 0.5 1578 642.20 38.53 2.56%
k
soil
(W/m-
o
C)
Horizontal Slinky 2.5 (4) 0.2 (1) 3.0 1520 68.00 13.60 0.92%
Vertical Slinky 2.5 (4) 0.2 (1) 3.0 1543 70.00 14.00 0.93%
Depth to constant sink (m)
Horizontal Slinky 7.5 (4) 2.5 (4) 3.0 1510 -5.33 13.33 0.90%
Vertical Slinky 7.5 (4) 2.5 (4) 3.0 1535 -6.00 15.00 0.99%
ε of water surface (--)
Horizontal Slinky 0.97 (5) 0.02 (1) 0.9 1473 185.71 3.71 0.25%
Vertical Slinky 0.97 (5) 0.02 (1) 0.9 1495 185.71 3.71 0.25%
Windspeed (m/s)
Horizontal Slinky 1 (a) 0.10 (6) 2.0 (a) 1847 361.00 36.10 2.43%
Vertical Slinky 1 (a) 0.10 (6) 2.0 (a) 1880 372.00 37.20 2.47%
Solar radiation
Horizontal Slinky 1 (a) 0.05 (6) 2.0 (a) 1242 -244.00 12.20 0.82%
Vertical Slinky 1 (a) 0.05 (6) 2.0 (a) 1256 -252.00 12.60 0.84%
Air temperature (
o
C)
Horizontal Slinky 0 (b) 0.35 (6) 2.0 (b) 1284 -100.86 35.30 2.38%
Vertical Slinky 0 (b) 0.35 (6) 2.0 (b) 1305 -101.43 35.50 2.35%
Relative humidity (%)
Horizontal Slinky 1 (a) 0.05 (6) 1.05 (a) 1452 -674.13 33.71 2.27%
Vertical Slinky 1 (a) 0.05 (6) 1.05 (a) 1474 -679.80 33.99 2.25%
Supply fluid temperature (
o
C)
Horizontal Slinky 0 (b) 0.12 (7) 0.2 (b) 1400 -431.90 51.83 3.49%
Vertical Slinky 0 (b) 0.12 (7) 0.2 (b) 1413 -477.43 57.29 3.80%
Supply fluid flow rate (gpm)
Horizontal Slinky 0 (b) 0.025 (7) 0.05 (b) 1486 6.66 0.17 0.01%
Vertical Slinky 0 (b) 0.025 (7) 0.05 (b) 1508 7.04 0.18 0.01%
Total Uncertainty in Results
Horizontal Slinky 6.13%
Vertical Slinky 6.35%
NOTES:
(a) - a multiplication factor is used to vary the hourly input value of this item for this uncertainty analysis
(b) - an addition factor is used to vary the hourly input value of this item for this uncertainty analysis
(1) - estimated value
(2) - value given by pipe manufacturer Phillips Driscopipe
(3) - based on data given by Mills (1995)
(4) - based on in-situ thermal conductivity test data
(5) - value given by Incropera and DeWitt (1996)
(6) - measurement error given by Elliot et al. (1994)
(7) - based on standard error of calibration curves
Perturbed
Model
Value
Item Adjusted Value
Used in
Final
Analysis
Uncertainty
in
Parameter
148
TABLE A-2
Pavement Model Parameter Uncertainty Analysis
Modeled cummulative heat rejected from Nov.12, 1998 to Dec. 19, 1998: Horizontal Slinky 1255 kW-hr
Perturbed ∆(result) Uncertainty Model
Model ∆(parameter) in Parameter
Result Result Uncertainty
(kW-hr) (kW-hr) (%)
k
pipe
(W/m-
o
C)
Slinky test section 0.391 (2) 0.06 (3) 0.5 1293 348.62 20.92 1.67%
k
concrete
(W/m-
o
C)
Slinky test section 1.663 (4) 0.2 (1) 2.9 1314 47.70 9.54 0.76%
k
soil
(W/m-
o
C)
Slinky test section 0.4 (5) 0.2 (1) 1.0 1146 -181.67 36.33 2.90%
α of concrete (--)
Slinky test section 0.4 (6) 0.1 (6) 0.6 1135 -600.00 60.00 4.78%
ε of concrete (--)
Slinky test section 0.9 (6) 0.05 (6) 0.8 1192 630.00 31.50 2.51%
Windspeed (m/s)
Slinky test section 1 (a) 0.10 (7) 2.0 (a) 1327 72.00 7.20 0.57%
Air temperature (
o
C)
Slinky test section 0 (b) 0.35 (7) 2.0 (b) 1092 -81.26 28.44 2.27%
Relative humidity (%)
Slinky test section 1 (a) 0.05 (7) 1.05 (a) 1250 -103.99 5.20 0.41%
Supply fluid temperature (
o
C)
Slinky test section 0 (b) 0.12 (8) 0.2 (b) 1089 -831.18 99.74 7.95%
Supply fluid flow rate (gpm)
Slinky test section 0 (b) 0.025 (8) 0.05 (b) 1256 16.48 0.41 0.03%
Total Uncertainty in Results
Slinky test section 10.47%
NOTES:
(a) - a multiplication factor is used to vary the hourly input value of this item for this uncertainty analysis
(b) - an addition factor is used to vary the hourly input value of this item for this uncertainty analysis
(1) - estimated value
(2) - value given by pipe manufacturer Phillips Driscopipe
(3) - based on data given by Mills (1995)
(4) - based on data given by Tinker and Cabrera (1992)
(5) - value given by Spitler et al. (1996) for light soil
(6) - value given by ASHRAE (1997) light-colored surfaces
(7) - measurement error given by Elliot et al. (1994)
(8) - based on standard error of calibration curves
Perturbed
Model
Value
Item Adjusted Value
Used in
Final
Analysis
Uncertainty
in
Parameter
149
The influence coefficient for each parameter of interest is determined by running
the model with a “perturbed” value of the parameter. The Uncertainty in Result, with
respect to the parameter, is then determined by:
Uncertainty in Result = (Influence Coefficient )x(Uncertainty in Parameter) (A-2)
where Uncertainty in Parameter is the estimated parameter error (for example, t 0.5
W/m-
o
C). The non-dimensional Model Parameter Uncertainty is then determined by:
Result
Result in y Uncertaint
y Uncertaint Parameter Model · (A-3)
The errors introduced to the model result by each parameter are assumed to be
independent of each other. Therefore, the Total Uncertainty in Result is given by:

·
·
n
1 i
i
y) Uncertaint Parameter (Model Result in y Uncertaint Total
2
(A-4)
where n is the number of parameters considered in the analysis.
The parameters investigated in this uncertainty analysis can be divided into two
groups: (1) TRNSYS model parameters and (2) TRNSYS model inputs. For the pond
model, these items consisted of: (1) (a) pipe thermal conductivity (k
pipe
), (b) soil thermal
conductivity (k
soil
), (c) depth to constant sink temperature, and (d) emissivity of the water
150
surface (ε) and (2) (a) wind speed, (b) solar radiation, (c) air temperature, (d) relative
humidity, (e) supply fluid temperature, and (f) supply fluid flow rate. For the pavement
model, these items consisted of: (1) (a) pipe thermal conductivity (k
pipe
), (b) concrete
thermal conductivity (k
concrete
), (c) soil thermal conductivity (k
soil
), (d) absorptivity of the
concrete surface (α) and (d) emissivity of the concrete surface (ε) and (2) (a) wind speed,
(b) air temperature, (c) relative humidity, (d) supply fluid temperature, and (e) supply
fluid flow rate. The model values used in the final analysis and the uncertainty in each
value were taken from the sources shown in Tables A-1 and A-2.
For the pond model, the supply fluid temperature has the greatest effect on the
model uncertainty, on the order of 3.5%. Four other parameters have lesser uncertainties
on the model results, on the order of 2.5%; these are k
pipe
, wind speed, air temperature,
and relative humidity. Of still lesser significance are k
soil
, the depth to constant sink
temperature, and the solar radiation, which produce model uncertainties on the order of
0.9%. The emissivity of the water surface has a relatively insignificant uncertainty of
0.25%. The fluid flow rate has an insignificant effect on the model uncertainty,
contributing only 0.01% to the total.
For the pavement model, the supply fluid temperature also has the greatest effect
on the model uncertainty, on the order of 8%. The absorptivity of the concrete surface
possesses the next highest uncertainty, on the order of 5%. Air temperature, k
soil
, and ε of
the concrete surface all have uncertainties on the order of 2.5%. k
pipe
has an uncertainty
of 1.7% and wind speed and relative humidity have relatively low uncertainties on the
151
order of 0.5%. As with the pond model, the fluid flow rate has an insignificant effect on
the model uncertainty, contributing only 0.03% to the total.
Based on the results shown in Table A-1, the total uncertainty in the results of the
pond model are t 6.13% for the horizontal slinky case and t 6.35% for the vertical
slinky case. Based on the results shown in Table A-2, the total uncertainty in the results
of the pavement model is t 10.47% for the slinky case.
The results of this uncertainty analysis can be used to identify items of concern to
one who may use the model(s) for the purposes of GSHP system design or system
simulation. For this study, estimates of the TRNSYS model parameters are likely no
better than those that could be made by a typical user, and therefore, similar uncertainties
can be expected for those items. However, the TRNSYS model inputs cannot be known
exactly by a typical user because they consist of weather factors and system fluid
temperature and flow rate.
A user performing some type of system design or simulation must use typical or
synthetic weather data based on historical observations for the location of interest. The
use of “typical weather data” is one inherent difficulty in building simulation studies
because the it is obviously impossible to quantify uncertainties in future weather
conditions. However, the longer the simulation time (10-20 years), the greater the
probability is that average actual weather conditions approach the “typical weather
152
conditions”. For shorter-term simulations (up to 5 years), the conservative user may want
to use weather data with more extreme values, if available.
Other model inputs include the supply fluid temperature and flow rate which will
be provided either as a constant value or by another model. From the results of this
uncertainty analysis, model results are much more sensitive to the supply fluid
temperature, and a user should therefore pay more attention to this item.
Experimental Uncertainty
This section presents an analysis of the experimental errors contributing to the
measurement of heat rejected to the ponds and to the concrete slab. The purpose of this
analysis is to quantify the error in the experimental results that were used to validate the
pond and pavement models. The results of this analysis are shown in Tables A-3 and are
discussed below.
The main sources of error in the measurements of heat rejection to the ponds and
to the concrete slab are identified as: (1) calibration error in the watt meter used for
power measurements and (2) heat losses or gains to the ground from the pipes between
the heat rejecter and the instrumentation building.
Error in the watt meter measurements was determined from the calibration
procedure. Heat losses/gains to the ground from the supply and return pipes were
153
estimated using the pavement model. The model was run from October 1, 1998 to
December 19, 1998 with actual weather conditions and assuming a constant fluid supply
temperature of 75
o
F (23.9
o
C) and constant fluid supply flow rate of 4 gpm (908 kg/hr).
The thermal conductivity of the ground and the pipe backfill material were taken as 1.0
Btu/hr-ft-
o
F (1.73 W/m-
o
C) and the average pipe burial depth was taken as 1.5 ft (0.48
m). The model was used to determine the heat losses per foot of pipe for the period of
November 12, 1998 to December 19, 1998, which is the period of interest for the model
validation. The cumulative heat lost through the supply/return pipes was then estimated
by considering the distances from the instrumentation building to each of the
supplemental heat rejecters (62 ft (18.9 m) for the pond with the vertical slinky, 37 ft
(11.3 m) for the pond with the horizontal slinky, and 12 ft (3.7 m) for the concrete slab).
TABLE A-3
Heat Rejection Experimental Uncertainty Analysis
Experimental
Uncertainty
Error in Power Measurement (a) 0.50%
Estimated Cumulative Heat Losses to Ground: (b)
Pond with horizontal slinky 20.25 kW-hr 1.36%
Pond with vertical slinky 33.83 kW-hr 2.24%
Slab with slinky 10.92 kW-hr 0.87%
Total Experimental Uncertainty
Pond with horizontal slinky 1.45%
Pond with vertical slinky 2.30%
Slab with slinky 1.00%
NOTES:
(a) - based on instrument calibration precision
(b) - includes heat lost through supply and return headers, estimated using
the pavement model
Value
--
Item
154
Based on the results shown in Table A-3, the total uncertainty in the cumulative
heat rejection is 1.45% for the pond with the horizontal slinky, 2.30% for the pond with
the vertical slinky, and 1.00% for the concrete slab. These errors are considered
acceptable.
Summary
In summary, the experimentally-determined cumulative heat rejection and the
model-predicted cumulative heat rejection for the test period are compared in Table A-4.
TABLE A-4
Summary of Experimental and Model Cumulative Heat Rejected
Two conclusions may be drawn from the above comparison: (1) the model
predictions match the experimental results within the bands of estimated uncertainty and
(2) the model predictions match the experiment better than would be expected by the
uncertainty analysis. The second conclusion implies that the uncertainty prediction may
be over-conservative.
Test Configuration Experimental Model
Cumulative Heat Rejected Cumulative Heat Rejected
(kW-hr) (kW-hr)
Pond with horizontal slinky 1532 +/- 22.2 1486 +/- 91.1
Pond with vertical slinky 1591 +/- 36.6 1508 +/- 95.8
Slab with slinky 1321 +/- 13.2 1255 +/- 131.4
155
APPENDIX A - REFERENCES
ASHRAE, 1997. ASHRAE Handbook, Fundamantals. American Society of Heating,
Refrigeration and Air-Conditioning Engineers, Inc., Atlanta, GA.
Elliot, R.L., F.V. Brock, M.L. Stone, and S.L. Sharp. 1994. Configuration Decisions for
an Automated Weather Station Network. Applied Engineering in Agriculture,
10(1): 45-51.
Incropera, F.P. and D.P. DeWitt, 1996. Introduction to Heat Transfer, 3
rd
Edition. John
Wiley & Sons.
Mills, A.F., 1995. Basic Heat and Mass Transfer. Prentice Hall.
Spitler, J. D., C. Marshall, R. Delahoussaye, and M. Manicham. 1996. Users Guide of
GLHEPRO, School of Mechanical and Aerospace Engineering, Oklahoma State
University, Stillwater, OK.
Spitler, J.D., D.E. Fisher, and D.C. Zietlow. 1989. A Primer on the Use of Influence
Coefficients in Building Simulation. Proceedings of Building Simulation ’89,
Vancouver, BC.
Tinker, J.A. and J.G. Cabrera. 1992. Modeling the Thermal Conductivity of Concrete
Based on Its Measured Density and Porosity. Proceedings of the
ASHRAE/DOE/BTECC Conference, Clearwater Beach, FL. American Society of
Heating, Refrigerating and Air-Conditioning Engineers, Inc., Atlanta, GA.

ADVANCES IN MODELING OF GROUND-SOURCE HEAT PUMP SYSTEMS

Thesis Approved:

Thesis Adviser

Dean of the Graduate College

ii

ACKNOWLEDGEMENTS

This thesis would not have been possible without the contributions of several people. To all of these people, I wish to express my gratitude.

First and foremost, I wish to thank my thesis adviser, Dr. Jeffrey Spitler. His expertise and respected reputation in the thermal sciences field allowed for the funding of many projects and the opportunity for me to gain knowledge and experience in several geothermal topics.

I also wish to thank to Dr. Simon Rees who served as an unofficial co-adviser along the course of this thesis. His expertise in computational methods was quite valuable in helping me to achieve the project goals.

Thanks also to Dr. Marvin Smith who provided input and experimental data regarding the supplemental heat rejecter studies

The review and contribution to this thesis by the committee members, Dr. R. Delahoussaye and Dr. A. Ghajar are appreciated.

Last, but not least, I thank my colleagues in B-10, both here and gone, namely Cenk Yavuzturk and Nagendra Jain, for their ideas and suggestions along the way.

iii

................................ 39 2........ 20 2.......... 26 2.................... 45 2.. 37 2............. 5 1........TABLE OF CONTENTS Chapter Page 1...... Conduction versus Advection in Geologic Materials ......................4...................5...........2... Introduction ........ The Numerical Model .. 12 1................................................2..........3........................................................................... 17 2.......2.................................................. Numerical Ground-Water Flow and Heat Transport Models ........................... Simulated In-Situ Thermal Conductivity Tests ....3.1......................... 30 2........... A Preliminary Assessment of the Effects of Ground-Water Flow on Closed-Loop Ground-Source Heat Pump Systems .............3............ 6 1................................ 43 2........ 1 1........1............................................4.. Validation of the Numerical Model .......... 33 2.......................................2................. Concluding Remarks and Recommendations for Future Work .... 24 2..................2.. Ground-Water Heat Pump Systems ...............................................2.........................4......................3...2... Horizontal Ground-Coupled Heat Pump Systems...................2....3.. Single-Borehole Simulations ............... Overview of Ground-Source Heat Pump Systems .. 20 2....... Thesis Objectives and Scope......... 27 2... 24 2.................. 4 1.....................1....... Borehole Field Simulations ............ 60 iv .........................................1................1................................1...........2...4.............................................................. 43 2...2........ 37 2......... 57 3.........1............................................4.................................2........2........... Boundary Conditions.............3...................5.........4......1.........3....................................3.......................... Heat Transport in Ground Water ..................2..........1........ Ground-Coupled Heat Pump Systems..........2.............................................................. Introduction............. 15 1..... Surface-Water Heat Pump Systems ............ Vertical Ground-Coupled Heat Pump Systems....3.................. 38 2.................................................4......1................................... Ground-Water Flow ... The Finite Element Mesh .................. 10 1.......1............................................... The Overall Modeling Approach................................. 13 1........... A Model for Simulating the Performance of a Shallow Pond as a Supplemental Heat Rejecter with Closed-Loop Ground-Source Heat Pump Systems ............ Typical Hydraulic and Thermal Properties of Soils and Rocks .. Coupled Ground-Water Flow and Heat Transport ..... 1 1... Results and Discussion............ 41 2........................................................... 49 2........... Model Description.....................1....................3.. Standing Column Well Systems .1........................

. 101 4.............. 97 4............................3.................. 92 3............ Test Slab Description and Data Collection..................................2........4......... Heat Transfer In Pavement Slabs ....3................3. Heat Transfer Due to Evaporation....4.........4... 77 3.....................................................1................................................................................... 116 v ............... Convection Heat Transfer at the Pond Surface .............4.....................................3......4..Chapter Page 3.................. 101 4................................. Parking Lot Test Section..................2.... 100 4............5..1...........................................................2. 66 3.....1.............. Weather Instrumentation and Data Collection.................. Governing Equations...8.....3.....2............................... 68 3.................. 72 3............. 110 4..1.3..................7.... 99 4.. Model Development . 66 3.................4..4......................2........... 99 4................... 84 3...................................................4.4.......................... Model Comparison to Experimental Results With Heat Rejection ............................... 87 3........................................... 79 3.................... 101 4......... Pond Description and Data Collection ..................4...................... Weather Instrumentation and Data Collection.. 76 3........ Boundary Conditions...2......................... Model Development ............................... Heat Transfer Due to Ground Water Seepage .4.................................................. 89 3............ Thermal Radiation Heat Transfer.............................5..............4.........1...... Bridge Deck Test Section ..................3.............. Computer Implementation... Heat Transfer Due to the Heat Exchange Fluid ................................1.... 106 4...3.........4........1............................................ Experimental Methods ........ 67 3.... 105 4......4.3...............................................2.............6......1..1.........1......................................................................................1....... Model Application .............................2..........5...... 111 4.. The Finite Difference Grid ................. 69 3.........................................................3..4... Heat Transfer to the Ground .......2....4........ 61 3.........6.....................................5................................ Results and Discussion..............................1.......5......... Heat Transfer In Ponds ........ Heat Transfer Due to the Heat Exchange Fluid ............3.3....... Governing Equations... 87 3.......................3.......................................4....................2...................1.................................................3............. Existing Pond and Lake Models ..4..1.........4......5....1................ 75 3..... 113 4.........1.................... 108 4........................ 111 4....4.......... Solving the Overall Energy Balance Equation......... 61 3. Introduction ..........4... Heat Flux Due to Rain and Snow.... A Model for Simulating the Performance of a Pavement Heating System as a Supplemental Heat Rejecter with Closed-Loop Ground-Source Heat Pump Systems....3................. 60 3...... Experimental Methods ..............1.....................4...... Introduction ...... 104 4.............. General Overview ........................ 62 3.........................3.................. 68 3.......... 108 4............... 71 3.....2..................... Solar Radiation Heat Gain ... Thermal Radiation Heat Flux........................1..... Convection Heat Flux at the Pavement Surfaces ... 83 3...................1..................... Concluding Remarks and Recommendations for Future Work .................3..................2............................. Model Comparison to Experimental Results With No Heat Rejection...... Solar Radiation Heat Flux.......... 106 4...........1..........1..................2........

.................................. 133 References............................2................. 121 4.4....... Summary and Conclusions.. Model Application .5.......5.........3......................................... 131 5. 121 4... 121 4.......................... 140 APPENDIX A – Model and Experimental Uncertainty Analysis.. Concluding Remarks and Recommendations for Future Work .. Computer Implementation.....................................................1....................................... 145 vi ............................5.... Results and Discussion....................................................Chapter Page 4....6...........4.................... 126 4.......................... 118 4........ Model Comparison to Experimental Results With Heat Rejection ... Model Comparison to Experimental Results With No Heat Rejection...............................................................5...................

.... Peclet Numbers Corresponding to Typical Values of Hydraulic and Thermal Properties of Soils and Rocks ............ 50 3-1......... Summary of Pavement Heating System Performance for Example GSHP System Simulation ............................. 96 4-1.. Typical Values of Hydraulic and Thermal Properties of Soils and Rocks.......... Summary of Borehole Field Design Parameters for Each Test Case.. Case Summary of Simulated In-Situ Ground Thermal Conductivity Tests ............................... 47 2-4. 131 vii .......................................LIST OF TABLES Table Page 2-1.................... 29 2-2..... Summary of Pond Performance for Example GSHP System Simulation ............. 32 2-3.........

..................... ............. 13 1-5... 49 2-7............................... time for coarse sand................... Average borehole fluid temperatures for the 12 simulated in-situ ground thermal conductivity test cases............................. 44 2-5........... .. fine sand............................... and shale.................................................................................... ......... A schematic of a surface-water heat pump system..... ....... 48 2-6....... 52 viii ...................... 7 1-4.............. Finite element mesh representing a 16 borehole field........ 39 2-4............................ ......................................................... Hourly ground loads for the test building............................................. 21 2-2.................. A schematic of a ground-water heat pump system........... A schematic of a vertical borehole and a horizontal ground-coupled heat pump system. 38 2-3... Schematic of cycles in a GSHP system in cooling and heating modes................................................ Schematic of typical apparatus for in-situ ground thermal conductivity testing........ 51 2-8............................................... Predicted effective ground thermal conductivity values versus ground-water flow velocity for the simulated in-situ thermal conductivity tests................... 2 1-2..........LIST OF FIGURES Figure Page 1-1. A schematic of a standing column well system................................................... 14 2-1........ Finite element mesh representing a single borehole................................................... Design borehole depths versus ground-water flow velocity for the simulated in-situ thermal conductivity tests............................. Average borehole fluid temperature vs............... 5 1-3............

......................................... ..................... . 6000 ft2 pond.......... 102 4-2............................. 54 2-10.............................Figure Page 2-9................. 63 3-2............ 107 4-5................. ................. Building thermal loads for the example building in Tulsa....................................................................................... Heat transfer mechanisms in shallow ponds................................................................ Pond model computer algorithm.... Entering heat pump water temperatures for the example GSHP system simulation with no pond and with a 2-feet deep...................... ........ 95 4-1................ 109 ix .................................... Comparison of observed and simulated average pond temperatures with no heat rejection for the (a) 2-feet deep pond and (b) 3................................... 2............................................ 85 3-3.................................. Comparison of observed and simulated heat rejected to the (a) 2-feet deep pond and (b) 3. ....................... and 10... .................... Pond model component configuration....... System schematic for the example model of a GSHP system with a shallow pond supplemental heat rejecter.. Heat transfer mechanisms in hydronically-heated pavement slabs with (a) no bottom exposure to the atmosphere and (b) bottom exposure to the atmosphere................................ OK.... ....... Temperature contours for Case 8 for the end of September of years 1. Comparison of observed and simulated heat exchange fluid return temperatures for the (a) 2-feet deep pond and (b) 3......................5 feet deep pond.................... 5... 56 3-1................... 92 3-8..... ............................. 90 3-6........ 88 3-5... Annual maximum and minimum average borehole fluid temperatures for the 16 borehole field simulations................................... 91 3-7........... Serpentine pipe configuration in a hydronically-heated pavement slab in (a) plan view and (b) cross-sectional view.......5 feet deep pond................... 102 4-3.. Slinky pipe configuration in a hydronically-heated pavement slab in (a) plan view and (b) cross-sectional view................ 103 4-4........................5 feet deep pond..... 86 3-4..... 94 3-9..... Finite difference cell geometry and notation.................... Model domain showing the finite-difference grid and boundary conditions.............. ...................................................................................

Pavement heating model component configuration............. 127 4-12.............. 125 4-11........... 119 4-7....... .. 120 4-8......... Entering heat pump water temperatures for the example GSHP system simulation with no pavement heating and with a 24.... 1996 and (b) March 24-23.. Comparison of observed and simulated slab surface temperatures and heat exchange fluid return temperatures for the bridge deck test section for the nights of (a) March 7-8..000 ft2 (2230 m2) parking lot with pavement heating.....Figure Page 4-6.... 124 4-10............................... 129 x ........................... Comparison of observed and simulated slab surface temperatures for the parking lot test section with no heat rejection to the slab.......... .... ............................................. 122 4-9...... System schematic for the example model of a GSHP system with a pavement heating system supplemental heat rejecter. 1996......... ........................... Comparison of observed and simulated (a) heat exchange fluid return temperatures and (b) heat rejected to the parking lot test section.... .... Pavement heating model computer algorithm......... ...................... ....

ft2-oF/hr-Btu (m2-oC/W) acceleration due to gravity .67 x 10-8 W/m2-K4) transmittance of solar radiation (-) area .577215664901) s dynamic viscosity .ft (m) borehole depth .ft2/s (m2/s) effective thermal diffusivity .R-1 (K-1) emissivity coefficient (-) Euler’ constant (0.ft2/s (m2/s) fouling factor . distance between nodes or thickness – ft (m) mass flow rate– lb m/hr (kg/s) mass flux – lbm/hr-ft2 (kg/s-m2) xi .Btu/lb m-oF (J/kg-oC) depth .Btu/hr-ft2-oF (W/m2-oC).ft (m). solar absorptivity (-) coefficient of thermal expansion .Btu/hr-ft2 (W/m2) hydraulic conductivity .lb/ft3 (kg/m3) reflectance of pond surface (-) Stephan-Boltzmann constant 0. hydraulic head .NOMENCLATURE Symbols α β ε γ µ µ’ ν θ ρ ρ’ σ τ A c.Btu/hr-ft-oF (W/m-oC) characteristic length .cp d D D* ff g h H i I K k L l & m & m" = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = thermal diffusivity .ft/ft (m/m) solar radiation flux on horizontal .ft (m) diffusion coefficient .lb m/ft-hr (Pa-s) extinction coefficient .ft (m) hydraulic gradient .ft/s2 (m/s2) heat or mass transfer coefficient .ft2/hr (m2/s).ft2 (m2) specific heat capacity .1714 x 10-8 Btu/hr-ft2-oR4 (5.ft/s (m/s) thermal conductivity .ft (m) length term described by Eskilson (1987) .ft2/hr (m2/s) angle of incidence of sun’ rays (radians) s density .ft-1 (m-1) kinematic viscosity .

ft (m).lb m water/lb m dry air (kg water/kg dry air) rectangular coordinates time step .s undisturbed mean temperature of the ground .oF (oC) overall heat transfer coefficient .z ∆t Subscripts ⊥ || a AB b = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = Lewis Number (-) porosity (-) quantity (-) Nusselt Number (-) perimeter .ft2-oF/hr-Btu (m2-oC/W) ground-water recharge -s-1 Rayleigh Number (-) Reynolds Number (-) specific storage coefficient .ft3 (m3) average linear groundwater velocity .Btu/hr (W] heat flux . specific discharge .Btu/hr-ft2-oF (W/m2-oC) volume . thermal resistance . unpolarized solar radiation component (-) retardation coefficient (-).Btu/hr (W).Le n N Nu P Pe Pr q q* ′ q′ Q Q* r R R* Ra Re Ss T t Tom U V v V* w x.ft-1 (m -1) temperature .ft/s (m/s) volumetric flow rate .Btu/hr-ft2 (W/m2) volumetric flow rate .ft3/s (m3/s) heat source/sink term .oF (oC) time . pressure .oF/s (oC/s) radius .ft/s (m/s) velocity . borehole convection flow circuit or spool diffuse radiation. xii .ft (m).s c circuit d = = = = = = = = = perpendicular component parallel component absorbed component of solar radiation transfer from material A (water) to material B (air) beam radiation.ft/s (m/s) ground thermal load .ft3/s (m3/s) humidity ratio .atmospheres Peclet number (-) Prandtl Number (-) heat transfer rate .

z (x.db dp eff fg fluid i i. = background or initial = outlet = thermal radiation. = refraction = solid phase = surface = steady-state.b t t-∆t w wb x.n o out r s surf sw. borehole wall = current time step.1) = diffusion = dry bulb = dew point = effective = latent heat of vaporization = heat exchange fluid = nodal index. = injected/extracted water = wet bulb = coordinate indices = coordinate index for a surface node xiii .j if in l m. = total = previous time step = water. = pipe inside = coordinate indices = latent heat of fusion = inlet = liquid phase = coordinate locations = pipe inside.

Overview of Ground-Source Heat Pump Systems Ground-source heat pump (GSHP) systems (also referred to as geothermal heat pump systems. These categories are: (1) high-temperature (>302 oF (>150 oC) ) electric power production.and low-temperature (<302 oF (<150 oC)) direct-use applications. Introduction 1. Thermodynamically. or surface water as a heat source and/or sink. both systems absorb heat at a low temperature level and reject it to a higher temperature level. The first loop is on the load side and is either an air/water loop or a water/water loop. The difference between the two systems is that a refrigeration application is only concerned with the low temperature effect produced at the evaporator. and GeoExchange systems) have received considerable attention in the recent decades as an alternative energy source for residential and commercial space heating and cooling applications. The term “ground-source heat pump” has become an all-inclusive term to describe a heat pump system that uses the earth. depending on the application.1 1. ground water. GSHP applications are one of three categories of geothermal energy resources as defined by ASHRAE (1995). earth energy systems. GSHP systems consist of three loops or cycles as shown in Figure 11. and (3) GSHP applications (generally <90 oF (<32 oC)). there is no difference between the well-known vapor-compression refrigeration cycle and the heat pump cycle.1. The GSHP applications are distinguished from the others by the fact that they operate at relatively low temperatures. The second loop is the refrigerant loop inside a watersource heat pump. while a heat pump may be concerned with . (2) intermediate.

a reversing valve is used to switch between heating and cooling modes by reversing the refrigerant flow direction. Return from Cooling Load Heat Pump “Load-Side” Cycle To Cooling Load Heat Pump Return from Heating Load “Load-Side” Cycle To Heating Load Water-Refrigerant or Air-Refrigerant Heat Exchanger Expansion Device Compressor Expansion Device Water-Refrigerant or Air-Refrigerant Heat Exchanger Compressor Refrigerant Cycle Reversing Valve Refrigerant Cycle Reversing Valve Water-Refrigerant Heat Exchanger Water-Refrigerant Heat Exchanger Warmer Fluid Return Ground-Source Heat Exchanger Cycle Cooler Fluid Supply Cooler Fluid Return Ground-Source Heat Exchanger Cycle Warmer Fluid Supply (a) (b) Figure 1-1.2 both the cooling effect produced at the evaporator as well as the heating effect produced at the condenser. In these dual-mode GSHP systems. The third loop in the system is the ground loop in which water or an antifreeze solution exchanges heat with the refrigerant and the earth. A higher COP (coefficient of performance) can be achieved by a GSHP because the source/sink earth temperature is relatively constant compared to air . Efficiencies of GSHP systems are much greater than conventional air-source heat pump systems. Schematic of cycles in a GSHP system in (a) cooling mode and (b) heating mode.

under normal geothermal gradients of about 0. . which is a more desirable heat transfer medium because of its relatively high heat capacity. the temperature of this interval within the earth is approximately equal to the average annual air temperature (Hart and Couvillion. 1986). the earth temperature begins to rise according to the natural geothermal gradient.3 temperatures. (3) surface water heat pump (SWHP) systems. A fourth category is added here for the sake of completeness.. 1986). Above this zone (less than about 20 feet (6. the earth temperature is a damped version of the air temperature at the earth’ surface. 1982). As a result. ASHRAE (1995) groups GSHP systems into three categories based on the heat source/sink used.7 m) deep (Hart and Couvillion. the earth temperature is roughly constant in a zone extending from about 20 ft (6. These categories are: (1) ground-water heat pump (GWHP) systems. heat is absorbed and rejected through water.1 m) deep). and (4) standing column well (SCW) systems. (2) ground-coupled heat pump (GCHP) systems.5 oF/100 ft (30 oC/km) (Grant et al. This constant temperature interval within the earth is the result of a complex interaction of heat fluxes from above (the sun and the atmosphere) and from below (the earth interior).1 m) deep to about 150 ft (45.7 m) deep). GSHP systems rely on the fact that. Additionally. Each of these is discussed in the following subsections. Below this zone (greater than about s 150 ft (45.

Design considerations for GWHP systems are fairly-well established. and small amount of ground area required relative to other GSHP and conventional systems.1. The main advantage of GWHP systems is their low cost. such as back to an aquifer. many legal issues have arisen over ground water withdrawal and re-injection in some localities. or to a sewer. conventional water wells and well pumps are used to supply ground water to a heat pump or directly to some application. simplicity. Disadvantages include limited availability and poor chemical quality of ground water in some regions. 1997). also referred to as open-loop systems. ground-water chemical quality. Corrosion protection of the heat pump may be necessary if ground water chemical quality is poor. . are the original type of GSHP system. In GWHP systems.1. Ground-Water Heat Pump Systems GWHP systems. to a surface-water body. well-drilling technologies and well-testing methods have been well-known for decades. and ground-water disposal method. A schematic of a GWHP system is shown in Figure 1-2.4 1. With growing environmental concerns over recent decades. Design considerations include: ground-water availability. to the unsaturated zone (as shown in Figure 1-2). The “used” ground water is typically discharged to a suitable receptor. The first GWHP systems were installed in the late 1940s (Kavanaugh and Rafferty. GWHP systems are not the focus of this thesis. so they will only be briefly described here.

were pioneered in the 1970s. Their main advantage over their water-well predecessors is that they eliminate the problems associated with ground water quality and availability and they generally require much less pumping energy than water well systems because there is less elevation head to overcome.2. In GCHP systems. also referred to as closed-loop ground-source heat pump systems. Ground-Coupled Heat Pump Systems GCHP systems.1. A schematic of a ground-water heat pump system. This fluid is either pure water or an antifreeze solution and is typically circulated through high-density polyethylene . GCHP systems can be installed at any location where drilling or earth trenching is feasible. 1. heat rejection/extraction is accomplished by circulating a heat exchange fluid through a piping system buried in the earth.5 Figure 1-2.

6

(HDPE) pipe installed in vertical boreholes or horizontal trenches as shown in Figure 1-3. Thus, these systems are further subdivided into vertical GCHP systems and horizontal GCHP systems.

1.1.2.1. Vertical Ground-Coupled Heat Pump Systems

Vertical borehole GCHP systems are the primary focus of this entire thesis. Therefore, they are described in some detail here and their design challenges are explained, laying the foundation for the motivation of this study.

In vertical borehole GCHP systems, ground heat exchanger configurations typically consist of one to tens of boreholes each containing a U-shaped pipe through which the heat exchange fluid is circulated. Some Swedish applications use boreholes inclined from the vertical. A number of borehole to borehole plumbing arrangements are possible. Typical U-tubes have a diameter in the range of ¾ in. (19 mm) to 1 ½ in. (38 mm) and each borehole is typically 100 ft (30.5 m) to 300 ft (91.4 m) deep with a diameter ranging from 3 in. (76 mm) to 5 in. (127 mm). The borehole annulus is generally backfilled with a material that prevents contamination of ground water.

7

(a)

(b) Figure 1-3. A schematic of (a) a vertical borehole ground-coupled heat pump system and (b) horizontal ground-coupled heat pump system.

8

The design of vertical ground heat exchangers is complicated by the variety of geological formations and properties that affect their thermal performance (ASHRAE, 1995). Proper subsurface characterization is not economically feasible for every project. One of the fundamental tasks in the design of a reliable GCHP system is properly sizing the ground-coupled heat exchanger length (i.e. depth of boreholes). Particularly for large systems, an extensive effort is made to design the ground loop heat exchangers so that they are not too large (resulting in too high of a first cost) or too small (resulting in the building’ thermal load not being met). s

In the early days of GCHP technology, the task of sizing the ground-loop heat exchanger was accomplished using rules of thumb (i.e. 250 feet of bore length per ton of heating or cooling capacity). These rules were slightly modified on a case-by-case basis using some estimates of thermal conductivity of the formation or using previous design experience, but additional costs of more detailed testing or calculations was judged to outweigh the costs of a conservative design. This relatively simple approach proved to be successful in most residential and other small applications, but in larger-scale commercial and institutional applications, some ground-loop heat exchangers failed to meet their design loads after the first few years of operation. Further, the practice of greatly over-designing large GCHP systems was found to be unacceptable because the first costs were simply not competitive with the first costs of conventional systems. Consequently, intensive research regarding methods to optimize ground-loop heat exchanger design has been ongoing for the last decade.

are generally coolingdominated and therefore reject more heat to the ground than they extract over the annual cycle. As a result. these effects have not been well understood. Another challenge in the design of GCHP systems arises from the fact that most commercial and institutional buildings. This has given rise to the concept of “supplemental heat rejecters” or so-called “hybrid GSHP systems”.9 Simple approaches to sizing the ground-loop heat exchanger in larger-scale applications are inadequate mainly because the heat transfer processes in the ground are complicated by thermally interacting boreholes and hourly periodic heat extraction/injection pulses. the GSHP system may be eliminated from consideration early in the design phase of the project due to excessive first cost. ground-loop heat exchanger designers need to consider hourly heating and cooling loads of the building and need to perform some simulation of the ground-loop temperatures over the life-cycle of the building. As a result. . This load imbalance may require the heat exchanger length to be significantly greater than the length required if the annual loads were balanced. This is the topic of Chapter 2 of this thesis. However. none of the programs consider the effects of ground water flow on ground-loop heat exchanger performance. perhaps because of the lack of relevant investigations. Annual heat rejection and heat extraction are usually not equal and peak temperatures rise or fall over a number of years. Recent research efforts have produced several methods and computer software programs for this purpose. even in moderate climates.

These are the topics of Chapters 3 (shallow ponds) and Chapter 4 (pavement heating systems) of this thesis. Therefore.2.91 m) to 6 ft (1. A number of piping arrangements are possible. 1. Design of these supplemental components adds to the challenge of designing the overall hybrid GCHP system because of their highly transient nature. In some applications. and pavement heating systems.2. conventional cooling towers or shallow ponds can provide a cost-effective means to reduce heat exchanger length. Horizontal Ground-Coupled Heat Pump Systems In horizontal GCHP systems. car washes. “Slinky” configurations (as shown in Figure 1-3 (b)) are popular and simple to install in trenches and shallow excavations. (19 mm) . Heat rejection systems are likely to operate more often during the night-time hours or when the building is not in use.83 m) deep. it is essential that the hourly (or less) behavior of these systems be examined during their design phase.1. In horizontal boreholes. ground heat exchanger configurations typically consist of a series of parallel pipe arrangements laid out in dug trenches or horizontal boreholes about 3 ft (0. In cases where the excess heat cannot be used beneficially.10 Supplemental heat rejecters have been integrated into building designs to effectively balance the ground loads and therefore reduce the necessary length of the ground-loop heat exchanger. Typical pipes have a diameter in the range of ¾ in. straight pipe configurations are installed. the excess heat that would otherwise build up in the ground may be used for domestic hot water heaters.

the use of a shallow pavement heating system) is the focus of Chapter 4 of this thesis.e. (38 mm) and about 400 ft (121. The main difference is that horizontal ground-loop heat exchangers are more affected by weather and air temperature fluctuations because of their proximity to the earth’ surface. geologic site characterization is relatively easier because soils can be readily seen and sampled. Also. This may be due to the fact that vertical systems tend to be preferred in larger applications since much less ground area is required. horizontal systems have received much less attention than vertical systems with respect to recent research efforts. Further.9 m) of pipe is installed per ton of heating and cooling capacity.11 to 1 ½ in. This may result in larger loop temperature fluctuations and therefore s lower heat pump efficiencies. A specific application (i. over-conservative designs are not as cost prohibitive as with vertical borehole designs because of the relatively low installation costs of the heat exchanger pipe. . since horizontal systems are installed at shallow depths. Aside from the invention of the Slinky coil itself and the use of these systems as supplemental heat rejecters. Recent research activities have focussed on using these systems as supplemental heat rejecters with vertical borehole GCHP systems. The thermal characteristics of horizontal GCHP systems are similar to those of vertical ones.9 m) to 600 ft (182.

Heat transfer mechanisms and the thermal characteristics of surface-water bodies are quite different than those of soils and rocks. A schematic of a SWHP system is shown in Figure 1-4. 1997).3. Surface-Water Heat Pump Systems The third category of GSHP systems is the surface-water heat pump (SWHP) system. depending on the climate. many successful installations are currently in operation and some guidelines do exist. (38 mm) and a length of 100 feet (30. the use of a shallow pond as a supplemental heat rejecter in vertical GCHP systems) is the focus of Chapter 3 of this thesis. water is extracted from the surface-water body through a screened intake area at an adequate depth and is discharged to a suitable receptor. pond. or other suitable open channel.44 m) per ton of heating or cooling capacity is recommended by ASHRAE (1995b). (19 mm) to 1 ½ in. At the present time. However. reservoir. The surface-water heat exchanger can be a closed-loop or open-loop type. design tools for surface-water heat pump systems are in their developmental infancy (Kavanaugh and Rafferty. Typical closed-loop configurations are the Slinky coil type (as shown in Figure 1-4) or the loose bundle coil type. In .48 m) to 300 feet (91. This subject will be further discussed in Chapter 3 of this thesis.12 1. In open-loop systems. A specific application of SWHP systems (i. heat rejection/extraction is accomplished by circulating a heat exchange fluid through HDPE pipe positioned at an adequate depth within a lake. In the closed-loop systems.e. Typical pipe diameters range from ¾ in.1.

and locating the coil at a proper depth in a water body with adequate thermal capacity.1.4. but are recently receiving much attention. .13 short. specifying a sufficient numbers of parallel loops. 1. the loop design involves selection of sufficient length of coil for heat transfer. specifying adequate diameter piping. These systems are about as old as the ground-water heat pump systems. A schematic of a surface-water heat pump system. they are only briefly discussed here. Since these are not the subject of this thesis. Figure 1-4. Standing Column Well Systems The fourth category of GSHP systems is known as a standing column well (SCW) system.

Properly sited and designed. primarily installed in hard rock areas. These systems.2 m). (15. use uncased boreholes with typical diameters of about 6 in. Figure 1-5.24 cm) and depths up to 1500 feet (457. SCW systems have been shown to have significant installation cost savings over closed-loop GCHP systems. This type of GSHP draws water to a heat pump from a standing column of water in a deep well bore and returns the water to the same well.14 A schematic of an SCW system is shown in Figure 1-5. The uncased borehole allows the heat exchange fluid to be in direct contact with the earth (unlike closed-loop heat exchangers) and allows ground water infiltration over the entire length of the borehole. Design guidelines for SCW systems are currently under development. A schematic of a standing column well system. .

and collecting and analyzing data at the site over a time period of years. Thesis Objectives and Scope This study deals with the modeling of vertical closed-loop and hybrid. instrumenting that site. The challenges associated with the design of these systems were discussed in the previous section. Given the time and cost constraints of finding a suitable site with significant ground water flow. Hydraulic and thermal properties of soils and rocks were compiled and AQUA3D. a commercially-available numerical . and (3) develop a design and simulation tool for modeling the performance of a hydronic pavement heating system as a supplemental heat rejecter with closed-loop GSHP systems. Chapter 2 of this thesis addresses the first objective. (2) develop a design and simulation tool for modeling the performance of a shallow pond as a supplemental heat rejecter with closed-loop GSHP systems. a computer modeling study was conducted as a preliminary assessment of the effects of ground-water flow on closed-loop GSHP systems. There are three primary objectives of this study. These are to: (1) examine the effects of ground-water flow on closed-loop GSHP systems.2. groundsource heat pump systems.15 1. A considerable amount of research in the past decade has been geared toward optimizing the performance of these types of systems and this study is part of those efforts.

16 ground-water flow and heat transport model developed by Vatnaskil Consulting Engineers. 1991). An example application of the pond model is also presented. The model has been developed in the TRNSYS modeling environment and can therefore be coupled to other GSHP system component models for short-time step (hourly or minutely) system analyses. This model has also been developed in the TRNSYS modeling environment . ducting. The development and validation of a design and simulation tool for modeling the performance of a shallow pond as a supplemental heat rejecter with closed-loop GSHP systems is presented. was used to simulate the impact of ground-water flow on the average heat exchange fluid temperature in single and multiple borehole systems. The development and validation of a design and simulation tool for modeling the performance of a hydronic pavement heating system as a supplemental heat rejecter with closed-loop GSHP systems is presented. Chapter 3 of this thesis addresses the second objective. The impact of ground-water flow on the estimation of soil/rock thermal conductivity from in-situ test data was also examined. Chapter 4 of this thesis addresses the third objective. By a simple language. is a widely-used modular transient thermal systems simulation program where each system component is described mathematically by a FORTRAN subroutine. the components are connected together in a manner analogous to piping. TRNSYS.. Iceland. Inc. and wiring in a physical system (Duffie and Beckman. developed at the University of WisconsinMadison. Reykjavik.

Mathematical models are solved analytically or numerically using manual or computer methods. (2) develop a conceptual model of the system. This study deals only with mathematical representations of systems. 1. (5) validate the model. (3) develop or define the mathematical model of the system. some definitions and approaches as applicable to this study are described here. Finally.3. Chapter 5 of this thesis summarizes the conclusions of the individual studies. The overall modeling approach consists of six stages: (1) define the purpose of the model. Since the term “model” can be used loosely. The Overall Modeling Approach Each of the objectives of this thesis deals with the development and/or application of a model. Each of these is described in the following paragraphs. A “model” is a physical or a mathematical representation of an actual system.17 for use in short-time step system analyses. American Society for Testing and Materials (ASTM) defines a mathematical model as “mathematical equations expressing the physical system behavior and including simplifying assumptions”. (4) implement the solution method. An example application of the pavement heating model is also presented. and (6) apply the model. .

18 The first stage in the modeling approach is to clearly define the purpose and objectives of model. The third stage in the modeling approach is to develop a mathematical model. . ASTM defines a conceptual model as “an interpretation or working description of the characteristics and dynamics of the physical system”. The solution method and limiting assumptions or simplifications are also identified. The second stage in the modeling approach is to develop a conceptual model of the system. This helps to determine the level of detail and accuracy desired by the model and helps in making decisions regarding the resources needed. The fourth stage in the modeling approach is to implement the solution method to solve the mathematical equations. In this stage. this stage involved using computer methods with a combination of commercially-available software and FORTRAN code developed specifically for this study. the conceptual model is translated to mathematical equations that can be solved for the desired unknowns. The purpose of the conceptual model is to describe the system by a set of assumptions and concepts that can be evaluated mathematically. With respect to this thesis.

The sixth stage in the modeling approach is to finally use the model to analyze the performance and behavior of actual thermal systems. to an analytical solution or to experimental data. With respect to this thesis. .19 The fifth stage in the modeling approach is to validate the model. this stage involved comparing model results. where applicable.

1996. yet adequately determining them is often the most difficult task in the design phase..e. 1996.1. Recent research efforts have produced several methods and commerciallyavailable design software tools for this purpose (Ingersoll. Current methods range from estimating values from published data to conducting laboratory experiments on soil/rock samples to conducting single-borehole. Introduction One of the fundamental tasks in the design of a reliable ground-coupled heat pump system is properly sizing the ground-coupled heat exchanger length (i. Kavanaugh. A schematic of the typical field . in-situ field tests. These parameters are perhaps the most critical to the system design. 1954. 2000). 1987. 1984. All of these design tools are based on principles of heat conduction and rely on some estimate of the ground thermal conductivity and volumetric specific heat. and Austin et al. depth of boreholes). A Preliminary Assessment of the Effects of Ground-Water Flow on Closed-Loop Ground-Source Heat Pump Systems 2. and Kavanaugh and Rafferty. 1997)..20 2. Methods of determining the thermal properties of the ground have also been the subject of considerable recent research (Eklof and Gehlin. 1991. thermal property values derived from in-situ field tests are most representative because the values are site-specific and a larger volume of material is evaluated under more realistic conditions than is possible in the laboratory. Spitler et al. IGSHPA. In general. Eskilson. The typical field procedure in in-situ tests is to measure the temperature response of a fluid flowing through a ground heat exchanger in a single borehole.

1882. 1985. and numerical solutions with parameter estimation .. These methods will not be discussed in detail here.. Determination of thermal conductivity from temperature-time data is an inverse problem. 1996. Ingersoll.21 apparatus is shown in Figure 2-1. numerical solutions (Mei and Emerson. Muraya et al. Several analytical and numerical methods exist for interpreting the data set. the “line-source” analytical solution (Kelvin. 1946). Portable Unit Water Heaters + Watt Transducer Temperature Sensors Pipe Insulation Flow Meter Ground Surface Water Circulating Pumps U-Tube Heat Exchanger Borehole Figure 2-1. 1997). but include the “cylinder-source” analytical solution (Carslaw and Jaeger. The fluid temperature at regular time intervals and heat added to the fluid stream are recorded over the course of the test. and Rottmayer et al. 1954). Schematic of typical apparatus for in-situ ground thermal conductivity testing.

Where ground water is present. The objectives of the work presented in this chapter have been to make a preliminary examination of the effects of ground water flow on both in-situ ground .22 (Shonder and Beck. flow will occur in response to hydraulic gradients and the physical process affecting heat transfer in the ground is inherently a coupled one of heat diffusion (conduction) and heat advection by moving ground water. 2000)..or under-designed. An unusually high thermal conductivity value was determined from in-situ test data at a site in Minnesota where significant ground-water flow was believed to occur (Remund. Complications in the borehole field design process due to the presence of flowing ground water arise from the fact that both current in-situ conductivity test data interpretation methods and ground-loop heat exchanger design methods are based on models that only consider heat conduction. Therefore. In general. Austin et al. 1999. s A further complication in the design of ground-coupled heat pump systems is the presence of ground water. ground-water flow can be expected to be beneficial to the thermal performance of closed-loop ground heat exchangers since it will have a moderating effect on borehole temperatures in both heating and cooling modes. Each of these methods is based on Fourier’ Law of heat conduction. ground-water flow may impact the design process in two ways: (1) thermal conductivities derived from in-situ tests may appear artificially high and (2) borehole fields designed from artificially high thermal conductivity values may be over. 1998).

This has been attempted firstly by examining the range of hydrogeological conditions that might be expected and estimating the order of magnitude of the corresponding ground-water flows.23 thermal conductivity measurements and long-term borehole field performance. were used to design a hypothetical multi-borehole field by employing conventional design tools and procedures. the derived thermal conductivities. For each test case. Conclusions are presented on the ability of conventional design procedures to correctly predict the long-term performance of closed-loop ground heat exchangers under different ground-water flow conditions. . along with the thermal loads from an actual building. a numerical model of the whole borehole field was used to simulate its longterm performance. A finite-element numerical ground-water flow and heat transport model has been used to simulate and observe the effects of ground-water flow on the average fluid temperature in a single U-tube borehole in various geologic materials. The model was used to simulate several in-situ ground thermal conductivity tests. and thermal conductivities were derived from these data using a standard approach. For different sets of hydrogeological conditions. A simple method of examining the importance of heat advection from groundwater flow is then presented.

In practice. Ground water is present nearly everywhere. group of formations. water in soil or rock pore spaces is at atmospheric pressure. In the unsaturated zone. Coupled Ground Water Flow and Heat Transport 2. Aquifers are described as being either confined or unconfined. Confined aquifers are bounded between two layers of lower permeablity materials. pores are fully saturated and water exists at pressures greater than atmospheric.1. In the saturated zone (below the water table).2. Unconfined aquifers are bounded at their upper surface by the water table.2. but it is only available in usable quantities in aquifers. or part of a formation that contains sufficient saturated permeable material to yield economical quantities of water to wells and springs. the boreholes of ground-loop heat exchangers may partially penetrate several geologic layers. In this paper. An “aquifer” is defined by Driscoll (1986) as a formation. The governing equation describing flow through porous media is Darcy’ Law: s . The surface separating the saturated zone from the unsaturated zone is known as the “water table”. we deal only with water in the saturated zone.24 2. Ground-Water Flow Underground water occurs in two zones: the unsaturated zone and the saturated zone. At the water table. The term “ground water” refers to water in the saturated zone. pores are only partially saturated and the water exists under tension at pressures less than atmospheric.

by: v= q n (2-2) where n is the porosity and is introduced to account for the difference between the unit cross-sectional area and the area of the pore spaces through which the ground water flows (Freeze and Cherry. an equation defining the hydraulic head distribution s can be derived. 1979. Fetter. 1988). Transient ground-water flow with constant density can then be expressed in Cartesian tensor notation as: Ss ∂ h ∂ ∂  h * K ij −  ∂ = R  ∂ ∂i  t x xi  (2-3) Since ground water at 110oF (43. The specific discharge is related to average linear ground water velocity. and h is the hydraulic head.3oC) (an extreme temperature limit expected in GSHP applications) has a specific gravity of approximately 0. the assumption of constant density flow for low-temperature geothermal applications may be considered valid. v.991. . By applying the law of conservation of mass to a control volume and by making use of Darcy’ Law (Equation 2-1). K is the hydraulic conductivity.25 q=− K dh dx (2-1) where q is the specific discharge (volume flow rate per unit of cross-sectional area).

Equation 2-4 reduces to a form of Fourier’ Law of heat s conduction. By applying the law of conservation of energy to a control volume.26 2. (2) heat transfer through the liquid phase by conduction. It is the second term in Equation 2-4 that represents advection of heat by the ground water and couples Equations 2-3 and 2-4 together.2. If the ground water velocity is zero.2. an equation for heat transport in ground water can be found and can be expressed as: nR ∂ T ∂ T ∂ + vi − ∂ t ∂i ∂i x x  ∂  T * D ij  ∂ = Q  xi   (2-4) where the velocity. Heat Transport in Ground Water Heat can be transported through a saturated porous medium by the following three processes: (1) heat transfer through the solid phase by conduction. The diffusion coefficient tensor Dij is modeled here as an effective thermal diffusivity given by: . and (3) heat transfer through the liquid phase by advection. vi is determined from the solution of Equation 2-3 and T is the temperature of rock/water matrix. The governing equation describing mass or heat transport in groundwater is a partial differential equation of the advection-dispersion type (Freeze and Cherry. 1979).

Heat transfer is dependent on the flow velocity and the thermal properties of the material.2. to consider the range of naturally-occurring soil and rock properties and possible values of hydraulic gradient. but heat is only advected by the water. the question arises as to what locations have significant ground-water flow. It is therefore useful. it is necessary to define a retardation coefficient R accounting for retardation of the thermal plume which results from differences in the liquid and solid volumetric heat capacities: R= 1 + (1 − n)c s ρs n clρl (2-7) 2.27 D* = k eff ρlcl (2-5) The effective thermal conductivity keff is a volume-weighted average thermal conductivity of the saturated rock matrix and can be expressed using the porosity as: k eff = nk l + (1 − n)k s (2-6) It is necessary to distinguish between the conductivity and thermal capacity of the water and soil/rock in this way to account for the fact that heat is stored and conducted through both the water and soil/rock. Similarly. Typical Hydraulic and Thermal Property Values for Soils and Rocks In assessing the significance of ground-water flow to closed loop heat exchanger performance. in making a preliminary assessment of the significance of ground-water flow.3. . Darcy’ Law indicates that flow is dependent on both the local hydraulic gradient and the s hydraulic conductivity of the geologic material.

Values of hydraulic gradient are somewhat more site-specific. Canada (Stephenson et al. 1988).002 to 0. Local pumping activities may further increase ground-water flow rates in aquifers. 361 ft/yr (110 m/yr) in the High Plains sand and gravel aquifer. 1988).6 x 10-3 m/yr) in glacial clay soils in Southern Ontario. Of these.0 x 10-4 m/yr) to 1. USA (Lindholm and Vaccaro.. porosity. porosity may be considered the most important property simply because of the origin and nature of soils and rocks.50 x 10-2 ft/yr (4.05.0001 to 0. 1988). Idaho. and degree of saturation. Some specific examples of natural ground water velocities include: 1796 ft/yr (547. the United States Environmental Protection Agency (1996) reports a typical range of hydraulic gradient values of 0. For saturated materials.3 x 10-3 ft/yr (4. increased porosity results in increased heat capacities and therefore lower thermal diffusivities. and 1. Lower porosities in rocks result in higher contact area between grains and therefore higher thermal conductivities than soils.28 Naturally-occurring ranges of values of hydraulic and thermal properties of soils and rocks are summarized in Table 2-1. The thermal properties of soils and rocks are functions of mineral content.5 m/yr) to 7185 ft/yr (2190 m/yr) under a hydraulic gradient of 0. Rocks originate under higher heat and pressure environments than soils and consequently generally possess lower porosities.012 in the Snake River Group basalt. regardless of mineral content. . western central USA (Weeks and Gutentag.

40) 0.80) 0.01 ft/ft (m/m).0E-10) 2.08E-01 (9.00E+06) - ------4.56E+06) 5.60) 3.83 (6. 3 *** thermal property data fromHellstrom (1991).50) 1.39 (2.10 (1.31 .3E-09 (1.20E+01 (5.34 .65 (4.8E-04 (3.2E-08) 4.0E-07) 3.6E-10 (2.20E+01 (5.0.50) - 1.65E-02) 2.4E-04 (7.40E+06) 4.40E+06) 2.0E-04) 1.40E-02) 4.0E-09) 9.17E+01 (2.85E+06) 4.15E+02) 2.94E+06) 3.7E-09) 9.1E-06 0 .70) 0.20E+06) Notes: Thermal conductivity values are taken to represent that of materials in the dry condition.00E+06) 8.10 0.0.50) 3.0E-02 (6.0E-06) 3.92E+01 (3.21E+01 (9.0E-09) 3.20 (7.13E+07) 3.025 - 9.0E-05 (6.09E-05) 0.80) 1.02 (3.54 (4.96 (3.0525 0.40E+06) 4. .0.28E+01 (2.05 .76E+02 (1.78E-04 (1.24 .3E-06 (1.39 (2.02E-01 (2.4E-12) 8.01E-04 (3.66E+01 (5.48 (4.30E+06) 5.50) 1.00E-03 2.33 (2.TABLE 2-1.28E+01 (2.40 (0.25E+01 (2.17E+02 (2.05 (2.F o (W/m.46 (0.6E-08 (8.6E-05 (2.31E+01 (3.84E-02 3.0.8E-10 (3.475 0.79E-04 (8.05E+00) 3.40) 1.87E+01 (3.0E-09) 3.47 (2.09E+01 (1.99E+02 (1.4E-07 0.61 0.00E-04 3.20E+01 (5.26 .0E-04) 6.17E+02 (2.0E-07) 6.60) 1.3E+06) Rocks Limestone.47 Range Arithmetic Average Btu/hr-ft.09E+01 (1. For sedimentary rocks.3E-11 (1.50 (1.50E-05) 3.50) 2.13E+07) 3.01E+01) 1.0E-03) 6.5E-07 0 .47E+01 (3.58E+01 (2.34E+07) 5.4E-07) 7.60E+06) 2.C) o Porous Medium Thermal Properties Thermal Conductivity *** Volumetric Heat Capacity *** (k) (ρs cs ) Btu/ft .20E+06) 3.40 0.60 Range (--) Soils Gravel Sand (coarse) Sand (fine) Silt Clay 9.40 (0.44 (2.0E-04) 6.0E-05 (6.84E-03 3.65 (4.0E-11) 0.44E-01) 3.6E-11 0 .3E-05) 2.5E-08 (4.51E-01 (7.05E+03 1.87 (1.0E-09) 9.50) 2.40E+06) 2.50) 1.6E-09 (2.C) Range Arithmetic Average 3 o ft/s (m/s) Range Geometric Average 0.40) 2.53 0.90) 0.80) 0.04 (1.47 (2.5E-06) 8.83 (6.98) ------3.52 (0.0.34 .0E-13) - 2.0.46 (0.05 .0E-10) 3.0E-12 0 .0.70) 0.3E-09 (1.1E-10 (2.52 (0.40) 1.50E+06) 7.78E+00) 1.09E+01 (1.30) 0. Typical Values of Hydraulic and Thermal Properties of Soils and Rocks Hydraulic Properties Hydraulic Conductivity* Porosity* (K) (n) Velocity** (v) ft/yr (m/yr) Arithmetic Average 0.37E+01 (3.13E+06) 3.30) 3.6E-07 (1.90) 0.10) 0. ** v is the average linear ground water velocity based on an assumed gradient of 0.46E+01 (5.0E-05) 1.90) 1.0E-02) 2.05 0.00E+04 3.0. Hellstrom lists only c In these cases.8E-14 (3.0E-06) 9.1E-05 (6.50E+06) ----- 1.44 (2. a density of 2500 kg/mis assumed.0.54E+01 (2.0E-06 (9.92E+01 (3. Dolomite Karst Limestone Sandstone Shale Fractured Igneous and Metamorphic Unfractured Igneous and Metamorphic 3.85) - 0.0.70) 0.84E-04 3.385 0.46 0.2E-10) 1.98E+02 (6.30 (4.6E-07 (2.30) 2.00E-02 2.50E+06) 8.3E-06) 4.275 0.56 (0.46 (0.38 0.76 (6.6E-04 (2.31 0.80) 0.69 (1.50) 1.0E-13) 2. s.49 (0.91 (3.3E-04 0.46E-04) 0.99E+02 (1.18 0.10 (1.58) 3.64 (1.3E-02 (1.0E-06) 6.38E+06) ----- - 8.87 (1.7E-08) 3.52 (0.0.4E-11) 5.58) 2.20) 0. * hydraulic conductivity and porosity data from Domenico and Schwartz (1990).3E-13 (1.40 (0.34E+07) 1.F 3 o (J/m .

e. However.2. A dimensionless parameter describing conduction versus convection is the Peclet number (Pe).e.4. According to Bear (1972). Some assessment of the significance of the flow can be made by considering the order of magnitude of the advection of heat compared to conduction (diffusion). Materials with higher porosity generally also have higher hydraulic conductivity. this correlation does not hold for fine-grained soils (see Table 2-1. Conduction Versus Convection in Geologic Materials It has already been noted that it is the presence of advection that distinguishes the heat transfer regime under ground-water flow conditions from that of heat conduction alone. so long as it is consistent with other comparisons. advection becomes significant when Pe . In principle. In this application. 2. in karst limestone) or to fracturing (i. Domenico and Schwartz (1990) define Pe for heat transport in ground water as: Pe = ρlcl q L k eff (2-8) The term L is defined as some characteristic length dependent on the situation. in rocks and cohesive soils). L can be chosen as any length dimension.). the Peclet number expresses the transport of heat by bulk fluid motion to the heat transported by conduction. Porosity and hydraulic conductivity of soils and rocks can be increased by so-called “secondary porosity” which is attributed to solution channels (i. 1979).30 The porosity of soils and rocks can also be an important controlling influence on hydraulic conductivity (Freeze and Cherry.

and kl as 62. when the characteristic length was chosen as mean grain size. cl. Above a Peclet number of about 5.4.8 ft (4. diffusion is the process controlling mass transport at Peclet numbers less than about 0. 1. In short.4 lb/ft3 (1000 kg/m3).347 Btu/hr-ft-oF (0.5 m) and assuming the fluid property values of ρl.60 W/m-oC). At Peclet numbers in the range of 0.4 to 5. mechanical dispersion (or advective dispersion) dominates. Many studies have been conducted and the data have been summarized by Bear (1972).31 is of order one. The Peclet number has been used to quantify the relative importance of mechanical (or advective) dispersion versus molecular diffusion in mass transport in ground water. An analysis of the Peclet number using the typical hydraulic and thermal values of soils and rocks presented in Table 2-1 may be used to assess the role of ground water flow in the design of closed-loop ground heat exchangers. The calculated Peclet numbers are listed in Table 2-2 using a typical borehole spacing of 14. No similar studies conducted for heat transport have been found. and 0. The exact value of Pe at which advection becomes significant is slightly dependent on the choice of L. The characteristic length could conceivably be chosen as (1) a typical borehole spacing or (2) the length of the borehole field in the direction of flow.0 Btu/lb-oF (4180 J/kg-oC). . a transition occurs where mechanical dispersion (or advective dispersion) and diffusion are of the same order of magnitude.

00E-07 .28E-02 3.05E-06 6.34E+01 1.32 A review of the data presented in Table 2-2 reveals that heat advection by ground water flow is significant process contributing to heat transfer in coarse-grained soils (sands and gravels) and in rocks exhibiting secondary porosities (fracturing and solution channels). that even when the Peclet number is of order one or higher. Dolomite Karst Limestone Sandstone Shale Fractured Igneous and Metamorphic Unfractured Igneous and Metamorphic 5. Peclet Numbers Corresponding to Typical Values of Hydraulic and Thermal Properties of Soils and Rocks Porous Medium Peclet Number where L = a typical borehole spacing of 14.77E-03 1.32E-02 1. When the characteristic length is defined as the borehole spacing.24E-05 5. Peclet numbers exceeding 1 exist only for sands.28E+00 1. TABLE 2-2. the effects of the ground-water flow on the temperature response may not be seen within the normal time scale of an in-situ thermal conductivity test.72E+02 1.92E-03 5. and karst limestones.15E+00 1. gravels.8 ft (4. It is possible however. This is one of the reasons for conducting numerical borehole field simulations for the duration of several years.5 m) [--] Soils Gravel Sand (coarse) Sand (fine) Silt Clay Rocks Limestone.

This code is also a three-dimensional. It was developed mainly for simulation of mass-transport problems but allows easy adaptation of boundary conditions to model heat transport without density-dependent ground-water flow. Numerical Ground-Water Flow and Heat Transport Models In assessing the effects of ground-water flow on U-tube heat exchanger performance. Numerous commercially-available and public domain numerical software codes exist for modeling mass and/or heat transport in ground water. • AQUA3D by Vatnaskil Consulting Engineers.2. Reykjavik. Density-dependent flow can also be simulated. Consequently. Therefore. . Iceland. The heat exchange fluid temperature is affected by the transient building thermal loads in addition to the heat transfer in the porous medium around the borehole. one is mainly interested in the temperature of the heat exchange fluid. this problem is characterized by an irregular model domain with time-varying boundary conditions and is best handled by a numerical model. modeling of the U-tubes in some detail is important in this problem. Pennsylvania State University. finite-element code. Of these. This code was developed to simulate mass transport in variably-saturated porous media.33 2. Yeh.5. the following 8 were selected for a more detailed review for potential application to this project: • 3DFEMFAT (3-Dimensional Finite Element Method Flow and Transport) by G.

have been recently added. mainly in the horizontal plane. Bethesda.. Germany. It was developed for simulating contaminanttransport problems in ground water. Denver.34 • FEFLOW (Finite Element FLOW) by WASY Institute for Water Resources Planning and Systems Research. This code is a two-dimensional finite difference code. contaminant-transport simulation capabilities. This code is also a three-dimensional finite-difference code. • MT3D 96 (Mass Transport in 3 Dimensions) by S. finite-element code. It solves the mass transport equation only and requires a solution to the ground-water flow equation from another code. Colorado. • HST3D (Heat and Solute Transport in 3 Dimensions) by USGS. It is capable of simulating both mass and heat transport in variable-density ground-water flow systems. Papadopulos & Associates. (WHI). Ltd. Inc. Maryland. It was developed mainly for simulating problems involving waste injection into aquifers. It is capable of simulating mass and heat transport in variable-density ground-water flow systems. Waterloo. Ontario. • Flowpath II by Waterloo Hydrologic. Inc. This code is also a threedimensional.. Berlin. It was developed originally for simulation of ground-water flow problems only. This code is a three-dimensional finite-difference code.S. .

Further. particular attention was paid to the following items: (1) the type of boundary conditions handled by the code. This code is a three-dimensional finite-difference code. variable-density ground-water flow systems. the selected code needed to be capable of handling timevarying. Sterling. • SWIFT (Sandia Waste Isolation Flow and Transport) by HSI GeoTrans. For simulation of periodic heat extraction or heat rejection to the ground. The type of boundary conditions handled by the code was perhaps the most important consideration in the code selection process. Colorado. It was developed mainly for simulating problems involving deep-well radioactive waste injection into geologic repositories. It is capable of simulating mass and heat transport in variable-density ground-water flow systems in porous or fractured media. (3) verification of the code. Each of these points is described in more detail below. . the selected code needed to be capable of reading time-varying conditions from an external file. This code is a two-dimensional finite-element code. It was mainly developed as a cross-sectional model for simulating salt-water intrusion into fresh-water aquifers. Denver. and (4) cost. Virginia. heat flux boundary conditions. since a relatively large number of timevarying data were to be used as input. It is capable of simulating mass or energy transport in variably-saturated. (2) the solution scheme employed by the code. In the code selection process.35 • SUTRA (Saturated-Unsaturated TRAnsport) by United States Geological Survey (USGS).

36

The software codes that were reviewed for this project employ either finite-difference methods (FDM) of finite-element methods (FEM) to solve the partial differential equations describing heat/mass transport in ground water (Equations 2-3 and 2-4). The solution scheme was considered important for two main reasons. First, FEM offers an advantage over FDM in the ability to represent complex or irregular geometries (i.e. circular U-tubes in a rectangular domain). Second, there has been controversy in the literature over advantages of FEM over FDM in solving the advection-dispersion equation (Equation 2-4). In general, experience has shown the FEM to be generally superior to FDM in solution stability (Wang and Anderson (1982) and Mercer and Faust (1986)). Consequently, codes employing an FEM solution scheme were preferred.

Documented verification of the code was an important consideration since it often requires years to find and fix bugs in these types of software programs. All of the 8 codes listed above have been originally developed in the 1980s and many validation examples exist.

The cost of the code was also a main consideration in the selection process since the project had an allocated budget for software.

37

2.3. The Numerical Model

2.3.1. Model Description

The computer code selected for this study was AQUA3D. It is a commerciallyavailable software package originally developed in 1983 by Vatnaskil Consulting Engineers of Reykjavik, Iceland. The partial differential Equations 2-3 and 2-4 are discretized spatially by a Galerkin finite element method using triangular elements with linear weighting functions (Vatnaskil, 1998). The temporal term of the equations is dealt with by first order backward differencing in time. AQUA3D does not allow for the explicit representation of the heat transport equation, but provides a general form of the mass transport equation (Equation 2-4). Temperatures were in fact calculated by suitable choice of the coefficients of the mass transport equation and corresponding adaptation of the boundary conditions.

The finite element ground-water flow and mass/heat transport model was used in this study as the primary means of assessing the effects of ground-water flow on closedloop heat exchangers. Use of a numerical model allows a wide range of conditions to be examined and is the only practical means of modeling a whole borehole field. In each test case, a uni-directional flow field was imposed over the whole numerical domain. As the flow was assumed to be fully-saturated and within homogeneous geologic material, it was only necessary to use a two-dimensional model.

38

2.3.2. The Finite Element Mesh

Finite element meshes for a single borehole geometry, and for a complete borehole field geometry have been constructed using triangular elements. Nodal spacing was kept relatively fine around the pipe walls where the steepest temperature gradients were expected. The mesh for the single borehole geometry was constructed within a square domain and consisted of 465 nodes, as shown in Figure 2-2.

14.4 ft (4.4 m)

3.46 in. (8.8 cm)

Figure 2-2. Finite element mesh representing a single borehole.

3. Boundary Conditions Two sets of boundary conditions are required: one set for the flow model and one set for the transport model. Finite element mesh representing a 16 borehole field.6 m) Figure 2-3.39 A mesh for a four-by-four configuration borehole field was constructed by using the single borehole mesh (Figure 2-2) as the basis for the mesh at each borehole and by expanding the mesh in the direction of ground-water flow as shown in Figure 2-3.3. 57. 328 ft (100 m) 1640 ft (500 m) . 2. This mesh consisted of 4532 nodes.7 ft (17.

The required heat flux is imposed using a source term in the ground-water flow equation at these nodes (representing injection of the heating/cooling water).40 In the flow model. Tw was set to achieve the required heat input (the ground thermal load). In order to impose the ground thermal loads as boundary conditions at the U-tube pipe walls.53 x 10-19 ft3/s [1. Dirichlet boundary conditions were set on all four sides of the model domain. a zero flux condition for the mass (heat) transport equation was applied at each of the sixteen nodes forming each pipe wall. some adaptation of the usual boundary condition was required.0 x 10-20 m3/s]) so as not to disturb the ground-water flow field. Second-type or fixed-gradient (Neumann) boundary conditions were set on the upper and lower boundaries of the flow domain and were specified as zero flux. This arises from the use of the mass transport equation to model heat transport. so that. V*. These conditions represent fixed background or far-field temperatures. In the transport model. was set negligibly small (3.0 Btu/lbo F (4180 J/kg-oC). The average temperature of the heat exchange fluid in each borehole is taken as the average of the nodal temperatures of the 32 nodes defining the U-tube pipe . The flow injected. First.4 lb/ft3 (1000 kg/m3) and 1. first-type or fixed-value (Dirichlet) boundary conditions were set on the left-hand and right-hand boundaries in order to impose a fixed hydraulic gradient across the domain. This work assumes that no ground-water recharge takes place across the water table within the model domain. The temperature of this injection flow. Tw ≈ q* ρlcl V * (2-9) The values of ρl and cl are taken as constants of 62.

41 in each borehole. Validation of the Numerical Model In order to check the accuracy of the AQUA3D model and the implementation of the boundary conditions. was set at a fixed value representative of in-situ thermal conductivity test conditions. However. q*. these are mostly specific to pollutant-transport problems (T is replaced by solute concentration in Equation 2-4) involving point or line sources with uniform concentration in time. However. Where the whole borehole field was modeled. an appropriate analytical solution was sought. The most appropriate analytical solution found was that described by Eskilson (1987) for steady-state heat extraction from a borehole in a ground-water flow field. s Eskilson (1987) describes the steady-state temperature at the borehole wall (Tsw. the heat input per pipe node.3.4. Numerous analytical solutions have been developed for the advection-dispersion equation (Equation 2-4). q* was determined from the time-varying building loads. Where single borehole cases were simulated. The literature gives little to no attention to analytical solutions describing the explicit transport of heat in ground water. Eskilson’ solution contains some approximations. Fetter (1988) and Bear (1972) summarize solutions for boundary and initial conditions describing situations that are commonly found in nature.b) as: . 2.

b = Tom q* + H  1  H   H  ln    2r  − Pw  l   2πk s   b        (2-10) H  The logarithmic term gives the steady-state resistance and Pw   is a correction term l  accounting for ground-water flow which is approximated as: 1 H H  + γ ln (3) − Pw   = ln 2l 2 l  for l defined as: (2-11) l= 2k s ρ lc l q where l >>rb (2-12) Comparisons of the steady-state temperature at the wall of a pipe were made using both the AQUA3D and the Eskilson (1987) solution. a single leg of a typical U-tube). a pipe depth of 250 ft (76. .9%. soil thermal property values of sand from Table 2-1.e. The temperature predicted by AQUA3D at steady-state time and Eskilson’ (1987) solution were 101.55oC). respectively. a far-field temperature of 63oF (17. and a heat flux of 8530 Btu/hr (2500 W). The following data were used as inputs: a pipe diameter of 0. the percent difference in error between the two solutions is 4.42 Tsw .2oC). s In terms of temperature increase above the far-field temperature.4oF (38. water properties as described above.2 m). (2 cm) (i.54oC) and 99.787 in. This error was considered to be acceptable.59oF (37.

simulations were made for each case but with zero ground-water flow. “typical” ground water flow . Single-Borehole Simulations The numerical model was initially used to determine average borehole temperatures for a range of soil and rock types over a two-year simulation time. Plots of average borehole fluid temperature versus time for three example geologic materials are shown in Figure 2-4. However. Results And Discussion 2.2 m) deep borehole. The model domain is that shown in Figure 2-2.43 2. After a one-year period. and appears to have reached a steady state.1.4. For comparison purposes. A constant heat flux of 8530 Btu/hr (2500 W) was applied on a U-tube in a 250 ft (76. A review of these plots reveals that a “typical” ground water flow rate in a coarse sand dramatically lowers the average borehole fluid temperature when compared to the zero-flow case. A simulation time of two years with a time step of 5 days was used for these simulations. The hydraulic and thermal property inputs are those listed in Table 2-1 and a hydraulic gradient of 0. The initial temperature and first-type boundary conditions were set at 63oF (17.01 was assumed. A small reduction in peak temperature is shown for the case of fine sand.4.2oC). the average fluid temperature in the borehole is approximately 15oF (8.3oC) lower than the average fluid temperature in the borehole where no ground-water flow was simulated. The objective was to examine trends in heat exchanger performance with increasing Peclet number.

25 0.00 Average Borehole Fluid Temperature ( o C) 0.50 1.50 0.00 48.9 43.6 ft/yr (5.8 ft/yr (60 m/yr) 48. Borehole Fluid Temperature ( o C) 0.75 1.75 Time (years) Figure 2-4.1 15.2 26.1 15.05 m/yr) 48.8 32.7 21.1 15.8 32.00 Average Borehole Fluid Temperature ( o C) Avg.3 37.5E-5 m/yr) 0.75 1.00 1.25 0.25 1.75 1.00 No Ground Water Flow Ground Water Flow Rate = 2.50 1.25 0.2 26.50 1.2 26.44 (a) coarse sand 120 110 Average Borehole Fluid Temperature ( o F) 100 90 80 70 60 0.00 1.7 21. . time for (a) coarse sand.3 37.7 21.00 No Ground Water Flow Ground Water Flow Rate = 196.8 32. (b) fine sand.6 2.50 0.75 Time (years) (b) fine sand 120 110 Average Borehole Fluid Temperature ( o F) 100 90 80 70 60 0.25 1.00 No Ground Water Flow Ground Water Flow Rate = 16. Average borehole fluid temperature vs.75 Time (years) (c) shale 120 110 Avg.25 1.9 43.00 1. Borehole Fluid Temperature ( o F) 100 90 80 70 60 0.6 2.8E-4 ft/yr (8.50 0.3 37.9 43. and (c) shale.6 2.

These data are used either with an analytical model or with a numerical model and parameter estimation technique to arrive at a value of thermal conductivity of the soil/ rock formation. Simulated In-Situ Thermal Conductivity Tests The second objective of the single-borehole simulations was to determine the effects of ground-water flow (in a material where ground-water flow is expected to be significant) on the interpretation of data from in-situ ground thermal conductivity tests. limestones. advection of heat by flowing ground water is a significant process contributing to heat transfer in the ground. the simulated in-situ thermal conductivity test calculations have been based on coarse sand properties.45 rates in rocks such as granites.4. the borehole temperature response . The trends shown in these results are in agreement with the previous Peclet number analysis. At Peclet numbers of order one or higher. conduction is the dominant heat transfer process and enhancement to the heat exchanger performance is negligible. 2. At Peclet numbers of order less than one. dolomites. Accordingly. and shales were found to have a negligible effect on the average borehole fluid temperature. in-situ thermal conductivity tests involve the application of a steady heat flux to a test borehole along with the measurement of the temperature response of the circulating water.2. Here. As previously discussed. The previous results showed the effects of ground-water flow to be most significant in the cases of gravel and coarse sand.

2oC) and the model domain is that shown in Figure 2-2. Twelve cases were simulated as listed in Table 2-3.5 minutes. the time to reach steady-state conditions decreases and (2) as ground water velocity increases. Also. ‘ effective’thermal conductivities have been estimated for different flow conditions. except the ground water flow velocity was varied from a “typical” value of 196. A review of Figure 2-5 shows that ground water flow in a coarse sand significantly impacts the average borehole fluid temperature over the time scales of an in-situ ground thermal conductivity test. In-situ test conditions were modeled by applying a constant heat flux of 8530 Btu/hr (2500 W) on a U-tube in a 250 ft (76.2 m) deep borehole. the steady-state temperature decreases. the deviation from the zero-flow condition can be seen to be dependent on the duration of . The simulation time periods were 50 hours and one week.5 ft/yr (600 m/yr) by adjusting the hydraulic conductivity value.46 is calculated for a range of ground-water flows using the AQUA3D model. Two noteworthy conclusions can be drawn from these simulations: (1) as ground water velocity increases. The initial temperature and first-type boundary conditions were set at 63oF (17. Hence. corresponding to typical durations of in-situ ground thermal conductivity tests.8 ft/yr (60 m/yr) to a more extreme value of 1968. Resulting temperature responses for the 12 cases are plotted in Figure 2-5. The model time step was 2. These data have been analyzed in exactly the same way as if the data had been measured in-situ. Model input hydraulic and thermal property values are those listed in Table 2-1 for a coarse sand.

temperatures are further reduced with increasing duration. However. the predicted effective thermal conductivity values from a conduction-based model are significantly different. These values are “effective” values since they include the effects of ground water advection.8 ft/yr (480 m/yr) 1968. depending on the duration of the simulated test. it is not clear if the . The actual values are listed by case number in Table 2-4.8 ft/yr (60 m/yr) 393.8 ft/yr (480 m/yr) 1968.5 ft/yr (600 m/yr) The effective thermal conductivity values predicted by the Austin et al. TABLE 2-3 Case Summary of Simulated In-Situ Ground Thermal Conductivity Tests Case 1 2 3 4 5 6 7 8 9 10 11 12 Simulation Time Period 50 hours 50 hours 50 hours 50 hours 50 hours 50 hours 1 week 1 week 1 week 1 week 1 week 1 week Ground Water Flow Velocity No Ground Water Flow 196. Hence.8 ft/yr (60 m/yr) 393.47 the test. the duration of the test could be expected to have an influence on the estimated thermal conductivity derived from an in-situ test.4 ft/yr (240 m/yr) 1574. at this stage of the design process.4 ft/yr (240 m/yr) 1574.7 ft/yr (120 m/yr) 787. (2000) model are plotted against the corresponding ground-water flow velocity for each of the two in-situ test simulation times (50 hours and 1 week) in Figure 2-6.7 ft/yr (120 m/yr) 787.5 ft/yr (600 m/yr) No Ground Water Flow 196. A review of these results shows that as ground water flow velocity increases.

48

50-hour data set or the 1-week data set produces more representative values, or if either data set produces representative values at all.

(a)
90 Average Borehole Fluid Temperature ( oF) 85 80 75 70 65 60 0 10 20 30 Time (hrs) 40 50 Case 1 Case 4 Case 2 Case 3 Case 5 32.2 Average Borehole Fluid Temperature ( oC)
Average Borehole o Fluid Temperature ( C)

29.4 26.7 23.9 21.1 18.3 15.6

Case 6

(b)
95

Case 7
Average Borehole o Fluid Temperature ( F) 90 85 80 75 70 65 60 0 20 40 60

Case 8 Case 9 Case 10 Case 12 Case 11

35.0 32.2 29.4 26.7 23.9 21.1 18.3 15.6

80 100 Time (hrs)

120

140

160

Figure 2-5. Average borehole fluid temperatures for the 12 simulated in-situ ground thermal conductivity test cases in a coarse sand with ground water velocities ranging from 0 to 1968 ft/yr (600 m/yr) for (a) 50 hours and (b) one week.

49

Ground Water Velocity (m/yr) 0 F) 16 14 12 10 8 6 4 2 0 0.0 328.1 656.2 984.2
50 hour simulated in-situ test 1 week simulated in-situ test
o

100

200

300

400

500

600 27.7
o

Predicted k Value (Btu/hr-ft-

24.2 20.8 17.3 13.8 10.4 6.9 3.5 0.0 1312.3 1640.4 1968.5

Ground Water Velocity (ft/yr)

Figure 2-6. Predicted effective ground thermal conductivity values versus ground-water flow velocity for 50-hour and 1-week simulated insitu thermal conductivity tests.

2.4.3. Borehole Field Simulations

In order to investigate the effects of ground-water flow on borehole field performance and system design procedures further, the predicted ground thermal conductivities were used to design a borehole field for a test building. The test building was an actual building located in north-central Oklahoma. This building is a single-story office building with 8 thermal zones and has a predominant demand for cooling. The hourly building loads were determined for one year using building energy simulation software (BLAST, 1986). The building loads were then converted to ground loads under

Predicted k Value (W/m- C)

50

the assumption that all heat pumps in the system have a constant coefficient of performance of 4.0. The ground loads for this building are shown in Figure 2-7.

TABLE 2-4 Summary of Borehole Field Design Parameters for Each Test Case
Case Number Simulation Ground-Water Duration Flow Rate Ground Thermal Design Borehole Conductivity Depth Predicted by Predicted by Numerical Model of Design Software of Austin et al. (2000) Spitler et al. (1996) Btu/hr-ft-oF (W/m-oC) 0.643 (1.11) 0.650 (1.12) 0.731 (1.26) 1.146 (1.98) 3.657 (6.33) 6.074 (10.51) 0.625 (1.08) 0.691 (1.20) 0.962 (1.66) 2.250 (3.89) 8.229 (14.24) 15.107 (26.14) ft (m) 239.98 (73.15) 238.56 (72.71) 224.10 (68.31) 171.56 (52.29) 87.24 (26.59) 61.58 (18.77) 243.86 (74.33) 230.86 (70.37) 191.58 (58.39) 115.91 (35.33) 48.02 (14.64) 26.90 (8.20)

(hours)

ft/yr (m/yr) 0 196.85 (60.00) 393.70 (120.00) 787.40 (240.00) 1574.80 (480.00) 1968.50 (600.00) 0 196.85 (60.00) 393.70 (120.00) 787.40 (240.00) 1574.80 (480.00) 1968.50 (600.00)

1 2 3 4 5 6 7 8 9 10 11 12

50 50 50 50 50 50 168 168 168 168 168 168

A 16 borehole field (four-by four boreholes in a square pattern) was deemed adequate for the test building ground loads (Figure 2-7).2oC) maximum and 35oF (1. a commercially-available ground-loop heat exchanger design software tool developed by Spitler et al. The borehole depths were sized for 20years of operation. representing heat extracted from the ground. . For this study. This was done with GLHEPRO. Hourly ground loads for the test building. (1996).7oC) minimum. Heating load is shown negative. The monthly loads and peak hourly loads are generally input in the design software. no peak hourly loads were specified for the sake of the computational time required for the subsequent borehole field simulations (see discussion below).51 250 200 Ground Loads (kBtu/hr) 150 100 50 59 44 Ground Loads (kW) 29 15 0 0 -50 -100 -150 -200 0 2190 4380 Time (hours) 6570 -15 -29 -44 -59 8760 Figure 2-7. Peak design entering fluid temperatures to the heat pump were specified at 90oF (32. cooling load is shown positive. Borehole field designs were produced for each of the twelve test cases. representing heat rejected to the ground.

Ground Water Velocity (m/yr) 0 250 Design Borehole Depth (ft) 50 hour simulated in-situ test 100 200 300 400 500 600 76.4 1968.2 1 week simulated in-situ test 984.0 45.3 0. the corresponding effective thermal conductivity shown in Figure 2-6 and Table 2-4 was input into the ground-loop heat exchanger design software (GLHEPRO). Design borehole depths versus ground-water flow velocity for 50hour and 1-week simulated in-situ thermal conductivity tests.1 656. The borehole depths predicted by GLHEPRO are plotted against the corresponding ground water flow velocity for each of the two in-situ test simulation times (50 hours and 1 week) in Figure 2-8. The model domain was that previously described for the multi-borehole field simulations shown in Figure 2-3.52 For each test case. The simulation time for all cases was 10 years and the model time step was 5 days. AQUA3D was further used to simulate the long-term performance of each borehole field designed from the simulated in-situ ground thermal conductivity test cases.2 Design Borehole Depth (m) 200 150 100 50 0 0.3 1640.5 Ground Water Velocity (ft/yr) Figure 2-8.5 15.7 30.2 61.0 328. The values are listed by case number in Table 2-4.0 1312. The simulated heat flux at the internal boundary nodes defining the Utube pipes was a time-varying source corresponding to the monthly ground loads for the .

Each 10-year simulation required approximately 60 hours of computation time on a personal computer with a 233 megahertz pentium II processor. except for the borehole depths.8 ft/yr (480 m/yr) and 1968. This is shown by cases 5 and 6 which represent thermal conductivity values determined from a 50-hour test at ground water flow velocities of 1574. all of the cases with ground-water flow show no changes in minimum and maximum temperatures from year to year. . and by cases 11 and 12 which are for the same flow rates but based on thermal conductivity values determined from one-week test data. This implies that for in-situ test cases where the average borehole fluid temperature reaches steady-state in a relatively short time (as demonstrated by case 4/10.53 test building. After the second year. The thermal conductivity values determined in cases 11 and 12 are unrealistically high and consequently the design borehole depths are too shallow.5 ft/yr (600 m/yr). Hydraulic and thermal property inputs for each borehole field case number is the same as the corresponding single-borehole case number. Some notable differences can be seen between the borehole field designs based on 50-hour test data compared to one week test data. respectively. the result is that the maximum peak temperature of the simulated borehole field in both cases exceeds the maximum design temperature during the first year. which are those listed in Table 2-4. Annual maximum and minimum peak temperatures are plotted for each case in Figure 2-9. Examination of the cases with no ground-water flow (cases 1 and 7) shows annual rises in peak temperature typical of cooling-dominated buildings. case 5/11.

3 o C) o F) 65 18.54 and case 6/12 in Figure 2-5).7 70 0 1 2 3 4 5 6 7 8 9 10 Time (years) Case 1 Case 2 Case 3 Case 4 Case 5 21.8 55 12.2 45 7. Minimum Temperature ( Minimum Temperature ( .2 80 26.8 55 12.3 o C) Maximum Temperature ( Maximum Temperature ( Minimum Temperature ( 100 37.1 25 0 1 2 3 4 5 6 7 8 9 10 Time (years) -3.9 Case 12 Case 7 Case 8 Case 9 Case 10 Case 11 Case 12 Figure 2-9.8 90 Maximum Design Temperature 32.7 35 Minimum Design Temperature 1.3 o C) Maximum Temperature ( Maximum Temperature ( Minimum Temperature ( 100 37.7 70 0 1 2 3 4 5 6 7 8 9 10 Time (years) Case 7 Case8 Case 9 Case 10 Case 11 21. (a) (b) 110 oF) 43.9 Case 6 Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 (c) (d) 110 oF) 43. Annual maximum (a) and (c) and minimum (b) and (d) average borehole fluid temperatures for the 16 borehole field simulations.3 o C) o F) 65 18.8 90 Maximum Design Temperature 32.2 45 7.2 80 26. increasing the duration of the in-situ test produces decreased confidence in the accuracy of the effective thermal conductivity value determined from the test.1 25 0 1 2 3 4 5 6 7 8 9 10 Time (years) -3.7 35 Minimum Design Temperature 1.

at the end of the cooling season). These cases have maximum peak temperatures of about 74oF (23.e. it is interesting to note from Figure 2-9 (a) that it is the cases where the ground-water flow is moderate (cases 2.3oC). when the average annual ground temperatures are the greatest (i. the annual maximum and minimum temperatures fell within the design conditions. and 4) that are most over-designed. It is at higher flow rates (cases 5 and 6) that the peak temperature was closest to the original design condition after ten years. Considerable drilling cost savings could be seen in cases like this where shallower borehole depths could have been adequate. some 16oF (8. 3.55 Except for cases 11 and 12. This illustrates the non-linearity introduced into the design problem by the presence of advection. It also illustrates the difficulty in adapting conventional design methods to accurately size closed-loop ground heat exchangers in cases with significant ground-water flow. The data are plotted for the end of September. Having followed conventional design procedures. .9oC) below the maximum design temperature. The temperature field predicted by the numerical model for case 8 is shown in Figure 2-10 in the form of a series of contour plots over the 10 year simulation period.

Temperature contours for Case 8 for the end of September of years 1. Year 10 100 m 500 m Figure 2-10. 5.56 ground water flow rate = 60 m/yr (197 ft/yr) end of Sept. and 10. . Year 1 Temperature Key (oC) end of Sept. Year 2 end of Sept. Year 5 end of Sept. 2.

57 A further feature that is shown in the predicted temperature field (Figure 2-10) is the development of a peak in the temperatures immediately downstream of the borehole field after year 2. it appears that it is only in geologic material with high hydraulic conductivities. A finite-element numerical ground-water flow and heat transport model was used to simulate and observe the effects of ground-water flow on the heat transfer from a single U-tube closed-loop ground heat exchanger in various geologic materials. such as coarse-grained soils (sands and gravels) and in rocks exhibiting secondary porosities such fractures and solution channels. A simple but useful method of assessing the relative importance of heat conduction in the ground versus heat advection by moving ground water is demonstrated through the use of the dimensionless Peclet number. From these simulations and from a Peclet number analysis. This arises from the advection of the heat rejected to the ground at the boreholes during the previous year. In the contours plotted for year 10. that ground-water flow could be expected to have a significant effect on closedloop heat exchanger performance. thermal plumes from previous years can be distinguished. Concluding Remarks and Recommendations for Future Work Using a compilation of “typical” hydraulic and thermal properties of soils and rocks. 2. a preliminary analysis of the effects of ground-water flow on the design and performance of closed-loop ground-coupled heat pump systems has been made. .5.

these values were artificially high. As expected. considerable progress has been made in developing both in-situ test methods and design procedures for borehole field design for situations where there is no ground-water flow. in all cases of ground-water flow. the presence of moderate ground-water flows had the effect of removing the year-by-year increase in ground temperature found in systems where there is no ground-water flow. it appears difficult to adapt results from current design and in-situ measurement methods to fully account for ground-water flow conditions. Research would . Over the last decade. The borehole fields designed using conventional methods and the derived effective thermal conductivities were generally over-designed. However. These data were analyzed as if it came from real in-situ sources to arrive at effective thermal conductivity values. For coarse-grained sands. The finite-element numerical ground water flow and heat transport model was also used to simulate the 10-year performance of borehole fields designed from application of conventional design procedures using the derived thermal conductivity data. in some cases at very high ground-water flow rates. From this preliminary assessment of the effects of ground-water flow. temperatures were found to rise above the design criteria.58 The effect of ground-water flow on in-situ thermal conductivity test results has been examined by numerically simulating test conditions around a single borehole under different flow conditions. Results from one week test data have been shown to be no more reliable than data from 50-hour tests.

. • Experimental investigation of potential in-situ test data analysis methods at sites with significant ground-water flow. that could be used for closed-loop ground heat exchanger design. or adaptations to existing methods.59 be required in a number of areas before the same progress could be made to deal with the situations of ground-water flow. • Development of design guidelines and software tools that could be used by practicing engineers for in-situ testing and system design tasks in situations of significant ground-water flow. These include: • Identification of suitable numerical and/or analytical models that include ground- water flow and could be used to analyze in-situ test data. • Identification of suitable design methods.

In order to adequately dissipate the imbalanced annual loads. car washes. In cases where the excess heat cannot be used beneficially. shallow ponds can provide a cost-effective means to balance the thermal loading to the ground and reduce heat exchanger length. supplemental components can be integrated into the groundloop heat exchanger design. In some applications. and pavement heating systems. Introduction Commercial buildings and institutions are generally cooling-dominated. GSHP systems that incorporate some type of supplemental heat rejecter are commonly referred to as hybrid GSHP systems. A Model for Simulating the Performance of a Shallow Pond as a Supplemental Heat Rejecter with Closed-Loop Ground-Source Heat Pump Systems 3. the required ground-loop heat exchanger lengths are significantly greater than the required length if the annual loads were balanced. Consequently.60 3. the excess heat that would otherwise build up in the ground may be used for domestic hot water heaters. and therefore reject more heat than they extract over the annual cycle. The objective of the work presented in this chapter has been to develop a design and simulation tool for modeling the performance of a shallow pond that can be usefully . To effectively balance the ground loads and reduce the necessary size of the ground loop heat exchanger.1. under these circumstances. ground-source heat pump systems may be eliminated from consideration during the feasibility study phase of the HVAC design process because of excessive first cost.

The pond model has been developed in the TRNSYS modeling environment (SEL. General Overview Pertinent concepts of heat transfer in ponds and lakes have been summarized by many sources.61 and cost-effectively integrated into a ground-source heat pump system as a supplemental heat rejecter. Thermal radiation can also account for a significant amount of cooling during night hours. Conduction heat . Convective heating or cooling to the atmosphere is less significant. Dake and Harleman (1969) conducted studies of thermal stratification in lakes and addressed the overall thermal energy distribution in lakes. As an example of the model’ s applicability.2. ASHRAE (1995). Oklahoma with a hypothetical closed-loop GSHP system with and without a shallow pond supplemental heat rejecter.1. Natural convection of water due to buoyancy effects is the primary mechanism for heat transfer within a surface water body.2. The main cooling mechanism is evaporation. and Kavanaugh and Rafferty (1997) describe heat transfer in lakes in relation to their use as heat sources and sinks. Solar energy is identified as the main heating mechanism for ponds and lakes. 3. GSHP system simulation results are presented for a commercial building located in Tulsa. ASHRAE (1995b). Heat Transfer In Ponds 3. The model has been validated by comparing simulation results to experimental data. 1997) and can be coupled to other GSHP system component models for short-time step (hourly or minutely) system analyses.

. Raphael (1962) developed a numerical model for determining the temperature of surface water bodies as heat sinks for power plants. generally greater than 20 . Raphael reported that the model successfully predicted the temperature changes in a river used as a heat sink for a power plant. The relevant heat transfer mechanisms occurring within shallow ponds are illustrated in Figure 3-1. If inflow and outflow rates are high enough. Shallow ponds are generally thermally unstratified. Existing Pond and Lake Models Several mathematical and computer models have been developed for simulation of lakes used as heat sinks/sources and for solar ponds. 3. Consequently. Input data to the model included weather data and inflow and outflow data for the water body. Therefore.2. the pond will not stratify. water in deeper ponds will completely over-turn.30 ft (6.62 transfer to the ground is generally a relatively insignificant process.1 m). Thermal stratification of the water body was not considered. over the annual cycle.9.2oF (4oC). thermal stratification occurs only in ponds and lakes that are relatively deep.1 . Natural stratification of deeper ponds and lakes is due to buoyancy forces and to the fact that water is at its greatest density at 39. except in cases where the water surface is frozen. Thermal stratification in ponds is also dictated by inflow and outflow rates or ground water seepage rates.2. with low inflow rates.

Cantrell and Wepfer (1984) developed a numerical model for evaluating the potential of shallow ponds for dissipating heat from buildings. Thermal stratification of the water body was not considered. Heat transfer mechanisms in shallow ponds. The results of that work showed that the heat transfer at the water/air interface is highly dependent on the natural water temperature and the wind speed. 10-feet (3.141 m2). Jobson (1973) developed a mathematical model for water bodies used as heat sinks for power plants.048 m) deep pond in . Thermal stratification of the pond was not considered.63 Wind Convection E v a p o r a t I o n Thermal Radiation Incident Solar Radiation Fluid from heat pump Coil Heat Rejection/Absorption Fluid return to heat pump Subsurface Subsurface Conduction Conduction Groundwater Discharge Figure 3-1.. The model takes weather data and building cooling load data as inputs and computes the steady-state pond temperature using an energy balance method. The model showed that a 3-acre (12.

This is accomplished by suppressing natural convection within the pond induced by bottom heating. Srinivasan and Guha (1987) developed a model similar to the model of Rubin et al. The Srinivasan and Guha (1987) model consisted of three coupled differential equations.78oC) over a daily cycle. Solar radiation was modeled as an exponentially decaying function through the pond depth. usually by adding a brine layer at the pond bottom. solar ponds have three distinct zones as described by Newell (1984): (1) a top layer which is stagnated by some method and acts as a transparent layer of insulation. As a result. Ohio could reject 1000 tons (3516 kW) of thermal energy with a maximum increase in pond temperature of about 5oF (2. (1984) for solar ponds. The purpose of a solar pond is to concentrate heat energy from the sun at the pond bottom. Large-scale convective currents in the pond were assumed to be negligible while small-scale convective currents were handled by allowing the coefficient of heat diffusion to vary through the pond depth. The model also successfully predicted seasonal variations in solar pond temperatures with various heat extraction rates. The model successfully predicted seasonal variations in solar pond temperatures. and (3) a lower layer where solar energy is collected. . The model of Rubin et al. Solar radiation in each zone is computed as a function of depth. (1984) applied an implicit finite difference scheme to solve a onedimensional heat balance equation on a solar pond. each describing a thermal zone within the solar pond. (2) a middle layer which is usually allowed to be mixed by natural convection. (1984) developed a model for solar ponds. Rubin et al.64 Cleveland.

As such. The model is driven by monthly average bin weather data and handles both heat extraction and heat rejection. the model had some difficulty in matching the intermediate zone (the thermocline). As concluded by Pezent and Kavanaugh (1990). especially during times of heat rejection.55oC). a numerical method is necessary to more accurately predict the thermocline profile. The temperatures within the upper zone of the lake (the epilimnion) and the lower zone of the lake (the hypolimnion) were predicted to within 4oF (2. The model essentially combined the models of Srinivasan and Guha (1987) to handle stratified cases and of Raphael (1962) to handle unstratified cases. The model presented in this paper is based on the assumption that thermal gradients in shallow ponds are negligible.22oC) and approximately 1oF (0. when lakes are generally unstratified. respectively. thermal stratification of a lake could be handled in the summer months.65 Pezent and Kavanaugh (1990) developed a model for lakes used as heat sources or sinks with water-source heat pumps. this model allows the pond performance to be simulated on hourly or minutely time scales. Further. and neglected in the winter months. the model favorably predicted a lake temperature profile in Alabama. With no heat extraction or rejection. when lakes are generally most stratified. . However. This model is developed in the TRNSYS modeling environment and can be coupled to other component models for larger system simulations. perhaps due to the fact that this zone possesses moving boundaries (unlike the boundaries of a solar pond which are more distinct).

3. the slinky heat .19 m) by 3 ft (0.5 ft (1. Each slinky pipe was made of high-density polyethylene plastic and is 500 feet (152. approximately 8 in. Experimental Methods 3.40 m) long with a nominal diameter of ¾ in. (0. (0. (0. The pipe was coiled such that the resulting slinky heat exchanger was 40 ft (12. In the 2-ft (0.3.66 3.3 cm) thick.91 m) and a 10-in.5-ft (1. Two ponds were constructed on a test site at the north end of the campus. The ponds are rectangular with a plan area of 40 ft (12. Pond Description and Data Collection Pond construction and data collection for this study have been conducted by researchers affiliated with the Division of Engineering Technology at the Oklahoma State University. Each pond was constructed with vertical sidewalls with one of the ponds being 2 feet (0.254 m) pitch (the separation distance between the apex of each successive loop).019 m).91 m).61-m) deep pond. In the 3. the slinky heat exchanger was positioned horizontally within the pond at a depth of approximately 10 in.61 m) deep and the other being 3.19 m) long with a diameter of 3 ft (0.1.07 m) deep.07-m) deep pond. The walls and the bottom of each pond were constructed of reinforced concrete. (20.254 m). Heat was rejected to each pond by circulating heated water through a “slinky” heat exchanger (a pipe coiled in a circular fashion such that each loop overlaps the adjacent loop) installed in each pond.

(3) near the pond surface at the end opposite from the supply end. (2) below the slinky at its center.3. Weather Instrumentation and Data Collection Weather data for this study were obtained from the Oklahoma Mesonet (mesoscale network).2. a water heating element. All sensor information was recorded by the data acquisition system at time intervals of 6 minutes. The tests were controlled to maintain a set supply water temperature by heating the supply water if the temperature fell below a set point. and a watt transducer. Two set point temperatures were used in this study: 90oF (32.67 exchanger was positioned vertically within the pond along the center-line of the long axis of the pond. Depending on the weather parameter. .9oC) in the winter season. Each system also included a flow meter. which is a weather station network consisting of weather monitoring sites scattered throughout Oklahoma. 3. data are recorded at time intervals ranging from 3 seconds to 30 seconds and averaged over 5-minute observation intervals. The temperature of the pond water was measured by thermistors positioned at four locations within the pond: (1) near the pond surface at the center of the slinky. Slinky supply and return water temperatures were measured by thermistors embedded in the slinky header. and (4) below the slinky at the end of the pond opposite from the supply end.2oC) in the summer season and 75oF (23.

wind direction.68 Weather data at 15-minute intervals for the Stillwater monitoring station were acquired for the time periods of interest for this study. Model Development 3. Equation 3-1 can be expressed to describe the rate of change in average pond temperature as: . The Stillwater station is located approximately 1 mile from the test pond site. and solar radiation. cp is the specific heat capacity of the pond water.1. Governing Equations The governing equation of the model is an overall energy balance on the pond using the lumped capacitance (or lumped parameter) approach: qin − q out = Vρc p dT dt (3-1) where qin is the heat transfer to the pond. 3. ρ is the density of the pond water. This dt approach assumes that temperature gradients within the water body are negligible.4. air temperature. relative humidity. qout is the heat transfer from the pond. and dT is the rate of change of temperature of the pond water. V is the pond volume. Data for the following parameters were obtained: wind speed. Considering the heat transfer mechanisms shown in Figure 3-1. Further details of the weather station network may be found in Elliott et al. (1994).4.

qgroundwater is the heat transfer due to groundwater inflow or outflow. The s s reflectance (ρ’ is computed from: ) ρ' = τ a − τ (3-3) . and ASHRAE (1997). qconvection is the convection heat transfer at the pond surface. Solar Radiation Heat Gain Solar radiation heat gain (qsolar) is the net solar radiation absorbed by the pond.69 dT q solar + qthermal + qconvection + q ground + q groundwater + q evaporation + q fluid = dt Vρc p (3-2) where.1. the angle of incidence (θ) of the sun’ rays is first computed at each time step from correlations given by s Spencer (1971). qsolar is the solar radiation heat gain to the pond. 3. Duffie and Beckman (1991). It is assumed that all solar radiation incident on the pond surface becomes heat gain except for the portion reflected at the surface. The angle of refraction (θr) of the sun’ rays at the pond surface is given by Snell’ Law. qevaporation is the heat/mass transfer due to evaporation at the pond surface.1. qground is the heat transfer to/from the ground in contact with the pond. To determine the reflected component of solar radiation. qthermal is the thermal radiation heat transfer at the pond surface. Each of the heat transfer terms used in the above equation is defined below.4. and qfluid is the total heat transfer to/from the heat exchange fluid flowing in all spools or coils in the pond.

These are computed after Duffie and Beckman (1991) as: τa = e − µ 'd cos θ r (3-4) and − µ 'd 1 1 − r|| 1 − r⊥  cos θr e  τ= + 2 1 + r|| 1 + r⊥    (3-5) where µ’is the extinction coefficient for water. r|| represents the parallel component of unpolarized radiation.70 where τ is the transmittance of solar radiation by the pond surface and the subscript ‘ a’ refers to the absorbed component. the total reflectance is approximated by the beam reflectance) and Apond is the area of the pond surface. and r⊥ represents the perpendicular component of unpolarized radiation which are computed after Duffie and Beckman (1991) as: tan 2 (θ r − θ ) r|| = tan 2 (θ r + θ ) r⊥ = sin 2 (θ r − θ ) sin 2 (θ r + θ ) (3-6) (3-7) Finally. the amount of solar radiation absorbed by the pond (qsolar) is expressed as: q solar = I (1 − ρ ' ) A pond (3-8) where I is the solar radiation flux incident on the pond surface (here. d is the pond depth. The .

and Tsky is the sky temperature in absolute units. in which case I is computed from: I = I b cosθ + I d (3-9) 3. Tpond is the pond temperature in absolute units. Tsky is computed from relationship given by Bliss (1961) which relates sky temperature to the dew point temperature (Tdp) and the dry bulb temperature (Tdb): Tsky Tdp 4   = Tdb 0.8 +  250    1 (3-11) where all temperatures are in absolute units.2. The model uses the TRNSYS psychrometric subroutine to compute Tdp if it is unknown. Tdp is computed from either of the following pairs of state properties: (1) dry bulb temperature (Tdb) and wet bulb temperature (Twb) or (2) Tdb and relative humidity.4. This model uses a linearized radiation coefficient (hr) defined as: T pond + Tsky hr = 4εσ   2      3 (3-10) where ε is the emissivity coefficient of the pond water. The thermal radiation heat transfer (qthermal) is then computed by: .1. Thermal Radiation Heat Transfer This heat transfer mechanism accounts for heat transfer at the pond surface due to thermal or long-wave radiation.71 model also accepts solar radiation in the form of beam (Ib) and diffuse (Id) components. σ is the Stefan-Boltzmann constant.

72 qthermal = hr Apond (Tsky − T pond ) (3-12) 3. For a pond surface. ∆T refers to the temperature difference between the pond and the air.3. Convection Heat Transfer at the Pond Surface This mechanism accounts for heat transfer at the pond surface due to free and forced convection. and L is the characteristic length. β is the volumetric thermal expansion coefficient of air. the Nusselt Number (Nu) is often correlated to the Rayleigh Number (Ra). α is the thermal diffusivity of air. and ν are all evaluated at the film temperature which can be approximated as the average of the air and pond temperatures. the thermal properties of air are computed at each time step using correlations given by Irvine and Liley (1984). For horizontal flat plates. 1996): . correlations for a horizontal flat plate are the most applicable. Several empirical formulations exist for determining the convection coefficient for different geometries.β. the characteristic length (L) can be defined as the ratio of the area (A) to the perimeter (P) (Incropera and DeWitt. The thermal properties α. ν is the kinematic viscosity of air. which describes the relative magnitude of the buoyancy and viscous forces in the fluid: gβ (∆T ) L3 Ra = να (3-13) where g is the acceleration due to gravity. In free convection heat transfer.4. In the model.1.

15Ra 1 3 (107 > Ra > 1011 – turbulent flow) (3-15b) The convection coefficient (hc) for free convection can then be determined from: hc = Nu k L (3-16) where k is the thermal conductivity of air evaluated at the film temperature as with the other thermal properties described above and L is the characteristic length described by Equation 3-14.73 L= A P (3-14) In external free convection flows over a horizontal flat plate.54Ra 4 (104 < Ra < 107 – laminar flow) (3-15a) and Nu = 0. two empirical relations for the Nusselt number are used in the model as described by Incropera and DeWitt (1996) for free convection from the upper surface of a heated plate or the lower surface of a cooled plate: 1 Nu = 0. The Reynolds number is described as: . In forced convection heat transfer. Therefore. Nu is a function of the Reynolds (Re) and Prandtl (Pr) Numbers. the critical Rayleigh Number is about 107.

664 Re Pr 1 2 1 3 (laminar flow regime) (3-19a) and 4 1 Nu = 0. the pond surface).037 Re 5 Pr 3 (mixed and turbulent flow) (3-19b) The convection coefficient (hc) for forced convection can then be determined by Equation 3-16 with the characteristic length value described by Equation 3-14 for a horizontal flat plate. and the characteristic length (L) is now defined by the length dimension over which the wind blows and is determined from the pond orientation (degrees from north) and wind direction. two empirical relations for the Nusselt number are used in the model as described by Incropera and DeWitt (1996) for forced convection over a flat plate: Nu = 0. and k is the thermal conductivity of air. 1996).74 Re = vL ν (3-17) where v is the wind speed. the critical Reynolds Number is approximately 105 (Incropera and DeWitt.e. Therefore. For external forced convection over a flat plate (i. . ν is the kinematic viscosity of air. The Prandtl Number is defined as: Pr = cpµ k (3-18) where cp is the specific heat capacity of air. µ is the dynamic viscosity of air. all evaluated at the air film temperature.

Heat Transfer to the Ground This heat transfer mechanism accounts for heat conduction to/from the soil or rock in contact with the sides and the bottom of the pond. the convection heat transfer at the pond surface (qconvection) is computed by: qconvection = hc Apond (Tair − T pond ) (3-20) where Tair is the ambient air temperature and hc is taken as the maximum of the free convection coefficient and the forced convection coefficient. climatic factors. This mechanism of heat transfer is highly site-specific and complex and depends on many factors such as soil/rock thermal properties. (1984) was chosen to determine heat losses/gains from the bottom and sides of the pond. (1984) used a three-dimensional numerical code to compute steady-state ground heat losses from solar ponds of varying sizes. the ground thermal conductivity (kground). geometries.4. and sidewall insulation types. a semi-empirical approach developed by Hull et al. (1984) expresses ground heat losses from any pond as a function of the pond area. pond geometry.75 Finally. Hull et al. In this model. 3. Hull et al. and thermal loading history. and the distance .4. This practice of choosing the larger of the free and forced convection coefficients is recommended by Duffie and Beckman (1991) and McAdams (1954) and is used in the absence of additional experimental evidence regarding combined free and forced convection.1. pond perimeter.

76 from the pond bottom to a constant temperature sink. this heat transfer mechanism can be used to account for other inflows and outflows. a heat transfer coefficient for ground heat transfer (Uground) can be computed from:  k ground U ground = 0. and Ppond is the pond perimeter. such as make-up water or rain water. dgroundwater is the depth to the water table or the constant source/sink from the ground surface. 3. dpond is the pond depth. However.5.37 ground pond  A  pond       (3-21) where kground is the thermal conductivity of the ground. 1984).1. For practical purposes. the constant temperature sink can be taken as the groundwater table (Kishore and Joshi.4. Although ground water contributions may not be expected in shallow heat rejecter ponds. The conduction heat transfer between the ground and the pond is then given by: q ground = U ground A pond (Tgroundwater − T pond ) (3-22) It is recognized that the above conduction heat transfer model is a relatively simple representation of the true transient behavior of heat transfer in the ground. ground heat conduction is a relatively minor process affecting the overall heat transfer within the pond as compared to other processes. For a rectangular pond with vertical side walls. .999 d  groundwater − d pond k  P  + 1. Heat Transfer Due to Ground Water Seepage This heat transfer mechanism accounts for inflows and outflows of ground water to the pond.

This model uses the j-factor analogy to compute the mass transfer of ′ &w ) evaporating water ( m′ at the pond surface: &w ′ m′= hd ( w air − w surf ) (3-25) where hd is the mass transfer coefficient. 1980). and wsurf represents the humidity ratio of saturated air at the pond surface. These properties of ground water are computed from relationships given in the Handbook of Chemistry and Physics (CRC. 3.77 The volumetric groundwater flow rate (Q) is computed by Darcy’ Law: s Q = Ki ( Ppond (d pond − d groundwater ) + Apond ) (3-23) where K is the hydraulic conductivity of the soil/rock surrounding the pond and i is the hydraulic gradient. wair is the humidity ratio of the ambient air.4.1.6. The wsurf term is . Heat Transfer Due to Evaporation This heat transfer mechanism is the most important one contributing to pond cooling. The heat transfer contribution from ground water (qgroundwater) is then given by: q groundwater = Qρc p (Tgroundwater − T pond ) (3-24) where ρ and cp represent the density and specific heat capacity of ground water.

cp is the specific heat capacity of the air evaluated at the pond-air film temperature. Le is computed as: Le = α D AB (3-27) where α is the thermal diffusivity of the air evaluated at the pond-air film temperature and DAB represents the binary diffusion coefficient. (2) Tdb and relative humidity or (3) Tdb and Tdp. The wair term may also be computed using the TRNSYS psychrometric subroutine depending on what two state properties of the air are known. The thermal properties (α and cp) of air are computed at each time step using correlations given by Irvine and Liley (1984).072 Pair (280K < T < 450K) (3-28) .87 x10 − 10 T 2.78 computed using the TRNSYS psychrometric subroutine by setting Tdb and Twb equal to the pond temperature. The mass transfer coefficient (hd) is defined using the Chilton-Colburn analogy as: hd = hc c p Le 2 3 (3-26) where hc is the convection coefficient defined previously. DAB is computed after Mills (1995) who references Marrero and Mason (1972): D AB = 1. and Le is the Lewis number. The model accepts the following pairs of state properties for the calculation of wair if it is unknown: (1) Tdb and Twb.

the previous converged value is used as an initial guess and calculation of Tfluid is iterative. The heat transfer due to the heat exchange fluid (qfluid) is computed by: . 3. The thermal properties of the fluid are computed at each time step from correlations given in the Handbook of Chemistry and Physics (CRC.7.79 where T refers to the pond-air film temperature in absolute units and Pair is the atmospheric pressure in atmospheres. The thermal properties are computed at the average fluid temperature (Tfluid). Solution of the pond temperature is also an iterative procedure as discussed below.4. The heat transfer due to evaporation (qevaporation) is then computed by: ′ &w qevaporation = h fg A pond m′ (3-29) where hfg is the latent heat of vaporization and is computed at each time step from the relationship given by Irvine and Liley (1984). Heat Transfer Due to the Heat Exchange Fluid Heat transfer due to the heat exchange fluid represents the pond thermal load. 1980) for water and from correlations given by Wadivkar (1997) for an antifreeze solution. Since the outlet temperature at any current time step is not known. This model has been developed to account for water or antifreeze as the heat exchange fluid.1. This temperature is computed as the average of the inlet and outlet temperatures at the given time step.

Lspool is the length of one spool or circuit. The resistance terms are defined as follows (in terms of inner pipe radius): Ri = 1 .e. and ff represents a fouling factor at both the inner and outer pipe walls. Ro is the thermal resistance at external pipe surface. Equation 3-30 is based on the assumption that one spool is one flow circuit and that the flow rate is divided evenly between the circuits in a parallel arrangement. The term UApipe is expressed in terms of the inside pipe area as: UApipe = 2πri Lspool ΣRt (3-31) where ri is the inner pipe radius. and ΣRt represents the composite thermal resistance which is defined by the following resistance network: ΣRt = Ri + R pipe + Ro + ff (3-32) where Ri is the thermal resistance due to fluid flow through the pipe. Thus.80 q fluid = UApipe (T fluid − T pond )( N circuit ) (3-30) where UApipe is the overall heat transfer coefficient for the pipe expressed in terms of inside pipe area and Ncircuit refers to the number of flow circuits (i. Rpipe is the pipe thermal resistance. hi (3-33) . the number of spools) installed in the pond.

For this case. Equation 8-53). A Reynolds number of 2000 is assumed to be critical. and the characteristic length (L) is the inner pipe diameter.81 R pipe = Ro = ri ro ri k pipe 1  h  o r ln  o r  i         (3-34) and (3-35) where hi is the convection coefficient due to fluid flow through the pipe. kpipe is the thermal conductivity of the pipe material. The above convection coefficients are determined using correlations for the Nusselt number in flow through a horizontal cylinder. A constant heat flux at the pipe surface is assumed. the Nusselt Number is a constant equal to 4. fullydeveloped flow in the pipe (Re<2000). For laminar. ho is the convection coefficient at the outer surface of the pipe. v represents the mass flow rate of the heat exchange fluid. 1996. the Dittus-Boelter relation is used to compute the Nusselt Number: 4 Nu i = 0. since no specific correlations exist for a slinky coil. Nu is a function of the Reynolds and Prandtl numbers. Determination of Re is described in Equation 3-17. respectively. ν represents the kinematic viscosity of the heat exchange fluid. For turbulent flow. In convection heat transfer due to internal flow.023 Re 5 Pr x (3-36) .36 (Incropera and DeWitt. and ri and ro are the inner and outer radii of the pipe.

82 where Pr is defined by Equation 3-18 for µ. The inside pipe convection coefficient can then be determined by using Equation 3-16 where Nu is Nui.3 if the entering fluid is greater than the pond temperature and x is equal to 0.cp. the volumetric thermal expansion coefficient of water.60 +      0. The value of the exponent x in Equation 3-36 is dependent upon whether the entering fluid is being heated or cooled. k is the thermal conductivity of the heat transfer fluid.387 Ra 1 6 9 16  1 + (0. In free convection at the external pipe surface. 1996).559 / Pr)       8 27          2 (3-37) . and L is the characteristic length which is defined in this case as the inner diameter of the pipe (Incropera and DeWitt. and ν represent the thermal diffusivity of water.β. x is equal to 0. the Nusselt Number is a function of the Rayleigh Number. Ra is computed using Equation 3-13 where α. and k representing the thermal properties of the heat exchange fluid evaluated at its average temperature. and the kinematic viscosity of water. respectively. Nu for free convection from a horizontal cylinder is defined as (Churchill and Chu (1975) ):    Nu o = 0.4 if the entering fluid is less than the pond temperature. which is approximated as the average of the pipe surface and pond temperatures at the given time step. all evaluated at the pipe film temperature. The term ∆T refers to the temperature difference between the pipe surface and the pond temperatures.

4.cp. Now the external pipe convection coefficient can be determined by using Equation 3-16 where Nu is Nuo. 3. This outlet temperature is used to compute the average fluid temperature at the next iteration as described above. k is the thermal conductivity of the pond water evaluated at the pipe film temperature.83 where Pr is defined by Equation 3-18 for µ. and qcircuit is the heat rejected/extracted by one flow circuit. The outlet fluid temperature (Tout) is computed from an overall energy balance on the pipe: Tout = T fluid − qcircuit & 2mc p (3-38) & where m is the mass flow rate of the heat exchange fluid per flow circuit. and k representing the thermal properties of the pond water evaluated at the pipe film temperature.1.8. cp is the specific heat capacity of the heat exchange fluid. Solving the Overall Energy Balance Equation The differential equation describing the overall energy balance on the pond (Equation 3-2) is rearranged in the form: dT = x1T + x 2 dt (3-39) . and L is the characteristic length which is defined now as the outer pipe diameter.

3. The solution is given by the TRNSYS differential equation solver subroutine as:  x  x Tt = Tt − ∆t + 1 e x1∆t − 1   x2  x2  (3-40) where Tt is the average pond temperature at the new time step and Tt-∆t is the average pond temperature at the previous time step. Many of the quantities in the heat transfer equations described above require that the average pond temperature at the current time step be known. A companion model was also developed which manipulates any weather data needed for the pond model. and x2 contains all terms of Equation 3-2 that are independent of T.8x10-5oF (1x10-5oC) is used. A convergence criterion for the pond temperature of 1. Thus. Equation 3-39 is a linear first-order ordinary differential equation which is solved at each time step using the exponential function as an integrating factor. The weather component model makes use of the TRNSYS psychrometric subroutine to compute moist air properties given two known state properties. the actual pond temperature is arrived at iteratively.2.4.84 where T represents the pond temperature. The two . x1 contains all terms of Equation 3-2 that multiply T. Computer Implementation Thc component configuration for the pond model is shown in Figure 3-2.

number of spools or coils 10. pipe outer diameter 12. spool length 11. antifreeze concentration if used 16. The weather component model also computes the sky temperature. extinction coefficient for water 9. the solar radiation on a horizontal surface. pond length 5. pond depth 7. pond width 6. emissivity coefficient 8. pipe thermal conductivity 13. ground water or far field temperature 19. and the solar incidence angle. air temperature sky temperature humidity ratio wind speed wind direction solar angle of incidence solar radiation inlet fluid temperature inlet mass flow rate POND MODEL PARAMETERS: 1. pond orientation from north 4. soil hydraulic conductivity 20.85 state properties are dry bulb temperature and one of wet bulb temperature. or dew point temperature. relative humidity. fouling factor 18. depth to water table pond temperature outlet fluid temperature mass flow rate heat rejected/absorbed Figure 3-2. Pond model component configuration. initial pond temperature 3. A computer algorithm is shown in Figure 3-3 in the form of a flow chart. . fluid type (water or antifreeze) 15. ground thermal conductivity 17. hydraulic gradient 2. pipe wall thickness 14.

86 Start 1 st call in simulation? Weather data from component model No Get model inputs for the TRNSYS time step Yes Get model parameters Fluid temperature and flow rate from upstream component Yes 1 st call in time step? No Compute the average temperature of the heat exchange fluid Set pond temperature equal to final value at previous time step Compute heat transfer quantities Call TRNSYS differential equation solver to compute the average pond temperature Compute new outlet fluid temperature No Tpond converged? Yes Compute heat transfer by heat exchange fluid Set the outputs END Figure 3-3. Pond model computer algorithm. .

86oC) for the 3. Results and Discussion 3.1. Simulated and actual pond average hourly temperatures are shown in Figure 3-4 for an 8-day period in July 1998 when no heat was rejected to the ponds.5-feet deep pond. Shallow ground water was not encountered at the site and therefore ground water contributions were not considered. A review of the plots in Figure 3-4 shows that the model temperatures compare favorably to the measured pond temperatures.5.87 3.55oF (0.5. Fouling of the heat exchanger pipe was also not considered.07oC) for the 2-feet deep pond and 1. This comparison allowed a validity check of the simulation of the several environmental heat transfer mechanisms occurring within the ponds. The simulated temperatures are within 3oF (1.93oF (1. in these cases.67 oC) of the observed temperatures throughout the test period. The difference between the average simulated pond temperature and the average observed pond temperature for the entire test period is 1. . the model is driven by weather data input only. Model Comparison to Experimental Results with No Heat Rejection The first step in the model verification process was to compare the model pond temperatures to measured pond temperatures during times when no heat was being rejected to the ponds. Therefore.

1 21-Jul 22-Jul 23-Jul 24-Jul 25-Jul 26-Jul 27-Jul 28-Jul 29-Jul 30-Jul Air Actual Pond Model Pond (b) 110 105 100 Temperature ( o F) 95 90 85 80 75 43.2 29.7 23.3 40.5-feet (1.8 35.6 37.7 23.88 (a) 110 105 100 Temperature ( o F) 95 90 85 80 75 43.2 29.0 32.9 70 21. Comparison of observed and simulated average pond temperatures with no heat rejection in the (a) 2-feet (0.9 Temperature ( o C) Temperature ( o C) 70 21.0 32.4 26.6 37.4 26.3 40.61 m ) deep pond and (b) 3.07 m) deep pond.8 35. .1 21-Jul 22-Jul 23-Jul 24-Jul 25-Jul 26-Jul 27-Jul 28-Jul 29-Jul 30-Jul Air Actual Pond Model Pond Figure 3-4.

The model performance was evaluated by comparing (1) the simulated to the observed return temperature of the heat exchange fluid and (2) the simulated cumulative heat rejected to the ponds to the measured water heating element and pump power input. A review of the temperature plots in Figure 3-5 shows that model fluid return temperatures compare favorably to the observed fluid return temperatures.3oC). Input data to the model consisted of weather data as described previously in addition to measured slinky heat exchanger supply water temperatures and flow rates on 6-minutely time intervals.5. 1998.7oC) and 70. however. .4oC) and 70.89 3.5oF (21.4oF (21.5feet (1.2oF (21. The error is small.11oC) error in return fluid temperature from the pond will cause only a slight difference in modeled heat pump performance. and is probably acceptable for purposes of simulating hybrid GSHP systems. These comparisons are shown in Figures 3-5 and 3-6 respectively. ground water contributions and fouling of the heat exchanger pipe were not considered.07-meter) deep pond were 69. respectively. Model Comparison to Experimental Results with Heat Rejection Heat rejection to the ponds was simulated over a 25-day period from November 12 to December 7.61meter) deep pond were 70. The average observed and modeled fluid return temperatures over the test period in the 2-feet (0.2oC). respectively. As with the previous comparisons.2. even a 2oF (1. The deeper pond has slightly larger differences between modeled and observed fluid return temperatures. and in the 3.2oF (20.

7 21.07-m) deep pond.2 16.9 18.4 18.6 20.9 23.7 21.9 23.4 23.61-m ) deep pond and (b) 3.7 7-Dec 17-Nov 22-Nov Actual Return 27-Nov 2-Dec Model Return (b) 76 75 74 73 72 71 70 69 68 67 66 65 64 63 62 12-Nov 24.3 17.8 22.4 18.1 20.5-feet (1.3 22.6 20.90 (a) 76 75 74 73 72 71 70 69 68 67 66 65 64 63 62 12-Nov 24.2 16.8 17.3 17. Comparison of observed and simulated heat exchange fluid return temperatures for the (a) 2-feet (0.0 19.1 20.4 23.2 21.8 22.3 22.2 21.9 18.8 17. Temperature (oC) Temperature ( oC) Temperature ( oF) .0 19.7 7-Dec Temperature ( oF) 17-Nov 22-Nov Actual Return 27-Nov 2-Dec Model Return Figure 3-5.

At the end of the 25-day test period. Heat Rejected (kW-hr) 1600 1400 1200 1000 800 600 400 200 0 12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec Heating Element + Pump Pow er Model Heat Rejected (b) 1800 Cumm.5-feet (1.61-m ) deep pond and (b) 3. A review of the plots in Figure 3-6 shows that the model cumulative heat rejected compares well to the measured heating element and pump power input. Heat Rejected (kW-hr) 1600 1400 1200 1000 800 600 400 200 0 12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec Heating Element + Pump Pow er Model Heat Rejected Figure 3-6. the percent difference between the cumulative simulated heat rejected and the cumulative measured heat rejected is –2.95 % for the 2-feet deep (0.07-m) deep pond.61-meter) .91 (a) 1800 Cumm. Comparison of observed and simulated heat rejected to the (a) 2-feet (0.

A simplified system schematic is shown in Figure 3-7.3.20 % for the 3.92 pond and –5. Model Application To illustrate the applicability of the model as well as the viability of using shallow ponds as supplemental heat rejecters in GSHP systems.07-meter) deep pond. Each of the component models is described briefly below. Building Loads Heat Pumps T Controller T Pond Loop Heat Exchanger Ground Loop Heat Exchanger Figure 3-7.5.5-feet (1. System schematic for the example model of a GSHP system with a shallow pond supplemental heat rejecter. . 3. These discrepancies may be due partly to heat losses from the pond supply/return pipes to the ground and to the atmosphere in the equipment building. a model of a hypothetical GSHP system was constructed in the TRNSYS modeling environment. A model and experimental uncertainty analysis is presented in Appendix A.

. A simple water-to-air heat pump model was developed for this and other GSHP system simulations. Oklahoma and is highly cooling dominated. and fluid mass flow rate. entering fluid temperature. 1999) that simulates the heat transfer over a vertical U-tube ground heat exchanger. transient twodimensional implicit finite volume model (Yavuzturk et al. The model of Yavuzturk and Spitler (1999) extends the work of Eskilson (1987) to hourly or less (“short-time step”) time intervals. The ground-loop heat exchanger model used in this application is that described by Yavuzturk and Spitler (1999) which is based partly on the work of Eskilson (1987) who developed “long time-step” (monthly) response factors for vertical ground-coupled U-tube heat exchangers. The development of the short time-step response factors are based on an analytically validated. one heat pump component model handles the heating load and a second heat pump component model handles the cooling load.2-m) deep boreholes arranged in a . The building is an actual four-story. In this application. and fluid mass flow rate. The building thermal loads are shown in Figure 3-8. The hourly building thermal loads are pre-computed using a proprietary building energy analysis program and are read from a file and passed to the heat pump subroutines. Outputs provided by the model include exiting fluid temperature. Inputs to the model include sensible and latent building loads. The model uses quadratic curve-fit equations to manufacturer’ catalog data to compute the heat of rejection in cooling s mode.93 The building is not modeled explicitly in this application.000-ft2 (4181-m2) office building located in Tulsa. power consumption. heat of absorption in heating mode. the modeled borehole field consisted of one hundred 250-feet (76. In this application. and the heat pump power consumption. 45.

The properties of each heat exchanger . indicating heat to be rejected to the GSHP system. indicating heat to be extracted from the GSHP system. Cooling loads are shown as positive values. Building thermal loads for the example building in Tulsa. When this temperature difference exceeds 9oF (5oC).909 m3/hr) of water. The total system flow was rate 270 gpm (61. flow diverters. flow is diverted to the pond such that each heat exchanger coil in the pond receives 4 gpm (0.36 m3/hr). The control strategy used to activate the circulating pump to the pond was chosen somewhat arbitrarily by using the temperature difference between the pond and the exiting fluid temperature from the heat pumps. heating loads are shown as negative values. OK.94 10 by 10 square pattern. Models for ancillary components such as pumps. t-pieces. Representative thermal properties of sedimentary rock were chosen. and the differential controller are described by SEL (1997). the circulating pump to the pond is energized and heat will be rejected to the pond. 1400 1200 Building Load (kBtu/hr) 1000 800 600 400 200 0 -200 1-Feb 1-Mar 1-Aug 1-Sep 1-Nov 1-Jan 1-Jun 1-Apr 1-Jul 1-May 1-Dec 1-Oct Cooling Load Heating Load 410 352 Building Load (kW) 293 234 176 117 59 0 -59 Time Figure 3-8. During these times of heat rejection to the pond.

0 0. Hourly input weather data for the pond model were taken from a typical meteorological year (TMY) record for Tulsa.5 3.8 32.0 Temperature ( oC) Figure 3-9.7 21. A review of the data presented in Figure 3-9 shows the advantages of using a pond supplemental heat rejecter.4 m2) pond.5 Time (years) 2.78oC) is desirable. 120 110 100 90 80 70 60 50 40 30 0. The first case was the GSHP system with no pond and the second case was the GSHP system with the pond. Entering heat pump water temperatures for the example GSHP system simulation with no pond and with a 2-feet (0. In fact. Hourly heat pump entering water temperatures are shown in Figure 3-9 for both cases. based on the results of a ground-loop heat Temperature ( oF) .95 coil in the example model are the same as those described in the experimental procedure. Oklahoma. 6000-ft2 (557.0 1. Assuming that a maximum heat pump entering water temperature of 100oF (37. The model was run for two cases for a duration of 3 years with a time step of one hour.6096-m) deep.9 43.0 2.0 w ith pond no pond 48.3 37.5 1. the system without the pond would fail during the second year of operation.2 26.6 10.1 15.

903 100.29 Heat Pump Maximum Heat Rejected Entering Fluid Temperature (oF) (oC) (kBtu) (MJ) 99.706.278. Summary of Pond Performance for Example GSHP System Simulation Year Hours ON Average Pond Temperature (oF) 74.94 2.. Using the TRNSYS model as a design tool. Such a system would likely be eliminated from consideration early on in the design phase because of excessive first cost. A more detailed system simulation could involve system life-cycle operating cost analyses.29 37.52 (oC) 23. By adding the pond supplemental heat rejecter in this example. the boreholes of a 10 by 10 square pattern would need to be approximately 400 feet (121. control strategy variations.88 2.224 1.96 exchanger sizing program (Spitler et al.961 2.498. 6000 ft2 557 (m2) pond with 50 slinky heat exchanger coils.080 2.9 m) deep to accommodate the coolingdominated loads of this building for 20 years of operation.77 24.904 1 2 3 3937 4873 5324 . the size of the pond supplemental heat rejecter was determined under the assumption that the 10 by 10 borehole field could not be feasibly deeper than 250 feet (76. The heat pump entering water temperatures for the GSHP system with the pond shown in Figure 3-9 were produced by simulating a 2-feet (0. the depth of the borehole field could be decreased by approximately 35%.618.79 76.160.452 100.2 m).75 1.65 25. A summary of pond performance is given in Table 3-1.18 37.95 37. Table 3-1.635.37 77. and design variable optimization. 1996).61-m) deep.

Environmental heat transfer mechanisms that are simulated by the model include solar radiation heat gain. conduction heat transfer to the surrounding soil or fill material. The model has been validated by comparing simulation results to experimental data. convection heat transfer to the atmosphere. Concluding Remarks and Recommendations for Future Work A design and simulation tool for modeling the performance of a shallow pond as a supplemental heat rejecter in ground-source heat pump systems has been developed. The solution scheme involves a lumped-capacitance approach and the resulting first-order differential equation describing the overall energy balance on the pond is solved numerically. The heat transfer fluid is assumed to be carried by a series of pipes in the form of bundle spools or “slinky” coils. The model accounts for several natural heat transfer mechanisms within a surface water body plus convective heat transfer due to a closed-loop heat exchanger coil. 1997) and can be coupled to other GSHP system component models for short-time step (hourly or minutely) system analyses.97 3. thermal or long-wave radiation heat transfer. exiting fluid temperature. Some outputs provided by the model include average pond temperature.6. and heat rejected to the pond. An example application has been presented to demonstrate the use of the model as well as the viability of the use of shallow ponds as supplemental heat rejecters in GSHP . and ground water discharge contributions. The model has been developed in the TRNSYS modeling environment (SEL. heat and mass transfer due to evaporation.

additional research using the simulation techniques developed in this paper is needed. Through this example. • The use of spray fountains and other aeration devices in the pond to enhance pond cooling. To more fully understand this. . would be useful. Further research is suggested in the following areas: • Optimization of the design procedure and control strategy. • Extension of the model to cover deep ponds for situations where an existing pond or lake is available. • Additional validation of the model. • The impact of pipe configuration within the pond on the overall system performance.98 systems. Hybrid ground source heat pump systems have many degrees of freedom. This research would also take into account the economic costs and benefits. using data from a working system. which we have not investigated. there are tradeoffs between the reduction in size of the ground loop heat exchanger. the size of the pond. The potential exists for significantly increasing the performance of shallow ponds used as supplemental heat rejecters in GSHP systems. it is shown that the size of ground-loop heat exchangers can be significantly decreased by incorporating a shallow pond into the GSHP system. and the control strategy.

The objective of the work presented in this chapter has been to develop a design and simulation tool for modeling the performance of a hydronic pavement heating system that can be usefully and cost-effectively integrated into a ground-source heat pump system as a supplemental heat rejecter. these types of systems can also provide a useful and cost-effective method for pavement de-icing.1. 1997) and can therefore be coupled to other GSHP system component models for short-time step (hourly or minutely) system analyses. Several combinations of so-called “hybrid GSHP systems” are possible.1. Introduction The reasons for using supplemental heat rejecters in vertical borehole GSHP systems have been described in Section 3. . GSHP system simulation results are s presented for a commercial building located in Tulsa. With additional heating equipment where applicable. The model has been validated by comparing simulation results to experimental data. As an example of the model’ applicability. The pavement heating model has been developed in the TRNSYS modeling environment (SEL. Chapter 3 has dealt with using a shallow pond as supplemental heat rejecter. This chapter deals with using a hydronic pavement heating system as a supplemental heat rejecter. Oklahoma with a hypothetical closed-loop GSHP system with and without a pavement supplemental heat rejecter. A Model for Simulating the Performance of a Pavement Heating System as a Supplemental Heat Rejecter with Closed-Loop GroundSource Heat Pump Systems 4.99 4.

Heat Transfer in Pavement Slabs Hydronically-heated pavement systems are commonly one of two types of configurations: (1) “serpentine” configuration (Figure 4-1) or (2) “slinky” configuration (Figure 4-2). Kilkis (1994). (1999). and latent heat transfer from melting snow and evaporating water. Chapman (1952). Heat fluxes at the pavement surface are due to a number of environmental interactions and include convection. Internal sources of heat are due to convection from flow of the heat transfer fluid through the pipes. Heat transfer within the slab itself is by conduction. heat fluxes are due to conduction to the ground and may or may not be . a pipe is coiled in a circular fashion such that each loop overlaps the adjacent loop. thermal (long-wave) radiation.2.100 4. ASHRAE (1995). In the slinky configuration. The slinky is typically installed in fill material near the base of the pavement slab. sensible heat transfer from precipitation. solar radiation. Heat transfer mechanisms acting upon the pavement slab are shown schematically in Figure 4-3. Pertinent concepts of heat transfer in pavement slabs have been addressed for snow melting applications by many sources including Adlam (1950). On the bottom and sides of the slab. and Ramsey et al. The pipes are embedded in the pavement material and are placed on even centers and connected with U-shaped tubing. The serpentine configuration is that commonly used in snow-melting systems.

0635 m) from the slab surface on 6.05 m) by 3 ft (0.91 m) and a thickness of 8 in. Test Slab Description and Data Collection Two hydronically-heated concrete slabs have been constructed on a test site at the Oklahoma State University and used for this study. 4. If the slab bottom is exposed.3.01905-m) and were embedded at a depth of 2.2032 m). Construction details are given by Liao and Hogue (1996).1.1651-m) centers. Heat was rejected to the slab by circulating a heated fluid through a pipe system installed in a serpentine configuration. Bridge Deck Test Section The first test slab was constructed resembling a concrete bridge deck. The pipes are made of polybutylene with a nominal diameter of ¾-inch (0. (0. The test slab is rectangular with a plan area of 10 ft (3. .3. Experimental Methods 4. 4. as in the case of a bridge or parking deck. heat transfer from the bottom surface is by convection and radiation to the surroundings. Each is discussed in the following subsections. The slab has been constructed on a steel frame and insulated on all four sides to minimize edge losses.1. (0.1.5-inch (0.5 in.101 significant as compared to the top surface heat fluxes.3.

. (a) Supply “slinky” pipes Return Supply X-sect cut Return 8-12 in. Slinky pipe configuration in a hydronically-heated pavement slab in (a) plan view and (b) cross-sectional view along the slinky centerline. typ. (b) Pavement Material Pavement Surface Pipes Fill Material Figure 4-2.3 in. typ. Pavement Top 2 . typ. Pipes Pavement Bottom Figure 4-1. Serpentine pipe configuration in a hydronically-heated pavement slab in (a) plan view and (b) cross-sectional view.102 (a) Pipes X-sect cut Supply Return (b) 6 in.

Heat transfer mechanisms in hydronically-heated pavement slabs with (a) no bottom exposure to the atmosphere and (b) bottom exposure to the atmosphere. Data used in this portion of the study were collected by Wadivkar (1997).103 (a) Incident Solar Radiation Wind Convection Sensible heat + heat of fusion to melt snow Heat of evaporation of rain and melted snow Thermal Radiation Conduction Convection due to internal pipe flow Pavement Conduction Fill (b) Incident Solar Radiation Wind Convection Sensible heat + heat of fusion to melt snow Heat of evaporation of rain and melted snow Thermal Radiation Conduction Convection due to internal pipe flow Bridge/Parking Deck Pavement Conduction Wind Convection Thermal Radiation Figure 4-3. An antifreeze heat exchange fluid was circulated through the bridge deck coupled to a closed-loop ground-source heat pump system over several night-time periods during the .

01905 m). (0. (0. Parking Lot Test Section The construction of this second test slab and data collection for this study have been conducted by researchers affiliated with the Division of Engineering Technology at the Oklahoma State University. This test slab was constructed resembling a parking lot or concrete sidewalk.1.3.1524 m). (0. (0. and the bridge surface temperature were recorded at 5-minutely intervals.2. .254 m) pitch (the separation distance between the apex of each successive loop). This test slab is rectangular with a plan area of 40 ft (12.40 m) long with a nominal diameter of ¾ in. The temperature of the concrete surface was measured by two thermistors embedded in the concrete at a depth of approximately ¼ in. The bridge deck fluid supply and return temperatures.19 m) by 4 ft (1.91 m) and a 10-in. The concrete slab was underlain by 6 in. The slinky pipe is made of HDPE plastic and is 500 feet (152. The pipe was coiled such that the resulting slinky heat exchanger is 40 ft (12. Heat was rejected to the slab by circulating heated water through a slinky heat exchanger installed at the concrete/sand fill interface. 4. Icing conditions on the concrete surface were also noted during the Wadivkar (1997) study. the fluid flow rate.0064 m) from the surface. (0.1524 m) of sand fill material.22 m) and a thickness of 6 in. The bridge surface temperature was measured using 28 thermocouples embedded in the concrete surface and the bridge fluid supply and return temperatures were measured using thermocouples embedded in the pipes.104 winter of 1996.19 m) long with a diameter of 3 ft (0.

Further details of the weather station network may be found in Elliott et al. The tests are controlled to maintain a set water temperature by heating the water if the temperature falls below a set point. relative humidity. a water heating element. which is a weather station network consisting of weather monitoring sites scattered throughout Oklahoma.2. All sensor information was recorded at time intervals of 6 minutes.3. (1994). The remainder of the heat rejection system included a flow meter.2oC) in the summer season and 75oF (23. Weather Instrumentation and Data Collection Weather data for this study were obtained from the Oklahoma Mesonet (mesoscale network). . and solar radiation. data are recorded at time intervals ranging from 3 seconds to 30 seconds and averaged over 5-minute observation intervals.105 One thermistor was placed near the slinky center between two pipes of the slinky and the other was placed near the end of the slinky above a pipe. The Stillwater station is located approximately 1 mile from the test site. and a watt transducer. Slinky supply and return water temperatures were measured by thermistors embedded in the slinky header. Data for the following parameters were obtained: wind speed. 4. Two set point temperatures were used in this study: 90oF (32. wind direction.9oC) in the winter season. air temperature. Weather data at 15-minute intervals for the Stillwater monitoring station were acquired for the time periods of interest for this study. Depending on the weather parameter.

A is the cell face area (assuming a unit depth). V is the cell volume (assuming a unit depth). The resulting general form of the explicit finite difference equation is: ∑ q" i =1 4 ( t − ∆t ) i T( m. The nodal equations are formulated using a node-centered.4.1.n ) (t − ∆t ) A = Vρc p   ∆t      (4-2) (t − ′∆t ) where. Governing Equations Transient heat transfer in the pavement slab is represented in the model in twodimensional (2-D) cross-section using the cartesian coordinate system. Model Development 4.106 4. The typical geometry and notation of the finite difference cells in the x-z cartesian coordinate plane are shown in Figure 4-4. The 2-D approach is reasonable if the cross-section is taken through the mid-section of the slab along the major direction of fluid flow in the pipe system. qi′ is the heat flux across the cell face i at the previous time step.4.n ) t − T( m . The transient 2-D heat conduction equation can be expressed as: ∂2T ∂2T 1 ∂ T + = 2 2 α ∂ t ∂ x ∂ z (4-1) This equation has been discretized using an explicit finite difference method. energy balance approach. ρ is the .

1) q4 (m .107 average density of the cell material.n)t-∆t is the nodal ′ temperature at the previous time step. and l is the distance between nodes i and (m.n ) ) l (4-3) where subscript i denotes a neighboring node (from 1 to 4). n . Finite difference cell geometry and notation. and ∆t is the time step.n).n)t is the nodal temperature at the current time step.1. n + 1) Figure 4-4. n) q2 (m + 1. n) (m. n .1) q1 (m + 1. k is the average thermal conductivity of the material between nodes i and (m.1.n). T(m. x (m . cp is the average specific heat capacity of the cell material.1) z (m.n ) = k (Ti − T( m.n) during a given time step is given by Fourier’ Law in discrete s form as: q"i → ( m . The heat flux. n + 1) ∆x (m + 1.1. . q ′ for . T(m. n + 1) (m. n . conduction into node (m. n) ∆z q3 (m .

the domain corresponds to a distance from the center-line of a pipe to half the distance to the adjacent pipe. 4. In this fully explicit method. the domain corresponds to the top of the slab and bottom of the slab or the base of the underlying fill material.4.108 The size of the time step is limited by the need to maintain stability. The Finite Difference Grid The finite difference grid used in the model is shown in Figure 4-5. the model domain was reduced to a width equivalent to one-half of the pipe spacing as shown in Figure 4-5. Because of symmetry and small temperature differences between adjacent pipe. 4. and neglecting edge effects.4. In the z direction. the stability criterion for two-dimensional problems is given by: α ∆t 1 ≤ 4 l2 (4-4) where α is the thermal diffusivity of the material within the cell and l is the nodal spacing. A uniform square nodal spacing equal to the pipe radius has been used. The temperature at each boundary node is given by .2.3. Boundary Conditions The boundaries of the model domain are treated as flux-type (Neumann) conditions as shown in Figure 4-5. In the x direction.

where qi′ represents the appropriate external flux and conduction flux from adjacent nodes. convection. Lines of symmetry on the left. thermal radiation.Pavement Bottom Surface 1/2 pipe spacing ∆x Figure 4-5.Pavement Top Surface (solar heat gain.1) ) is given by: q"( m .snow− sensible + q" rain.and right-hand boundaries are. zeroflux conditions. Shaded squares show example control volumes for different types of grid node geometries. Heat Flux Boundary .1) = q" solar + q"thermal + q"convection + q" rain. sensible heat from precipitation. heat of evaporation of precipitation) x Adiabatic Boundary z Adiabatic Boundary .snow− latent (4-5) . The model domain showing the finite-difference grid and boundary conditions. Heat flux at each top surface node ( q(m . Heat flux at the pipe surface nodes represents convection from the heat ′ ′ transfer fluid.line of symmetry Heat Flux Boundary Pipe Wall (fluid convection from internal pipe flow) Pipe radius Adiabatic Boundary ∆z Adiabatic or Convective Boundary. by definition. heat of fusion from snow melt.109 ′ the energy balance equation (Equation 4-2).

The bottom surface is treated either as an insulated surface or as a surface exposed to convective plus radiative conditions. Each of the heat flux terms is further described below.3.4. 4. and ASHRAE (1997). qthermal is the thermal radiation heat flux. The model also accepts solar radiation in the form of beam (Ib) and diffuse (Id) components. in which case I is computed from: I = I b cosθ + I d (4-7) The angle of incidence (θ) of the sun’ rays is computed at each time step from s correlations given by Spencer (1971). Solar Radiation Heat Flux ′ Solar radiation heat flux ( q ′ ) is the net solar radiation absorbed by the solar pavement surface and is given by: q" solar = αI (4-6) where I is the solar radiation incident on the pavement surface and α is the absorptivity coefficient for the pavement material.snow− sensible is the sensible heat flux from falling rain ′ rain and snow. Duffie and Beckman (1991). . The absorptivity coefficient is corrected for solar incidence angle (θ) dependence using an empirical correlation given by Duffie and Beckman (1991). q ′. and q ′.1. solar ′ ′ ′ qconvection is the convection heat flux.snow− latent is the latent heat flux from melting snow and rain evaporating/condensing water.110 ′ ′ ′ where q ′ is the solar radiation heat flux.

This model uses a linearized radiation coefficient (hr) defined as: T( m .4. The convection coefficient (hc) is a function of the Nusselt Number (Nu). If the bottom of the slab is exposed.3.3.n ) ) (4-9) 4.4. Convection Heat Flux at the Pavement Surfaces This mechanism accounts for heat transfer at the pavement top and bottom surfaces due to free and forced convection.3. which is approximated as the air temperature. where T2 represents the ground temperature in absolute units. Thermal Radiation Heat Flux This heat transfer mechanism accounts for heat flux at the pavement top surface and bottom surface (if exposed) due to thermal or long-wave radiation. For a pavement surface. σ is the Stefan-Boltzmann constant.111 4. correlations for . Several empirical formulations exist for determining the convection coefficient for different geometries. and T2 represents the sky temperature in absolute units.2. Equation 4-8 is also used to determine hr for the bottom surface. T(m.n) is the surface node temperature in absolute units.n ) + T2   hr = 4εσ    2   3 (4-8) where ε is the emissivity coefficient of the pavement material. The thermal radiation heat flux at each node ′ ′ ( qthermal ) is then computed by: q"thermal = hr (T2 − T( m. Tsky is computed from the relationship given by Bliss (1961).

the critical Reynolds Number is approximately 105. 1996).n). L is the characteristic length described for horizontal flat plates as the ratio of the area to the perimeter (Incropera and DeWitt. two empirical relations for the Nusselt number are used in the model as described by Incropera and DeWitt (1996) for forced convection over a flat plate (Equations 3-19a and . the critical Rayleigh Number is about 107.n) and the air. In external free convection flows over a horizontal flat plate. Therefore. and ν are evaluated at the film temperature. In forced convection heat transfer. β. two empirical relations for the Nusselt number are used in this model as described by Incropera and DeWitt (1996) for free convection from the upper surface of a heated plate or the lower surface of a cooled plate (Equations 3-15a and 3-15b). and the thermal properties α. which can be approximated as the average of the air and the pavement surface temperature at node (m. For external forced convection over a flat plate (i. the pavement surface).e. Therefore. Nu is usually correlated to the Reynolds (Re) and Prandtl (Pr) Numbers. In free convection heat transfer. Ra is determined as described by Equation 3-13 where ∆T refers to the temperature difference between the pavement surface at node (m. The convection coefficient (hc) for free convection can then be determined from Equation 3-16 where k is the thermal conductivity of air evaluated at the film temperature and L is the characteristic length described above. Nu is a function of the Rayleigh Number (Ra).112 a horizontal flat plate are the most applicable and therefore the convection coefficient (hc) is determined as described in Section 3.3.4.1.

The sensible heat flux due to falling rain or snow at each pavement surface node ′ ( q ′. Heat Flux Due to Rain and Snow This heat transfer mechanism includes both sensible and latent effects. the convection heat flux at each pavement surface node ( qconvection ) is computed by: q"convection = hc (Tair − T( m. The convection coefficient (hc) for forced convection can then be determined by Equation 3-16 with the characteristic length value described as the ratio of the length (parallel to the wind direction) to the perimeter.n ) ) (4-10) where Tair is the dry-bulb air temperature and hc is taken as the maximum of the free convection coefficient and the forced convection coefficient.3. This practice of choosing the larger of the free and forced convection coefficients is recommended by Duffie and Beckman (1991) and McAdams (1954) and is used in the absence of additional experimental evidence regarding combined free and forced convection. This model uses a simple mass/energy balance on water at the pavement surface to account for heat and mass transfer. 1980). 4. ′ ′ Finally.4.snow ) is given by: rain . The thermal properties of water are computed from correlations given in the Handbook of Chemistry and Physics (CRC.113 3-19b).4.

Accumulation of rain is not considered. cp is the specific heat capacity of the air evaluated at the pavement node . forming a thin film from which evaporation occurs. and Le is the Lewis number described by Equation 3-27 where α and DAB are each evaluated at the ′ ′ pavement node . wair is the humidity ratio of the ambient air.1) ) (4-12) where hd is the mass transfer coefficient.1) ) (4-11) & where m" rain.114 & q"rain .55oC).snow c p (Tair − T( m .snow is the rainfall or snowfall rate in water equivalent mass per unit time per unit area and cp is the specific heat capacity of water at the air temperature. The heat flux due to evaporation ( qevaporation ) is then given by: . The mass transfer coefficient (hd) is defined using the Chilton-Colburn analogy defined previously by Equation 3-26 where hc is the convection coefficient defined above. and w(m.air film temperature.snow = m"rain . rainfall is assumed to drain instantaneously from the pavement surface.air film temperature. Latent heat transfer is considered only if the air temperature or the slab surface temperature is above 33oF (0. This model uses the j-factor analogy to compute the mass flux of evaporating & water at each pavement surface node ( m" w ): & m" w = hd (w air − w (m.1) represents the humidity ratio of saturated air at the surface node.

55oC).ice is the conduction heat flux from the pavement surface into the ice layer and hif is the latent heat of fusion of water.115 & q"evaporation = h fg m" w (4-13) where hfg is the latent heat of vaporization and is computed from the relationship given by Irvine and Liley (1984). Although snow is a porous medium.ice hif (4-14) ′ ′ where qconduction. The mass flux of water due to melting ice & ( m"icemelt ) at the pavement surface is then given by: & m"icemelt = q" solar + q"thermal + q"convection + q" rain.snow− sensible + q"evaporation + q"conduction.1) = − m"icemelt hif (4-15) The thickness of the ice layer at the end of the time step ( l ice. The accumulation of ice at the beginning of each time step is determined from the sum of the rainfall rate and the snowfall rate if the air temperature or the slab surface temperature is below 33oF (0. Sublimation of ice is not considered. The heat flux due to melting snow and ice is determined using a mass balance on freezing precipitation that has accumulated at the pavement surface.new ) is computed by: .1) ) is given by: & q"( m . The other heat flux terms have been defined previously. If the ice thickness is greater than zero. it is treated in the model as an equivalent ice layer. the heat flux into each pavement ′ ′ surface node ( q(m .

116 & &  m" + m"icemelt l ice. 1980) for water and from correlations given by Wadivkar (1997) for an antifreeze solution.4. The thermal properties are computed at the average fluid temperature (Tfluid).old −  w  ρice   ∆t   (4-16) 4. This model has been developed to account for water or antifreeze as the heat exchange fluid.5.new = l ice. Heat Transfer Due to the Heat Exchange Fluid Heat transfer due to the heat exchange fluid is represented by heat flux at the pipe surface nodes. the previous converged value is used as an initial guess and calculation of Tfluid is iterative. ′ The heat flux due to the heat exchange fluid ( q ′ ) is computed per flow circuit fluid by: q" fluid = U pipe (T fluid − T( m. Since the outlet temperature at any current time step is not known. This temperature is computed as the average of the inlet and outlet temperatures at the given time step.n ) ) (4-17) where Upipe is the overall heat transfer coefficient for the pipe and is expressed as: U pipe = 1 1 h pipe + l k pipe (4-18) .3. The thermal properties of the fluid are computed at each time step from correlations given in the Handbook of Chemistry and Physics (CRC.

This outlet temperature is used to compute the average fluid temperature at the next iteration as described above. For turbulent flow. The value of the exponent x in Equation 3-36 is dependent upon whether the entering fluid is being heated or cooled.4 if the entering fluid is less than the slab temperature. 1996.117 where hpipe is the convection coefficient due to fluid flow through the pipe. the Dittus-Boelter relation is used to compute the Nusselt Number as described by Equation 3-36. A constant heat flux at the pipe surface is assumed. For laminar flow in the pipe (Re<2000).36 (Incropera and DeWitt. Equation 8-53).3 if the entering fluid is greater than the slab temperature and x is equal to 0. the Nusselt Number is a constant equal to 4. kpipe is the thermal conductivity of the pipe material. The outlet fluid temperature (Tout) is computed from an overall energy balance on the pipe flow circuit: Tout = T fluid − q" fluid Apipe & 2 mc p (4-19) & where A is the inside surface area of the pipe per flow circuit. x is equal to 0. The convection coefficient due to fluid flow in the pipe is determined using correlations for the Nusselt Number in flow through a horizontal cylinder. m is the mass flow rate of the heat exchange fluid per flow circuit. The convection coefficient (hfluid) is given by Equation 3-16 where k is the thermal conductivity of the heat transfer fluid and the characteristic length (L) is defined as the inner diameter of the pipe. and l pipe is the wall thickness of the pipe. . and cp is the specific heat capacity of the heat exchange fluid.

The weather component model also computes the sky temperature. The weather component model makes use of the TRNSYS psychrometric subroutine to compute moist air properties given two known state properties. A computer algorithm is shown in Figure 4-7 in the form of a flow chart. .118 The total heat transfer due to the fluid flow (qtransfer. and the solar incidence angle. the solar radiation on a horizontal surface. A companion model was also developed which manipulates any weather data needed for the pavement heating model. Computer Implementation The component configuration for the pavement heating model is shown in Figure 4-6. The two state properties are dry bulb temperature and one of wet bulb temperature.4. fluid = mc p (Tin − Tout ) N circuit (4-20) where Ncircuit is the number of flow circuits. or dew point temperature. relative humidity.fluid) is given by: & q transfer . 4.4.

layer 1 11. thermal conductivity. flag for bottom boundary cond. thermal conductivity. Pavement heating model component configuration. pipe wall thickness 18.119 air temperature sky temperature wind speed wind direction solar angle of incidence rainfall rate humidity ratio solar radiation snowfall rate inlet fluid temperature inlet mass flow rate PAVEMENT HEATING MODEL PARAMETERS: 1. layer 2 16. pipe material 17. number of flow circuits 21. absorptivity coefficient 14. thermal conductivity. emissivity coefficient 13. volumetric heat capacity. pipe diameter 8. layer 1 15. flag for fluid type (water or antifreeze) 19. pipe length per flow circuit 22. . antifreeze concentration if used 20. depth to material 1-2 interface 10. time step for finite difference method 2. volumetric heat capacity. slab surface temperature outlet fluid temperature mass flow rate heat rejected/absorbed Figure 4-6. slab width 4. layer 2 12. thickness of slab + fill 6. slab orientation from north 5. pipe spacing 7. pipe depth below surface 9. slab length 3.

120 Start 1st call in simulation? Weather data from component model No Get model inputs for the TRNSYS time step Yes Get model parameters Fluid temperature and flow rate from upstream component Yes 1st call in simulation? No Compute fluid thermal properties Set nodal temperatures to initial values Compute grid parameters Compute average fluid temperature Perform heat balance on nodal network Compute outlet fluid temperature Avg. fluid temperature converged? Yes No Compute average pavement surface temperature Compute heat transferred by fluid Set the outputs END Figure 4-7. Pavement heating model computer algorithm. .

1996. in this case.5. This comparison allowed a validity check of the simulation of the several environmental heat transfer mechanisms occurring within the pavement.5.1.2. 1996 and March 24-25. Therefore. A review of the plot in Figure 4-8 shows that the model slab surface temperatures compare favorably to the measured slab surface temperatures.121 4.5. Heat rejection to the parking lot test section was simulated over a 36-day period from November 12 to . a thin layer of ice had formed on the bridge deck prior to the heat rejection test. Simulated and actual pavement average hourly temperatures are shown for the parking lot test section in Figure 4-8 for an 8-day period in July 1998 when no heat was rejected to the pavement. 4. the model is driven by weather data input only. During the night of March 7-8. These discrepancies are generally within 3oF (1. 1996. Model Comparison to Experimental Results with Heat Rejection Heat rejection to the bridge-deck test section was simulated for the nights of March 7-8. The simulated peak daily and nightly temperatures are generally lower than the measured values. Results and Discussion 4.67oC) . Model Comparison to Experimental Results with No Heat Rejection The first step in the model verification process was to compare the model pavement slab temperatures to measured slab temperatures during times when no heat was being rejected to the slab.

the .9 46.8 35.minutely time intervals for the bridge deck field tests and 6-minutely time intervals for the parking lot field tests.122 December 18.4 51. Input data to the model consisted of weather data as described previously in addition to measured slab heat exchanger supply water temperatures and flow rates on 5. For comparison of the model results to the parking lot field test data.1 43.4 26. 1998.7 23.3 40.2 29.1 22-Jul 23-Jul 24-Jul 25-Jul 26-Jul 27-Jul 28-Jul 29-Jul 30-Jul Air Actual Slab Model Slab Figure 4-8.0 32.9 Temperature ( o C) 70 21.6 37. the model performance was evaluated accordingly. Comparison of observed and simulated slab surface temperatures for the parking lot test section with no heat rejection to the slab. Since different parameters had been measured during the bridge deck field tests and the parking lot field tests. For comparison of the model results to the bridge deck field test data. the model performance was evaluated by comparing (1) the simulated to the observed bridge deck surface temperature and (2) the simulated to the observed return temperature of the heat exchange fluid.7 48. 130 125 120 115 Temperature ( o F) 110 105 100 95 90 85 80 75 54.

respectively.9oF (2. and the average observed and modeled fluid return temperatures were 84.33oC) and 85. Another important purpose of these simulations was to verify the model’ applicability to the serpentine pipe arrangement.67oC). and the average observed and modeled fluid return temperatures were 85. One main purpose for the bridge deck simulations was to verify the model’ applicability to system s performance over relatively short time periods.2oF (-2.83oC). A review of the plots in Figure 4-9 shows that model predicted the bridge deck surface temperature and fluid return temperature quite well. respectively.27oC) and 27.0oF (29. For the night of March 7-8. s The quantity of data was limited regarding snow and ice conditions and further experiments need to be conducted to fully evaluate the model’ capability to account for s snow and ice events. For the night of March 24-25. .123 model performance was evaluated by comparing (1) the simulated to the observed return temperature of the heat exchange fluid and (2) the simulated cumulative heat rejected to the pavement slab to the measured water heating element and pump power input. These comparisons are shown in Figures 4-9 and 4-10 respectively.50oC).44oC) and 85. the average observed and modeled surface temperatures were 27. the average observed and modeled surface temperatures were 36.1oF (29.8oF (2. 1996. respectively.72oC) and 36.67oC). respectively.9oF (-2.8oF (29. 1996.7oF (29.

1996 and (b) March 24-25. Comparison of observed and simulated slab surface temperatures and heat exchange fluid return temperatures for the bridge deck test section for the nights of (a) March 7-8.7 -12.1 15.1 -6.2 26.8 32.0 4.0 4. Bridge Surface Temperature Model Fluid Return Temperature Experimental Bridge Surface Temperature Experimental Fluid Return Temperature 37. . 1996. 7. 1996 Model Avg.8 32.1 -6.6 10.4 -1.7 (b) 100 90 Temperature ( o F) 80 70 60 50 40 30 20 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Elapsed Time (hours) from 8:30 PM Mar. Bridge Surface Temperature Model Fluid Return Temperature Experimental Bridge Surface Temperature Experimental Fluid Return Temperature 37.7 21. 24.2 21.2 Temperature ( o C) Temperature ( o C) Temperature ( o F) 26.124 (a) 100 90 80 70 60 50 40 30 20 10 0 0 1 2 3 4 5 6 7 8 9 10 11 Elapsed Time (hours) from 11:30 PM Mar.1 15.2 Figure 4-9.7 -12.4 -1. 1996 Model Avg.6 10.

3 22.2 21.125 (a) 80 78 76 74 72 70 68 66 12-Nov 26.0 18.4 23.1 20.7 25. . Comparison of observed and simulated (a) heat exchange fluid return temperatures and (b) heat rejected to the parking lot test section.6 Temperature ( oC) 24.9 17-Nov 22-Nov 27-Nov Actual Return 2-Dec 7-Dec 12-Dec 17-Dec Temperature ( o F) Model Return (b) 1400 Cummulative Heat Rejected (kW-hr) 1200 1000 800 600 400 200 0 12-Nov 17-Nov 22-Nov 27-Nov 2-Dec 7-Dec 12-Dec 17-Dec Heating Element + Pump Power Model Heat Rejected Figure 4-10.

3.126 A review of the plots in Figure 4-10 shows that model also predicted the fluid return temperature and heat rejected to the parking lot test section favorably. an effective pipe spacing of 8. However.5. Model Application To illustrate the applicability of the model as well as the viability of using pavement heating systems as supplemental heat rejecters in GSHP systems. A model and experimental uncertainty analysis is presented in Appendix A. (0. respectively. These discrepancies may be due partly to heat losses from the supply/return pipes to the ground and to the atmosphere in the equipment building. A .4oF (23. 4. Explicit modeling of a slinky pipe is difficult because there are no true lines of symmetry along its cross-section. For a 10-inch pitch. the pipe spacings are not uniform. As shown in Figure 4-2b.01 %. a heuristic approach was used to determine an effective pipe spacing along the center-line of the slinky coil.8oC) and 73. the percent difference between the cumulative simulated heat rejected and the cumulative measured heat rejected is –5. this approach may not be as appropriate for other slinky configurations. Further explanation is required regarding the parking lot test section simulations. Another complicating factor is quantifying the thermal interferences induced by overlap of the pipe coils.1oF (22. The average observed and modeled fluid return temperatures over the test period were 73.0oC).4 in.21 m) was determined by taking the arithmetic average of the pipe spacing at the points where the pipe coils overlap. a model of a hypothetical GSHP system was constructed in the TRNSYS modeling environment. To circumvent these difficulties. At the end of the 36-day test period.

The building thermal loads are shown in Figure 3-8. a slinky coil installed at the interface of a 6-inch (0.1524 m) concrete slab and underlying sand fill material.3. . For this example. the same one described in Section 3. is an actual four-story. 45. System schematic for the example model of a GSHP system with a pavement heating system supplemental heat rejecter.000-ft2 (4181-m2) office building located in Tulsa. Each of the component models is described briefly below. The building. Building Loads Heat Pumps T Controller T Pavement Heating System Ground Loop Heat Exchanger Figure 4-11. That is. a pavement heating system similar to the parking lot test section described above was used.127 simplified system schematic is shown in Figure 4-11.5. Oklahoma and is highly cooling dominated. The hourly building thermal loads are pre-computed using a proprietary building energy analysis program and are read from a file and passed to the heat pump subroutines. The building is not modeled explicitly in this application.

36 m3/hr) was assumed.128 A simple water-to-air heat pump model was developed for this and other GSHP system simulations. In this application. The model uses quadratic curve-fit equations to manufacturer’ catalog data to compute the heat of rejection in cooling s mode. A total system flow rate of 270 gpm (61. Outputs provided by the model include exiting fluid temperature. Inputs to the model include sensible and latent building loads. the modeled borehole field consisted of one hundred 250-feet (76. The model of Yavuzturk and Spitler (1999) extends the work of Eskilson (1987) to hourly or less (“short-time step”) time intervals. The development of the short time-step response factors are based on an analytically validated. and the heat pump power consumption. one heat pump component model handles the heating load and a second heat pump component model handles the cooling load. The control strategy used to activate the circulating pump to the pavement was chosen somewhat arbitrarily by using the . transient twodimensional implicit finite volume model (Yavuzturk et al. flow diverters.2-m) deep boreholes arranged in a 10 by 10 square pattern. heat of absorption in heating mode. 1999) that simulates the heat transfer over a vertical U-tube ground heat exchanger. The ground-loop heat exchanger model used in this application is that described by Yavuzturk and Spitler (1999) which is based partly on the work of Eskilson (1987) who developed “long time-step” (monthly) response factors for vertical ground-coupled U-tube heat exchangers. and the differential controller are described by SEL (1997). entering fluid temperature. Ancillary components such as pumps. t-pieces.. power consumption. and fluid mass flow rate. Representative thermal properties of sedimentary rock were chosen. and fluid mass flow rate. In this application.

Hourly heat pump entering water temperatures are shown in Figure 4-12 for both cases. The first case was the GSHP system with no pavement and the second case was the GSHP system with the pavement.9 43.5 3.0 1.5 Time (years) 2. The properties of each heat exchanger coil in the example model are the same as those described in the parking lot test section experiments.0 Temperature ( C) o Figure 4-12. the circulating pump to the pavement is energized and heat will be rejected to the pavement. 120 110 100 90 80 70 60 50 40 30 0.129 temperature difference between the pavement surface and the exiting fluid temperature from the heat pumps.3 37.2 26. Temperature ( F) o . Hourly input weather data for the model were taken from a typical meteorological year (TMY) record for Tulsa.6 10. During these times.000 ft2 (2230 m2) parking lot with pavement heating.1 15. Oklahoma.5 1.0 w ith pavement no pavement 48. When this temperature difference exceeds 9oF (5oC). Entering heat pump water temperatures for the example GSHP system simulation with no pavement heating and with a 24.0 0. The model was run for two cases for a duration of 3 years with a time step of one hour.0 2. all flow is diverted to the pavement system.8 32.7 21.

Using the TRNSYS model as a design tool. 1992) ). the boreholes of a 10 by 10 square pattern would need to be approximately 400 feet (121. control strategy variations.9 m) deep to accommodate the cooling dominated loads of this building for 20 years of operation. an area per car of 320 ft2 (29. the depth of the borehole field could be decreased by approximately 35%. and design variable optimization.130 A review of the data presented in Figure 4-12 shows the advantages of using a pavement heating system as a supplemental heat rejecter. In fact. the system without the pavement system would fail during the second year of operation. The heat pump entering water temperatures for the GSHP system with the pavement heating system shown in Figure 412 were produced by simulating a 24. Assuming that a maximum heat pump entering water temperature of 100oF (37. . Such a system would be eliminated from consideration early on in the design phase because of excessive first cost. the size of the pavement supplemental heat rejecter was determined under the assumption that the 10 by 10 borehole field could not be feasibly deeper than 250 feet (76.78oC) is desirable.. A summary of the pavement system performance is given in Table 4-1. based on the results of a ground-loop heat exchanger sizing program (Spitler et al. 1996). This size of a parking lot would have a capacity of 75 cars (in parking lot design.000 ft2 (2230 m2) with 200 slinky heat exchanger coils.73 m2) will allow for access through lots (Lindeburg.2 m). A more detailed system analysis could involve system life-cycle operating cost analyses. By adding the pavement supplemental heat rejecter in this example.

71 36.605.22 3.58 17.45 3.15 (oC) 17. Summary of Pavement Heating System Performance for Example GSHP System Simulation Year Hours ON Average Surface Temperature (oF) 63.418. Concluding Remarks and Recommendations for Future Work A design and simulation tool for modeling the performance of a pavement heating system as a supplemental heat rejecter in ground-source heat pump systems has been developed. Some outputs provided by the model include .79 38.048. The heat transfer fluid is assumed to be carried by a series of pipes in a “serpentine” configuration or a “slinky” configuration.837.51 3.131 Table 4-1.277 3.759 4. and sensible and latent heat and mass transfer due to rain and snow.65 63. 1997) and can be coupled to other GSHP system component models for short-time step (hourly or minutely) system analyses. The model uses an explicit finite-difference method to solve the transient two-dimensional heat conduction equation. convection heat transfer to the atmosphere. thermal or long-wave radiation heat transfer.670.6.96 64.409 100.068 1 2 3 6245 6448 6603 4.368 3.871. The model has been developed in the TRNSYS modeling environment (SEL.86 Heat Pump Maximum Heat Rejected Entering Fluid Temperature (oF) (oC) (kBtu) (MJ) 97.695 99. The model accounts for several environmental heat transfer mechanisms plus convective heat transfer due to a closed-loop heat exchanger coil.41 37.76 17. Environmental heat transfer mechanisms that are simulated by the model include solar radiation heat gain. The model has been validated by comparing simulation results to experimental data.

The potential exists for significantly increasing the performance of pavement heating systems used as supplemental heat rejecters in GSHP systems. • Application of the model to other uses such as modeling the performance of horizontal ground-loop heat exchanger systems and snow melting systems. rain.e. using data collected under a wider range of weather conditions (i. and the amount of heat rejected to the pavement slab. would be useful. it is shown that ground-loop heat exchanger sizes can be significantly decreased by incorporating a pavement heating systems into the GSHP system. and ice conditions). • Additional validation of the model. . the exiting fluid temperature. using data from a working system. Through this example.6.132 the pavement surface temperature. snow. would be useful. An example application has been presented to demonstrate the use of the model as well as the viability of the use of pavement areas as supplemental heat rejecters in GSHP systems. Further research is suggested in the following areas: • Optimization of the design procedure and control strategy as described for the shallow pond supplemental heat rejecter in Section 3. • Additional validation of the model. particularly with regard to accounting for slinky pipe spacing in models. • The impact of the pipe configuration on the overall system performance.

Summary and Conclusions Ground-source heat pump (GSHP) systems have received considerable attention in recent decades as an alternative energy source for residential and commercial space heating and cooling applications. there are inherent inabilities to make direct observations in the subsurface environment with respect to both space and time. a cooling tower). a GSHP system may either use the earth. Second. a considerable amount of research in the past decade has been geared toward optimizing the design and performance of GSHP systems and this study is part of those efforts. Hybrid GSHP systems use combinations of these or a combination of a GSHP system with conventional equipment (i. The challenges associated with the design of these systems originate from the fact that they present a unique type of heat transfer problem. Depending on the configuration. ground water. GSHP systems offer proven advantages over conventional heating and cooling systems. groundsource heat pump systems. This study has dealt with the modeling of vertical closed-loop and hybrid. and overall operating costs. Consequently. or surface water as a heat source and/or sink. maintenance costs.133 5. heat transfer within the subsurface environment can be highly transient. (2) to develop a design and simulation tool for modeling the performance of a shallow pond as a supplemental heat rejecter with closedloop GSHP systems. specifically with respect to efficiency. The objectives of this study were threefold: (1) to examine the effects of groundwater flow on closed-loop GSHP systems. and (3) to develop a design and simulation tool for modeling the . First.e.

The finite-element numerical ground-water flow and heat transport model was also used to examine the effect of ground-water flow on in-situ thermal conductivity test results. it appears that it is only in geologic material with high hydraulic conductivities. As expected. in all cases of ground-water flow.134 performance of a hydronic pavement heating system as a supplemental heat rejecter with closed-loop GSHP systems. This was done by numerically simulating test conditions around a single borehole under different flow conditions. A finite-element numerical ground-water flow and heat transport model was used to simulate and observe the effects of ground-water flow on the heat transfer from a single U-tube closed-loop ground heat exchanger in various geologic materials. Chapter 2 of this thesis has presented a preliminary assessment of the effects of ground-water flow on closed-loop ground-source heat pump systems. A simple but useful method of assessing the relative importance of heat conduction in the ground versus heat advection by moving ground water was demonstrated through the use of the dimensionless Peclet number. A compilation of “typical” hydraulic and thermal properties of soils and rocks was used in the study. From those simulations. . such as coarse-grained soils (sands and gravels) and in rocks exhibiting secondary porosities such fractures and solution channels. that ground-water flow could be expected to have a significant effect on closed-loop heat exchanger performance.

However. . temperatures were found to rise above the design criteria. in some cases at very high ground-water flow rates. These include: • Identification of suitable numerical and/or analytical models that include ground- water flow and could be used to analyze in-situ test data. considerable progress has been made in developing both in-situ test methods and design procedures for borehole field design for situations where there is no ground-water flow.135 these values were artificially high. Results from one week test data have been shown to be no more reliable than data from 50-hour tests when significant ground-water flow is present. it appears difficult to adapt results from current design and in-situ measurement methods to fully account for ground-water flow conditions. Even the presence of moderate ground-water flows had the effect of removing the year-by-year increase in ground temperature found in systems where there is no groundwater flow. From this preliminary assessment of the effects of ground-water flow. Over the last decade. Research would be required in a number of areas before the same progress could be made to deal with the situations of ground-water flow. The borehole fields designed using conventional methods and the derived effective thermal conductivities were generally over-designed. The finite-element numerical ground water flow and heat transport model was also used to simulate the 10-year performance of borehole fields designed from application of conventional design procedures using the derived thermal conductivity data.

convection heat transfer to the atmosphere. or adaptations to existing methods. thermal or long-wave radiation heat transfer. that could be used for closed-loop ground heat exchanger design. Environmental heat transfer mechanisms that are simulated by the model include solar radiation heat gain. • Development of design guidelines and software tools that could be used by practicing engineers for in-situ testing and system design tasks in situations of significant ground-water flow. conduction heat transfer to the surrounding soil or fill material. The model has been validated by comparing simulation results to experimental data. Chapter 3 of this thesis has described the development and validation of a model for simulating the performance of a shallow pond as a supplemental heat rejecter with closed-loop ground-source heat pump systems. The model has been developed in the TRNSYS modeling environment and can be coupled to other GSHP system component models for short-time step (hourly or minutely) system analyses. heat and mass transfer due to evaporation. The model accounts for several natural heat transfer mechanisms within a surface water body plus convective heat transfer due to a closed-loop heat exchanger coil. • Identification of suitable design methods. and ground water discharge contributions. The heat transfer fluid is assumed to be carried by a series of pipes in the form of bundle spools or “slinky” coils. The solution scheme involves a lumped-capacitance approach .136 • Experimental investigation of potential in-situ test data analysis methods at sites with significant ground-water flow.

Further research is suggested in the following areas: • Optimization of the design procedure and control strategy. • Additional validation of the model. which we have not investigated. Through this example. and the control strategy. • Extension of the model to cover deep ponds for situations where an existing pond or lake is available. there are tradeoffs between the reduction in size of the ground loop heat exchanger.137 and the resulting first-order differential equation describing the overall energy balance on the pond is solved numerically. the size of the pond. To more fully understand this. This research would also take into account the economic costs and benefits. exiting fluid temperature. would be useful. An example application has been presented to demonstrate the use of the model as well as the viability of the use of shallow ponds as supplemental heat rejecters in GSHP systems. it is shown that the size of ground-loop heat exchangers can be significantly decreased by incorporating a shallow pond into the GSHP system. . and heat rejected to the pond. using data from a working system. additional research using the simulation techniques developed in this paper is needed. The potential exists for significantly increasing the performance of shallow ponds used as supplemental heat rejecters in GSHP systems. Some outputs provided by the model include average pond temperature. Hybrid ground source heat pump systems have many degrees of freedom.

the exiting fluid temperature. Environmental heat transfer mechanisms that are simulated by the model include solar radiation heat gain. convection heat transfer to the atmosphere. and sensible and latent heat and mass transfer due to rain and snow. thermal or long-wave radiation heat transfer. The model has been developed in the TRNSYS modeling environment and can be coupled to other GSHP system component models for short-time step (hourly or minutely) system analyses. The model has been validated by comparing simulation results to experimental data. The model accounts for several environmental heat transfer mechanisms plus convective heat transfer due to a closed-loop heat exchanger coil. Chapter 4 of this thesis has described the development and validation of a model for simulating the performance of a pavement heating system as a supplemental heat rejecter with closed-loop ground-source heat pump systems.138 • The use of spray fountains and other aeration devices in the pond to enhance pond cooling. and the amount of heat rejected to the pavement slab. The heat transfer fluid is assumed to be carried by a series of pipes in a “serpentine” configuration or a “slinky” configuration. • The impact of pipe configuration within the pond on the overall system performance. Some outputs provided by the model include the pavement surface temperature. The model uses the finite-difference method to solve the transient twodimensional heat conduction equation. .

rain. using data collected under a wider range of weather conditions (i. Through this example.e. particularly with regard to accounting for slinky pipe spacing in models.139 An example application has been presented to demonstrate the use of the model as well as the viability of the use of pavement areas as supplemental heat rejecters in GSHP systems. and ice conditions). would be useful. it is shown that ground-loop heat exchanger sizes can be significantly decreased by incorporating a pavement heating systems into the GSHP system. Further research is suggested in the following areas: • Optimization of the design procedure and control strategy as described for the shallow pond supplemental heat rejecter. . • The impact of the pipe configuration on the overall system performance. would be useful. using data from a working system. • Additional validation of the model. snow. • Application of the model to other uses such as modeling the performance of horizontal ground-loop heat exchanger systems and snow melting systems. • Additional validation of the model. The potential exists for significantly increasing the performance of pavement heating systems used as supplemental heat rejecters in GSHP systems.

1997. 1946. Inc. 1975. BLAST 1986. 96-102.K.N. Chu. International Journal of Heat and Mass Transfer. Heating and Ventilating. American Society of Heating. Bear. Cleveland. Atlanta. Cantrell. GA.. Atlanta. American Society of Heating. T. ASHRAE Handbook. Chemical Rubber Company (CRC). 1980. HVAC Applications. GA.H. Development of an In-Situ System for Measuring Ground Thermal Properties. Inc. S. Spitler.. H. D. Austin.J. ASHRAE Handbook. OH. U.F. GA. BLAST Support Office. Dake.W. Refrigeration and Air-Conditioning Engineers. 2000. Handbook of Chemistry and Physics.P. 1972. Bliss. C. Yavuzturk. Design of Snow Melting Systems. Conduction of Heat in Solids. C. ASHRAE. 5(2): 484-495.W. Carslaw. Water Resources Research. Inc. Chapman. Atlanta. Inc. and W.M. BLAST (Building Loads and System Thermodynamics). W. 1961. and J. R. III. 1952. Wepfer. Fundamantals. 1995. J.140 REFERENCES Adlam. The Industrial Press. 61st Edition.. UrbanaChampaign: University of Illinois.M. J. and H. April. and D. Solar Energy. ASHRAE. Submitted for publication to ASHRAE Transactions.: Claremore Press. pp. 5(103). Thermal Stratification in Lakes: Analytical and Laboratory Studies. and J. Oxford. Shallow Ponds for Dissipation of Building Heat: A Case Study. 1969.K. Snow Melting. . Refrigeration and Air-Conditioning Engineers.R. New York.. 1995b. CRC Press. ASHRAE Transactions 90(1): 239-246. NY. Refrigeration and Air-Conditioning Engineers. S. 18: 1049-1053. Commercial/Institutional Ground-Source Heat Pump Engineering Manual. New York: Dover Publications. W. Correlating Equations for Laminar and Turbulent Free Convection from a Horizontal Cylinder. Churchill. American Society of Heating. Atmospheric Radiation Near the Surface of the Ground.S. 1950. 1984. ASHRAE. J. Harleman. Jaeger.A. Dynamics of Fluids in Porous Media.

Academic Press. Solar Engineering of Thermal Processes. 1979. OH. Chicester.141 Domenico. Schwartz. March 16-17. Hellstrom. 1972. Sweden: University of Lund. Elliot. Beckman. 1988. (Bose. 1984. Nordell. McGraw-Hill Book Company. S. R. Nielsen. Applied Hydrogeology. Grant. 1996. Applied Engineering in Agriculture. and J. 2nd Edition. 1990. K. 1991. A.V.F. 2nd Edition. TED – A Mobile Equipment for Thermal Response Test. and C.E. P. Heat Transfer. Sweden. Duffie. Gehlin. Cherry. I. J. Groundwater. Doctoral Thesis. New York. J. 1986. Sweden. Sha. Ground Heat Storage.V. Bixley. Freeze. Inc. Editor) Design and Installations Standards. Lund. Configuration Decisions for an Automated Weather Station Network. Paul Minnesota: Johnson Filtration Systems. Hull. s Eskilson. M. Department of Mathematical Physics. W. 1991.P. Oklahoma: International Ground Source Heat Pump Association.T. Physical and Chemical Hydrogeology. John Wiley and Sons. Department of Mathematical Physics. A.A. and W. Thermal Analyses of Duct Storage Systems. Sharp. National Water Well Association. F. Eklof. Lulea University of Technology. Geothermal Reservoir Engineering.. Columbus. and P. Holman. M. . Kamal.A. Liu. 33(1): 25-33.P. Earth-Coupled Heat Transfer. Solar Energy. Driscoll. D. Donaldson. 1987. Thermal Response Tests of Boreholes – Results from In Situ Measurements. and B. Lund.G. and S. W.. P.R. J. C. 1994. E. Fetter. 1991. Transaction of The Geothermal Project at Richard Stockton College Conference. New Jersey: Prentice Hall. Toronto. St. Englewood Cliffs.. Dependence of Ground Heat Losses Upon Solar Pond Size and Perimeter Insulation – Calculated and Experimental Results.. 3rd Edition. Brock. J. F. Groundwater and Wells. 10(1): 45-51. 1982.A. R. Singapore: John Wiley & Sons. Inc. Hart. and R. and F. W. G. Stone. J. Master’ Thesis 1996:198E. and S. G. Couvillion. Stillwater. University of Lund. 1998. Ohio: Merrill Publishing Co. 1986. Dublin.L. Brisbane. IGSHPA.A. 1998. Thermal Analysis of Heat Extraction Boreholes. C.L.L.. Gehlin.

W. Columbia Lava Plateau (Chapter 5). Thomson. F. F. McAdams. O-2.Mason. Introduction to Heat Transfer.R.. J.B. T.N.142 Incropera.. 89-103. Hogue. Kishore. 1882. Heat Transmission. Atlanta. Lindeburg. Thermal Stress Predictions for Geothermally Heated Bridge Decks. Seaber.P.A. Simulation and experimental verification of vertical groundcoupled heat pump systems. pp. Vaccaro. J. Jr. and K. 1996. S. Vol. 423-433. and P. New York: McGraw-Hill. May. 6th Edition.P. Inc.. 1988. Zobel. and J. Geological. In: The Geology of North America. 1996. Civil Engineering Reference Manual. and E. Hydrogeology. McGraw-Hill Book Company. 1972. Professional Publications.S. L. 1997.. and V. Oklahoma: Oklahoma State University. W. Rosenshein. Inc. dissertation. Gaseous Diffusion Coefficients. II. I. C. Proceedings of the American Society of Civil Engineers. Boulder. Academic Press. Colorado: The Geological Society of America. Journal of the Power Division. 1984. Inc. American Society of Heating. S. Ingersoll. Liley. Mathematical and Physical Papers.R. O. Sir W.D. Ingersoll. Inc. ASHRAE Transactions. and D. R. . Ground-Source Heat Pumps. 3rd Edition. Kavanaugh. A Practical Collector Efficiency Equation for Nonconvecting Solar Ponds.H. Rafferty. Ph. J. Belmont. Stillwater.P. Kavanaugh. 3rd Edition.E... P. Marrero. Solar Energy. 1962. John Wiley & Sons. Liao. Editors) Region 2. 99(P01). (Back.J. C. CA. Irvine.D. 1954. V. and A. and P. The Dissipation of Excess Heat from Water Systems. GA. 100(1):pp. 1: 3-118. Steam and Gas Tables with Computer Equations. Refrigeration and Air-Conditioning Engineers. Jobson. 33(5): 391-395. Heat Conduction with Engineering. 1994.V. p. T.R. 1954.41 ff. Prepared for the State of Oklahoma Department of Transportation. 1992. Joshi. H.F. M. Kilkis. G. DeWitt. and T.B. 1984. Design of Geothermal Systems for Commercial and Institutional Buildings. Design of Embedded Snow Melting Systems: Part 1. Heat Requirements – An Overall Assessment and Recommendations. and Other Applications. Kelvin. Journal of Physical and Chemical Reference Data.. 1984. Lindholm.

492. 91(2B):12161224. A Transient Systems Simulation Program. Prentice Hall. and S. Mei. 1997. H. 105(1). Shonder. 32(6): 771-778. s .. M. M. ASHRAE Transactions 102(2):12-21.M. 1962. B. 1995. J. J. Journal of the Power Division.D. Ramsey. Updated Design Guidelines for Snow Melting Systems.H. Kavanaugh. and W. Mills. July.A. Beckman. J. ASHRAE Transactions. Proceedings of the American Society of Civil Engineers.J.A. Prediction of Temperatures in Rivers. Military Engineer.. 88(P02).2.F. and C. Hewett. Emerson.. A.C. 1997. Solar Ponds – Alternate Energy Systems. Basic Heat and Mass Transfer. South Dakota State University. ASHRAE Transactions.M. 105(1). P. User’ Manual.. C.143 Mercer. Modeling the Performance of a Solar Pond as a Source of Thermal Energy. D. Raphael. 1984. National Water Well Association. and J. ASHRAE Transactions. ASHRAE Transactions. 1986. Ground-Water Modeling. O’ Neal. Petersen. A. V. Rottmayer. Bachu. C.P. Muraya. and J. J. Kuehn. Journal of the Society of American Military Engineers. Mitchell. 1984. Benedict. V. No. SD. pp.. T. New Approach for Analysis of Ground-Coil Design for Applied Heat Pump Systems. Determining Effective Soil Formation Thermal Properties From Field Data Using a Parameter Estimation Technique. J. Brookings. Rubin. A. Remund. Version 14. S. Simulation of a Single Vertical U-Tube Ground Heat Exchanger in an Infinite Medium. 1996.L. Beck. 1998. and S.. N. Pezent. and C. T.R. ASHRAE Transactions 96(1): 574-582. Faust. Solar Energy Laboratory (SEL). K. W. Personal Communication. 1999. TRNSYS. 103(2):651-658. W. 1990.. Newell. March-April 1984: 110-113. 1985. and S. Thermal Interference of Adjacent Legs in a Vertical U-Tube Heat Exchanger for a Ground-Coupled Heat Pump. University of Wisconsin-Madison. Northern Geothermal Support Center.W. Heffington. Solar Energy. 1999.W. 157-181. Development and Verification of a Thermal Model of Lakes Used With Water Source Heat Pumps.

C. OK. Seaber. Fleming. Hydrogeology. 1998. O. R. and J. Oklahoma State University. Editors) Glacial Deposits (Chapter 35). ASHRAE Transactions. 105(2).W. Spitler. R. M. C. Marshall. A Short Time Step Response Factor Model for Vertical Ground Loop Heat Exchangers. and E. J. Cincinnati. Introduction to Groundwater Modeling – Finite Difference and Finite Element Methods. Mickelson. B. Colorado: The Geological Society of America. Vatnaskil Consulting Engineers. OK. BIOSCREEN. Srinivasan. and P. J. 2(5). User’ Manual. Rosenshein. In: The Geology of North America. Seaber. Colorado: The Geological Society of America. Weeks. United States Environmental Protection Agency.H. Anderson. Yavuzturk.. 1982. J. 1996. ASHRAE Transactions. and A. Freeman and Company. Stillwater. D. (Back. Unpublished Masters Thesis. Rees. Spitler. and S. and P. Guha. and M. Vol.A. 1988. A. Solar Energy. Rosenshein.. Office of Research and Development. 1999. O-2. Wadivkar. Users Guide of GLHEPRO. O-2. Oklahoma State University. Boulder. W. Editors) Region 17. Delahoussaye. 1987. J. D. D. Stillwater. Wang. D. and M. Ohio: s National Risk Management Research Laboratory. R. 1996. S.. J. AQUA3D Groundwater Flow and Contaminant Transport Model. Gutentag.F. Yavuzturk. C. Natural Attenuation Decision Support System. Fourier Series Representation of the Position of the Sun. Hydrogeology. W. W. B. Stephenson. Search.. Manicham. Vol. Spitler. A Transient two-dimensional Finite Volume Model for the Simulation of Vertical U-Tube Ground Heat Exchangers. 1997.. Inc.D. H. Boulder. S. J. 39(4): 361-367.144 Spencer. School of Mechanical and Aerospace Engineering.. Reykjavik. 105(2). and D. H. J.. 1999. Inc.. 1971.P. 1988. High Plains (Chapter 20). J. The Effect of Reflectivity on The Performance of a Solar Pond. B. New York. (Back. . Iceland: Vatnaskil Consulting Engineers. An Experimental and Numerical Study of the Thermal Properties of a Bridge Deck De-Icing System. In: The Geology of North America..

145 APPENDIX A Model and Experimental Uncertainty Analysis .

including TRNSYS model parameters as well as TRNSYS model inputs such as weather factors. The pond and pavement model uncertainty is quantified using influence coefficients as described by Spitler et al. . (1989).146 This appendix presents an analysis of parameter uncertainty and experimental uncertainty regarding the validation and application of the pond and pavement models. for the purposes of this study. is approximated as: Influence Coefficient = ∆Result ∆Parameter (A-1) where Result is the cumulative heat rejected and Parameter is a variable with some uncertainty. The term Parameter should not be confused with TRNSYS model parameters. respectively. and fluid flow rate. and are discussed below. Parameter refers to any variable with a significant uncertainty. An influence coefficient is the partial derivative of a simulation result with respect to a parameter which. Model Uncertainty This section presents a sensitivity analysis to quantify the uncertainty in the final model results caused by uncertainties in the estimates of input and parameter values used in the models. fluid supply temperature. The results of this analysis are shown in Tables A-1 and A-2 for the pond and the pavement models. Here.

33 15.82% 0.2 0. 7.05 (b) (b) 1486 1508 6.13% 6.00 0.00 -252.25% 0 0 (b) (b) 0.9 1473 1495 185.05 0.5 1554 1578 2.9 0.13 -679.00 13.56% kpipe (W/m-oC) Horizontal Slinky Vertical Slinky ksoil (W/m.00 36.12 (7) (7) 0.35 0.71 0.10 (6) (6) 2.52% 2.02 0.53 (%) 2.06 (3) (3) 0.measurement error given by Elliot et al.05 0.05 0.05 (6) (6) 1.147 TABLE A-1 Pond Model Parameter Uncertainty Analysis Modeled cumulative heat rejected from Nov.0 (b) (b) 1284 1305 -100.84% 0 0 (b) (b) 0.25% 0.43 38.50 2.025 (7) (7) 0.30 35.90% 0.00 12.5 2.66 7.391 (2) 0.an addition factor is used to vary the hourly input value of this item for this uncertainty analysis (1) .10 0.00 70.43 35.49% 3.based on standard error of calibration curves .estimated value (2) .00 0.35% 1 1 (a) (a) 0.93% 7.85 642.17 0.25% 1 1 (a) (a) 0.a multiplication factor is used to vary the hourly input value of this item for this uncertainty analysis (b) .20 2.0 2.97 (5) (5) 0. (1994) (7) .025 0.0 (a) (a) 1847 1880 361.12 0.02 (1) (1) 0.0 1510 1535 -5.83 57.71 3.47% 1 1 (a) (a) 0.35 (6) (6) 2.38% 2.5 2.based on in-situ thermal conductivity test data (5) .5 7.00 372.80 33.29 3.60 14.5 (4) (4) 3.0 (a) (a) 1242 1256 -244.80% 0 0 (b) (b) 0.99% 0.97 0.10 37.60 0.90 -477.391 (2) 0.based on data given by Mills (1995) (4) .0 2.05 (a) (a) 1452 1474 -674.2 (b) (b) 1400 1413 -431.05 (6) (6) 2.27% 2.5 (4) (4) 0.0 1520 1543 68. 1998 to Dec.20 12.33 -6.0 3.35% NOTES: (a) .5 0.92% 0.05 1.00 13.01% 0.5 (4) (4) 2.0 2.71 3.12.86 -101.2 (1) (1) 3.0 3.71 185.71 33.C) Horizontal Slinky Vertical Slinky Depth to constant sink (m) Horizontal Slinky Vertical Slinky ε of water surface (--) Horizontal Slinky Vertical Slinky Windspeed (m/s) Horizontal Slinky Vertical Slinky Solar radiation Horizontal Slinky Vertical Slinky Air temperature ( C) Horizontal Slinky Vertical Slinky Relative humidity (%) Horizontal Slinky Vertical Slinky Supply fluid temperature ( C) Horizontal Slinky Vertical Slinky Supply fluid flow rate (gpm) Horizontal Slinky Vertical Slinky Total Uncertainty in Results Horizontal Slinky Vertical Slinky o o o 0. 1998: Horizontal Slinky Vertical Slinky 1486 1508 kW-hr kW-hr Item Adjusted Value Used in Final Analysis Uncertainty in Parameter Perturbed Model Value ∆(result) Model Perturbed Uncertainty ∆(parameter) Model in Parameter Uncertainty Result Result (kW-hr) (kW-hr) 623.20 37.2 0.43% 2.04 0.value given by Incropera and DeWitt (1996) (6) .43 51.01% 6.06 0.99 2.value given by pipe manufacturer Phillips Driscopipe (3) .18 0.

148

TABLE A-2 Pavement Model Parameter Uncertainty Analysis
Modeled cummulative heat rejected from Nov.12, 1998 to Dec. 19, 1998: Horizontal Slinky 1255 kW-hr

Item Adjusted

Value Used in Final Analysis

Uncertainty Perturbed Model in Value Parameter

∆(result) Perturbed Uncertainty Model Model ∆(parameter) in Parameter Uncertainty Result Result (kW-hr) (kW-hr) 348.62 20.92 (%) 1.67%

kpipe (W/m- C) Slinky test section kconcrete (W/m- C) Slinky test section ksoil (W/m- C) Slinky test section α of concrete (--) Slinky test section ε of concrete (--) Slinky test section Windspeed (m/s) Slinky test section Air temperature ( C) Slinky test section Relative humidity (%) Slinky test section Supply fluid temperature ( C) Slinky test section Supply fluid flow rate (gpm) Slinky test section Total Uncertainty in Results Slinky test section
o o o o

o

0.391

(2)

0.06

(3)

0.5

1293

1.663

(4)

0.2

(1)

2.9

1314

47.70

9.54

0.76%

0.4

(5)

0.2

(1)

1.0

1146

-181.67

36.33

2.90%

0.4

(6)

0.1

(6)

0.6

1135

-600.00

60.00

4.78%

0.9

(6)

0.05

(6)

0.8

1192

630.00

31.50

2.51%

1

(a)

0.10

(7)

2.0

(a)

1327

72.00

7.20

0.57%

0

(b)

0.35

(7)

2.0

(b)

1092

-81.26

28.44

2.27%

1

(a)

0.05

(7)

1.05

(a)

1250

-103.99

5.20

0.41%

0

(b)

0.12

(8)

0.2

(b)

1089

-831.18

99.74

7.95%

0

(b) 0.025

(8)

0.05

(b)

1256

16.48

0.41

0.03%

10.47%

NOTES: (a) - a multiplication factor is used to vary the hourly input value of this item for this uncertainty analysis (b) - an addition factor is used to vary the hourly input value of this item for this uncertainty analysis (1) - estimated value (2) - value given by pipe manufacturer Phillips Driscopipe (3) - based on data given by Mills (1995) (4) - based on data given by Tinker and Cabrera (1992) (5) - value given by Spitler et al. (1996) for light soil (6) - value given by ASHRAE (1997) light-colored surfaces (7) - measurement error given by Elliot et al. (1994) (8) - based on standard error of calibration curves

149

The influence coefficient for each parameter of interest is determined by running the model with a “perturbed” value of the parameter. The Uncertainty in Result, with respect to the parameter, is then determined by:

Uncertainty in Result = (Influence Coefficient )x(Uncertainty in Parameter)

(A-2)

where Uncertainty in Parameter is the estimated parameter error (for example, ± 0.5 W/m-oC). The non-dimensional Model Parameter Uncertainty is then determined by:

Model Parameter Uncertainty =

Uncertainty in Result Result

(A-3)

The errors introduced to the model result by each parameter are assumed to be independent of each other. Therefore, the Total Uncertainty in Result is given by:

Total Uncertainty in Result =

∑ (Model Parameter Uncertainty)
i =1

n

2

i

(A-4)

where n is the number of parameters considered in the analysis.

The parameters investigated in this uncertainty analysis can be divided into two groups: (1) TRNSYS model parameters and (2) TRNSYS model inputs. For the pond model, these items consisted of: (1) (a) pipe thermal conductivity (kpipe), (b) soil thermal conductivity (ksoil), (c) depth to constant sink temperature, and (d) emissivity of the water

150

surface (ε) and (2) (a) wind speed, (b) solar radiation, (c) air temperature, (d) relative humidity, (e) supply fluid temperature, and (f) supply fluid flow rate. For the pavement model, these items consisted of: (1) (a) pipe thermal conductivity (kpipe), (b) concrete thermal conductivity (kconcrete), (c) soil thermal conductivity (ksoil), (d) absorptivity of the concrete surface (α) and (d) emissivity of the concrete surface (ε) and (2) (a) wind speed, (b) air temperature, (c) relative humidity, (d) supply fluid temperature, and (e) supply fluid flow rate. The model values used in the final analysis and the uncertainty in each value were taken from the sources shown in Tables A-1 and A-2.

For the pond model, the supply fluid temperature has the greatest effect on the model uncertainty, on the order of 3.5%. Four other parameters have lesser uncertainties on the model results, on the order of 2.5%; these are kpipe, wind speed, air temperature, and relative humidity. Of still lesser significance are ksoil, the depth to constant sink temperature, and the solar radiation, which produce model uncertainties on the order of 0.9%. The emissivity of the water surface has a relatively insignificant uncertainty of 0.25%. The fluid flow rate has an insignificant effect on the model uncertainty, contributing only 0.01% to the total.

For the pavement model, the supply fluid temperature also has the greatest effect on the model uncertainty, on the order of 8%. The absorptivity of the concrete surface possesses the next highest uncertainty, on the order of 5%. Air temperature, ksoil, and ε of the concrete surface all have uncertainties on the order of 2.5%. kpipe has an uncertainty of 1.7% and wind speed and relative humidity have relatively low uncertainties on the

the fluid flow rate has an insignificant effect on the model uncertainty. estimates of the TRNSYS model parameters are likely no better than those that could be made by a typical user. the total uncertainty in the results of the pavement model is ±10. For this study. Based on the results shown in Table A-2.47% for the slinky case.151 order of 0. the total uncertainty in the results of the pond model are ±6. A user performing some type of system design or simulation must use typical or synthetic weather data based on historical observations for the location of interest. the longer the simulation time (10-20 years). similar uncertainties can be expected for those items. and therefore. the TRNSYS model inputs cannot be known exactly by a typical user because they consist of weather factors and system fluid temperature and flow rate.5%.35% for the vertical slinky case. contributing only 0.13% for the horizontal slinky case and ±6. Based on the results shown in Table A-1. the greater the probability is that average actual weather conditions approach the “typical weather . The results of this uncertainty analysis can be used to identify items of concern to one who may use the model(s) for the purposes of GSHP system design or system simulation. However. However. As with the pond model.03% to the total. The use of “typical weather data” is one inherent difficulty in building simulation studies because the it is obviously impossible to quantify uncertainties in future weather conditions.

Other model inputs include the supply fluid temperature and flow rate which will be provided either as a constant value or by another model. The main sources of error in the measurements of heat rejection to the ponds and to the concrete slab are identified as: (1) calibration error in the watt meter used for power measurements and (2) heat losses or gains to the ground from the pipes between the heat rejecter and the instrumentation building. model results are much more sensitive to the supply fluid temperature.152 conditions”. the conservative user may want to use weather data with more extreme values. Error in the watt meter measurements was determined from the calibration procedure. Heat losses/gains to the ground from the supply and return pipes were . Experimental Uncertainty This section presents an analysis of the experimental errors contributing to the measurement of heat rejected to the ponds and to the concrete slab. and a user should therefore pay more attention to this item. The results of this analysis are shown in Tables A-3 and are discussed below. if available. From the results of this uncertainty analysis. For shorter-term simulations (up to 5 years). The purpose of this analysis is to quantify the error in the experimental results that were used to validate the pond and pavement models.

estimated using the pavement model . The cumulative heat lost through the supply/return pipes was then estimated by considering the distances from the instrumentation building to each of the supplemental heat rejecters (62 ft (18.00% NOTES: (a) .50% 20. 1998.5 ft (0. TABLE A-3 Heat Rejection Experimental Uncertainty Analysis Item Value Experimental Uncertainty Error in Power Measurement (a) Estimated Cumulative Heat Losses to Ground: (b) Pond with horizontal slinky Pond with vertical slinky Slab with slinky Total Experimental Uncertainty Pond with horizontal slinky Pond with vertical slinky Slab with slinky -- 0.153 estimated using the pavement model. The model was used to determine the heat losses per foot of pipe for the period of November 12.24% 0. and 12 ft (3. 1998 to December 19.45% 2.36% 2.48 m).25 33. 37 ft (11.9oC) and constant fluid supply flow rate of 4 gpm (908 kg/hr). 1998 with actual weather conditions and assuming a constant fluid supply temperature of 75oF (23.73 W/m-oC) and the average pipe burial depth was taken as 1.83 10. 1998 to December 19.based on instrument calibration precision (b) .7 m) for the concrete slab). The model was run from October 1.87% 1. The thermal conductivity of the ground and the pipe backfill material were taken as 1.9 m) for the pond with the vertical slinky.92 kW-hr kW-hr kW-hr 1.includes heat lost through supply and return headers.30% 1.3 m) for the pond with the horizontal slinky. which is the period of interest for the model validation.0 Btu/hr-ft-oF (1.

36. and 1.2 Model Cumulative Heat Rejected (kW-hr) 1486 +/.91.2 1591 +/.8 1255 +/. the total uncertainty in the cumulative heat rejection is 1.131. TABLE A-4 Summary of Experimental and Model Cumulative Heat Rejected Test Configuration Experimental Cumulative Heat Rejected (kW-hr) 1532 +/.13.154 Based on the results shown in Table A-3. 2.00% for the concrete slab.45% for the pond with the horizontal slinky. Summary In summary. These errors are considered acceptable.6 1321 +/.1 1508 +/.30% for the pond with the vertical slinky. The second conclusion implies that the uncertainty prediction may be over-conservative. . the experimentally-determined cumulative heat rejection and the model-predicted cumulative heat rejection for the test period are compared in Table A-4.95.22.4 Pond with horizontal slinky Pond with vertical slinky Slab with slinky Two conclusions may be drawn from the above comparison: (1) the model predictions match the experimental results within the bands of estimated uncertainty and (2) the model predictions match the experiment better than would be expected by the uncertainty analysis.

Introduction to Heat Transfer.REFERENCES ASHRAE. Fundamantals. 1989. A. DeWitt. Tinker. A Primer on the Use of Influence Coefficients in Building Simulation. Applied Engineering in Agriculture. F. School of Mechanical and Aerospace Engineering. D. Modeling the Thermal Conductivity of Concrete Based on Its Measured Density and Porosity. C. 10(1): 45-51. R.P. 1996. Brock. ASHRAE Handbook.. Cabrera.L. Vancouver. Spitler. Clearwater Beach.. Inc. Atlanta. Proceedings of the ASHRAE/DOE/BTECC Conference. John Wiley & Sons. J. Proceedings of Building Simulation ’ 89.D. 3rd Edition. J. GA. and D. Sharp. . Oklahoma State University. Stillwater. Refrigeration and Air-Conditioning Engineers. J. Refrigerating and Air-Conditioning Engineers.. D. Fisher.L. and J. Incropera. American Society of Heating..V. American Society of Heating. and M. FL. Inc.F.A. GA. Spitler. Marshall. Prentice Hall. Elliot. Atlanta. and S. Stone.C. 1995. Manicham.L.. and D.E.. 1992.G. R. Delahoussaye. 1994. 1997. Zietlow. BC. OK. Mills.155 APPENDIX A .P. Configuration Decisions for an Automated Weather Station Network. Basic Heat and Mass Transfer. F. 1996. Users Guide of GLHEPRO. M.

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->