LOW-TEMPERATURE GEOTHERMAL ENERGY:

SYSTEMS MODELING, RESERVOIR SIMULATION,

AND ECONOMIC ANALYSIS

A Dissertation

Presented to the Faculty of the Graduate School

of Cornell University

in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

by
Koenraad Johan Hilde Ferdinand Beckers
May 2016
⃝
c 2016 Koenraad Johan Hilde Ferdinand Beckers

ALL RIGHTS RESERVED

 LOW-TEMPERATURE GEOTHERMAL ENERGY:
SYSTEMS MODELING, RESERVOIR SIMULATION,

AND ECONOMIC ANALYSIS

Koenraad Johan Hilde Ferdinand Beckers, Ph.D.

Cornell University 2016

Performance of low-temperature geothermal energy systems has been investigated

though systems modeling, reservoir simulation, and economic analysis. Both utilization
of deep geothermal energy with focus on direct-use heat though Enhanced Geothermal

Systems (EGS) and shallow geothermal energy exploited with hybrid heat pump systems

have been studied.

To assess power output and economic competitiveness of deep geothermal energy for

production of heat and/or electricity, a computer tool GEOPHIRES has been developed

which combines cost correlations and economic models with reservoir, wellbore, and sur-

face plant models. Simulations show that low-grade EGS resources (with geothermal

gradients of ∼30◦ C/km) are unattractive for solely electricity production with estimated

levelized costs of electricity between 20 and 60 ¢/kWhe . Utilizing low-grade resources

instead for low-temperature (<120◦ C) direct-use heat applications, results in competitive

levelized costs of heat (LCOH) between 6 and 14 $/MMBTU (2.0 and 4.8 ¢/kWhth ). Given
that low-grade resources are widely available and the market for low-temperature heat is

significant, geothermal energy becoming a major low-temperature heat supplier should

be considered.
To evaluate the energetic and economic performance of hybrid geothermal heat pump
(GSHP) systems for cooling-dominated applications, a TRNSYS systems model has been
developed and validated with data collected at a full-size experimental hybrid GSHP
system providing cooling for a Verizon Wireless cellular tower shelter in Varna, NY with
average continuous cooling load of 11 kWth . Simulations indicate that for the Varna Site
weather and operational conditions in the base case scenario, GSHP-based systems allow
the owner to save up to 30% of lifetime electricity consumption in comparison with air-
source heat pump (ASHP)-based systems. However, mainly because of lower upfront
capital costs, ASHP-based systems can have up to 10% lower total cost of ownership.
A novel approach for simulating transient heat transfer with slender bodies in a
conductive medium, e.g. geothermal wells and slinky-coil heat exchangers, using the
slender-body theory (SBT) has been developed. An efficient numerical implementation

is obtained based on a judicious choice of the discrete elements used to represent the
body and implementation of the Fast Multipole Method (FMM). The SBT requires a one-
dimensional spatial discretization only along the axis of the body in contrast to the three-

dimensional discretization for finite element models. Two case studies, heat transfer from

two parallel cylinders and heat transfer from a slinky-coil heat exchanger, are used to

show the speed and accuracy of the SBT model and its ability to model interacting slen-

der bodies of finite length and bodies with centerline curvature and internal advective

heat flow.
 BIOGRAPHICAL SKETCH

The author is born and raised in Belgium where he graduated magna cum laude with a

B.S. (2007) and M.S. (2009) in Mechanical Engineering at KU Leuven (University of Leu-
ven). He developed an interest in transport phenomena, systems modeling and simulat-

ing, and energy science and engineering. He conducted his M.S. Thesis on the design and
construction of a lab-scale gasification reactor at TU Munich (Technical University of Mu-
nich) in Germany, and gained valuable work experience during a 3-month engineering
internship with EGAT (Electricity Generating Authority of Thailand) at the Rajjaprabha

Hydro Power Plant near Surat Thani, Thailand.

In 2010, Koenraad moved to the U.S. and enrolled in the Chemical Engineering M.Eng.
program in Energy Economics and Engineering at Cornell University on a Fulbright

Scholarship and continued his education with doctoral research on geothermal energy

under the supervision of Prof. Jeff Tester. Throughout his graduate program, he has been

grateful for several extracurricular opportunities, including being teaching assistant at the

National Geothermal Academy and President at Von Cramm Cooperative Hall, working

as a research engineer at ExxonMobil Upstream Research Company, and visiting 35 U.S.

states.

iii
“We are not facing a shortage of resources, but a longage of expectations”

(adapted from Nate Hagens)

iv
 ACKNOWLEDGEMENTS

My 5 and half years at Cornell would not have been as wonderful and rewarding with-

out the support, guidance, friendship and love from my colleagues, friends and family.
Following is a humble attempt to express my gratitude to several individuals and orga-

nizations:
First and foremost, thank you to my advisor Prof. Jeff Tester, not only for teaching me
the depths of geothermal and the principles of sound research, but more importantly, for
being a friend, for encouraging me to follow my own ideas, and for inspiring me with

relentless energy and dedication.

Thank you to my other committee members, Prof. Teresa Jordan and Prof. Donald
Koch, for providing valuable research ideas and constructive feedback. My sincere thanks

to Polly Marion, Tara Walworth, Hilary Cullen, and Teri Carey for in essence keeping the

Energy Institute and IGERT Program running.

I am grateful for the help and advice from Prof. Brian Anderson from West Virginia

University on EGS economic modeling, Prof. Cy Yavuzturk from University of Hartford

on underground heat transfer modeling, Maria Richards and Casey Brokaw with measur-

ing the thermal conductivity of the shale samples, and Jim Feeney and colleagues from

Verizon on the cell tower shelter modeling and data logging.

Partial financial support was provided by the U.S. Department of Energy, the Cornell

Energy Institute, the Atkinson Center for a Sustainable Future, the Cornell University
Robert Frederick Smith School of Chemical and Biomolecular Engineering, Verizon and
Verizon Wireless, the Geothermal Resources Council, the Fulbright Program, the King
Baudouin Foundation, and Vesuvius plc, and is greatly appreciated.
Work wouldn’t have been as fun if it weren’t for my friends in the Tester Group, who
brought creative ideas to the thermo classes and gossip to the roof-top lunches. In partic-
ular, I would like to thank Sean Hillson for never taking my beliefs for granted, Mitchell

Ismael for the many bike rides, and Maciej Lukawski, for being my wingman.

v
 Thank you also to the Crammies, including Alex Garcia, Steph Ellman, Hannah
Holmes, Brenton White, Rennie Xie, and many others, for providing me a welcoming
home for several years on West Campus and introducing me to the “real American expe-
rience”. And thank you to my other friends from Cornell and beyond, including Veronica
Prush, for the many joyful moments we have experienced together.
Finally and most importantly, thank you to my parents and family for their endless
love and support.

Koenraad F. Beckers
Ithaca, NY, USA

March 2016

vi
 TABLE OF CONTENTS

Biographical Sketch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii

1 Introduction and Motivation 1

1.1 Geothermal Energy: a Versatile and Reliable Clean Energy Source . . . . . . 3
1.2 Geothermal Energy: an Underused Heat Supplier . . . . . . . . . . . . . . . 7
1.3 Geothermal Energy: Challenges and Solutions . . . . . . . . . . . . . . . . . 9
1.4 Dissertation Topics and Structure . . . . . . . . . . . . . . . . . . . . . . . . . 14

2 Objectives and Approach 21

2.1 Dissertation Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Dissertation Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

I Techno-Economic Analysis of Deep Geothermal Energy for Direct-

Use Heat, Electricity and Cogeneration 23
3 Rethinking Deep Geothermal Energy Beyond Electricity 24

4 GEOPHIRES Simulation Tool 27

4.1 Motivation and History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2 GEOPHIRES Model Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.3 Correlations and Component Models in GEOPHIRES . . . . . . . . . . . . . 29
4.3.1 Reservoir Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.3.2 Wellbore Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.3.3 Power Plant Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3.4 Cost Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3.5 Levelized Cost Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

5 GEOPHIRES Case-Studies 57
5.1 Case-Study 1: Deep Geothermal for Electricity and Direct-Use Heat . . . . . 57
5.1.1 Parameter Values for EGS Scenarios . . . . . . . . . . . . . . . . . . . 57
5.1.2 GEOPHIRES Simulation Results . . . . . . . . . . . . . . . . . . . . . 61
5.1.3 Discussion of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.1.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2 Case-Study 2: Deep Geothermal for District Heating in NY and PA . . . . . 78
5.3 Case-Study 3: Hybrid Low-Grade Geothermal-Biomass Cogeneration
System for Cornell University Campus . . . . . . . . . . . . . . . . . . . . . . 79
5.3.1 Introduction and Background Information . . . . . . . . . . . . . . . 79
5.3.2 Existing Energy System at Cornell University . . . . . . . . . . . . . 83

vii
 5.3.3 Proposed Hybrid Geothermal-Biomass Cogeneration System for
Cornell University . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.3.4 Simulation Parameters and Results . . . . . . . . . . . . . . . . . . . . 89
5.3.5 Discussion of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

6 Techno-Economic Analysis of Deep Geothermal Energy Systems: Key Conclu-

sions 105

II Techno-Economic Analysis of Hybrid Geothermal Heat Pump

Systems for Cooling-Dominated Applications 107
7 Introduction to Hybrid Geothermal Heat Pump Systems 108
7.1 Heat Pump Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.2 Opportunities, Challenges, and Examples of Hybrid Heat Pump Systems . 111
7.3 Design and Modeling of Hybrid Geothermal Heat Pump Systems . . . . . . 115

8 Cornell-Verizon Hybrid Geothermal Heat Pump Project:

Background Information and Varna Site 122
8.1 Background Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
8.2 Varna Site System Set-Up and Logging Data . . . . . . . . . . . . . . . . . . 125
8.3 Varna Site Geothermal Reservoir Characterization . . . . . . . . . . . . . . . 135
8.3.1 Drill Cuttings Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
8.3.2 Thermal Response Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
8.3.3 Shale Sample Thermal Conductivity Measurement . . . . . . . . . . 142

9 Cornell-Verizon Hybrid Geothermal Heat Pump Project:

TRNSYS System Analysis 148
9.1 TRNSYS Systems Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
9.1.1 Shelter Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
9.1.2 Alternating Current to Direct Current Plant Component . . . . . . . 150
9.1.3 Weather Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
9.1.4 Heat Pump Component . . . . . . . . . . . . . . . . . . . . . . . . . . 152
9.1.5 Borehole Heat Exchangers Component . . . . . . . . . . . . . . . . . 153
9.1.6 Dry-Cooler Component . . . . . . . . . . . . . . . . . . . . . . . . . . 155
9.1.7 Air-Economizer Component . . . . . . . . . . . . . . . . . . . . . . . 156
9.1.8 Controller Component . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
9.1.9 Air-Source Heat Pump Component . . . . . . . . . . . . . . . . . . . 157
9.1.10 Performance Metrics and Financial Parameters . . . . . . . . . . . . . 158
9.2 Overall TRNSYS Model Validation . . . . . . . . . . . . . . . . . . . . . . . . 159
9.3 System Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
9.3.1 Overview of Cases Studied . . . . . . . . . . . . . . . . . . . . . . . . 161
9.3.2 Base Case Simulation Results and Discussion . . . . . . . . . . . . . . 164
9.3.3 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

viii
10 Techno-Economic Analysis of Hybrid Geothermal Heat Pump Systems for
Cooling-Dominated Applications: Key Conclusions 177

III Slender-Body Theory for Transient Heat Conduction 179

11 Introduction to Slender-Body Theory 180

12 Slender-Body Theory for Transient Heat Conduction: Theoretical Derivation 187

12.1 Slender Body Geometry, Constraints and Heat Transfer Problem . . . . . . . 187
12.2 SBT Derivation for Transient Heat Conduction . . . . . . . . . . . . . . . . . 189
12.2.1 Inner Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
12.2.2 Outer Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
12.2.3 Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
12.3 Extension of Model to Short Time-Scales . . . . . . . . . . . . . . . . . . . . . 194
12.4 Other Model Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

13 Slender-Body Theory for Transient Heat Conduction: Numerical Framework 201

13.1 Space and Time Discretization of SBT Model . . . . . . . . . . . . . . . . . . 201
13.2 Equations for fi, j,m,n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
13.3 Fast Multipole Method, other Implementation Strategies, and Computa-
tional Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

14 Slender-Body Theory for Transient Heat Conduction: Case-Studies 208

14.1 Two Parallel Cylinders at Constant Temperature . . . . . . . . . . . . . . . . 208
14.2 Slinky-Coil Heat Exchanger . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211

15 Slender-Body Theory for Transient Heat Conduction: Conclusions 214

16 Overall Conclusions and Recommendations for Future Work 215

16.1 Overall Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
16.2 Recommendations for Future Work . . . . . . . . . . . . . . . . . . . . . . . . 216

A Performance Datasheets for ClimateMaster Heat Pump TT 026 219

B Borehole Heat Exchanger Transient Heat Transfer Model 224

B.1 Borehole Heat Exchanger Heat Transfer Problem . . . . . . . . . . . . . . . . 224
B.2 Existing BHE Heat Transfer Models . . . . . . . . . . . . . . . . . . . . . . . 225
B.3 Novel Hybrid BHE Heat Transfer Model . . . . . . . . . . . . . . . . . . . . . 227
B.4 Validation Case-Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

C TRNSYS Simulation Results 238

D Non-Dimensional Numbers, Heat Source Model Equations, and Decision Trees

for Calculating fi, j,m,n 242

ix
 NOMENCLATURE

List of Abbreviations

1-U Single-U
2-U Double-U
AC Alternating Current
ACAP Accelerated Climate Action Plan
ACDC Alternating Current to Direct Current
AE Air-Economizer
ASHP Air-Source Heat Pump

BHE Borehole Heat Exchanger

CAP Climate Action Plan

CF Capacity Factor

CFM Cubic Feet per Meter

CHP Combined Heat and Power

COP Coefficient of Performance

CP Circulation Pump; Current Pulse

CRF Capital Recovery Factor

CT Cooling Tower

CURBI Cornell University Renewable Bioenergy Initiative

DB Dry-Bulb
DC Dry-Cooler or Dry-Fluid Cooler
DLL Dynamic-Link Library
DOE Department of Energy
DST Duct Storage Model
E End of Heat Pulse
EAT Entering Air Temperature
EER Energy Efficiency Ratio

x
EGS Enhanced or Engineered Geothermal System

EIA Energy Information Administration

EPA Environmental Protection Agency
EWT Entering Water Temperature
ExWT Exiting Water Temperature
FCR Fixed Charge Rate
FCS Finite Cylindrical Source

FEM Finite Element Modeling or Methods

FL Full Load
FLS Finite Line Source

FMM Fast Multipole Method

GEOPHIRES Geothermal Energy for Production of Heat and electricity (“IR”)

Economically Simulated

GETEM Geothermal Energy Technology Evaluation Model

GPM Gallon Per Minute

GPRS General Purpose Research Simulator

GSHP Ground-Source or Geothermal Heat Pump

GUI Graphical User Interface

HDPE High-Density Polyethylene

HDR Hot Dry Rock

HHV Higher Heating Value

HP Heat Pump
HR Total Heat Rejection
HVAC Heating, Ventilating, and Air Conditioning

ICS Infinite Cylindrical Source

IEA International Energy Agency
IGSHPA International Ground-Source Heat Pump Association

xi
ILS Infinite Line Source

INDC Intended Nationally Determined Contributions

JEDI Job and Economic Development Impact
LCA Life Cycle Assessment
LCOE Levelized Cost of Electricity
LCOH Levelized Cost of Heat
LHV Lower Heating Value

MD Measured Depth
NPV Net Present Value
NREL National Renewable Energy Laboratory

NTU Number of Transfer Units

O&M Operation & Maintenance

ORC Organic Rankine cycle

PCCI Power Capital Costs Index

PL Part Load

PN Point Neglected

PP Previous Pulse

PS Point Source
PV Photovoltaic

RD&D Research, Development & Demonstration

RHS Right-Hand Side

S Start of Heat Pulse
SA Supply Air
SBT Slender-Body Theory

SC Sensible Cooling
SLR Spacing to Length Ratio
SMU Southern Methodist University

xii
 SPF Seasonal Performance Factor

TC Total Cooling
TCO Total Cost of Ownership
TDR Time Durations Ratio
TESS Thermal Energy System Specialists
TMY3 Typical Meteorological Year 3
TRNSYS Transient System Simulation Program

TRT Thermal Response Test

TT Two-Stage
VFD Variable Frequency Drive

WB Wet-Bulb

List of Symbols

A Single Side Area of Fracture in GEOPHIRES [m2 ]

b Half-Width Aperture of Fracture in GEOPHIRES [m]

b Coordinates of Point B in SBT Model [{m, m, m}]

B Exergy or Availability in GEOPHIRES [W]

Bi Biot Number [-]

c Specific Heat Capacity for Incompressible Solids [J/(kg·K)]

C Heat Capacity [J/K]

Cost [U.S. $]

CAP Capital Cost [U.S. $]

cf Correction Factor [-]
CF Cash Flow [U.S. $/year]
COP Coefficient of Performance [-]
cp Specific Heat Capacity at Constant Pressure [J/(kg·K)]
CRF Capital Recovery Factor [-]

xiii
d Wellbore Inner Diameter [m]

d Distance Vector in SBT Model [{m, m, m}]

D Sample Diameter in Thermal Conductivity Test [m]
db Debt Fraction [-]
e Pipe Roughness [m]
E Electricity or Heat Produced [kWh; MMBTU]
el Electricity Rate [¢/kWhe ]

eq Equity Fraction [-]

f Function; Time Function [various units]
Darcy Friction Factor in GEOPHIRES [-]

Unit Heat Response Function in SBT Model [K·m/W]

F(s) Location in Coordinate System [{m, m, m}]

FCR Fixed Charge Rate [-]

Fo Fourier Number [-]

g g-function in BHE Model [-]

Gravitational Acceleration in GEOPHIRES [9.81 m/s2 ]

G Green’s Function [various units]

h Heat Transfer Coefficient [W/(m2 ·K)]

Specific Enthalpy in GEOPHIRES [J/kg]

H Heaviside Function in SBT model [-]

Height (of Fractures) in GEOPHIRES [m]

i Interest Rate; Discount Rate; Inflation Rate [-]
I Income or Revenue in GEOPHIRES [U.S. $]
k Thermal Conductivity [W/(m·K)]
K Assigned Term in SBT Derivation [◦ C·m/W]
Curve fit Parameter in GEOPHIRES [various units]
l Length of Straight Line Element in SBT Model [m]

xiv
L Borehole Length [m]

Lt Total Slender Body Length Along Center-Line [m]

LCOE Levelized Cost of Electricity [¢/kWhe ]
LCOH Levelized Cost of Heat [U.S. $/MMBTU]
lt Lifetime [year]
m Mass Flow Rate [kg/s]
M Assigned Term in SBT Derivation [◦ C]

Molecular Mass in GEOPHIRES [g/mol]

MD Measured Depth [m]
n Arbitrary Number [-]

N Number of Wells in GEOPHIRES [-]

Number of Layers in BHE Model [-]

MT N Maintenance Cost [U.S. $/year]

NE Number of Elements [-]

NL Spatial Discretization of Slender Body [-]

NPV Net Present Value [U.S. $]

Nt Number of Time Steps [-]

Nu Nusselt Number [-]

p Percentage Temperature Decrease per Year [-]

P Pressure [Pa]

q Heat Exchange [W; BTU/h]

Q Heat Exchange per BHE or Slender Body Length [W/m]
r Radius [m]
r′ Radial Coordinate in Local SBT Coordinate System [m]
R Slender Body Radius [m]
Thermal Resistance in BHE Model [K·m/W]
Rb Borehole Thermal Resistance [K·m/W]

xv
RC Slender Body Radius of Curvature [m]

Re Reynolds Number [-]

Rt Effective Pipe Thermal Resistance [K·m/W]
s Arc-Length Along Slender Body Center-Line [m]
Laplace Variable in GEOPHIRES Reservoir Models [-]
Specific Entropy in GEOPHIRES [J/(kg·K)]
S Spacing [m]

S LR Spacing to Length Ratio [-]

t Time (s; year)
T Temperature [◦ C; ◦ F; K]

TCO Total Cost of Ownership [U.S. $]

T DR Time Durations Ratio [-]

Th Thickness [m]

Ti Imposed Temperature Field [◦ C; ◦ F; K]

ts Time-Scale in BHE Model [s]

U Velocity [m/s]

v Volume Flow Rate per Fracture and Unit Depth of [m2 /s]

Fracture
V Volume of Rock Blocks [m3 ]

w Spacing Between Center of Pipes in BHE [m]

W Electricity or Heat Production; Power Output [W; kW; MW]

x 1st Coordinate in Cartesian Coordinate System [m]
y 2nd Coordinate in Cartesian Coordinate System [m]
z 3rd Coordinate in Cartesian Coordinate System [m]
Depth in GEOPHIRES [m]
z′ Vertical Coordinate in Local SBT Coordinate System [m]

xvi
List of Greek Symbols

α Thermal Diffusivity [m2 /s]

β Parameter in Multiple Parallel Fractures Model [-]

γ Euler’s Constant [0.5772...]

Γ Parameter in Ramey’s Model [m]
δ Spatial Length [m]
∆P Pressure Drop [Pa]

∆T Temperature Drop [◦ C]
∆z Layer Thickness in BHE Model [m]
ϵ Parameter in 1-D linear Heat Sweep Model [-]

η Efficiency [-]

κ Linear Fit Slope in TRT Analysis [◦ C]

λ Linear Fit Intercept in TRT analysis [◦ C]

Λ Heat Storage Ratio in 1-D Linear Heat Sweep Model [-]

µ Dynamic Viscosity [kg/(m·s)]

ρ Density [kg/m3 ]

τ Time-Scale [s]

ϕ Parameter in Multiple Parallel Fractures Model [-]

Φ Rock Porosity [-]

χ Parameter in Multiple Parallel Fractures Model [-]
ϕ Parameter in SBT Model [-]

ω Geothermal Gradient [◦ C/m]

List of Subscripts

0 Initial; Far-Field; Dead-State

A Point A in SBT Derivation
AE Air-Economizer

xvii
AS HP Air-Source Heat Pump

amb Ambient
ave Average
b Borehole
c Cooling; Cold Reservoir
cap Capital
cas Casing

CP Current Pulse in SBT Model

cr Critical
d Downwards

db Debt

DC Dry-Cooler

distr Distribution

e End-Point

ef Effective

E End of Pulse in SBT Model

eq Equity

expl Exploration
f Fluid

fc Fixed Costs

fr Fracture
g Grout
grt Gross Revenue Taxes
GS HP Ground-Source or Geothermal Heat Pump

h Hating; Hot Reservoir

i Year; Pipe Element Index in SBT Model; Inner in BHE model
in f Inflation

xviii
in j Injection Well

it Income Taxes
itc Income Tax Credits
in Inner Solution in SBT Model
inner Inner Radius
j Pipe Element Index in SBT Model
lt Lifetime

m Medium; Mid-Point;Time Step Index in SBT Model

n Time Step Index; Grout Layer Index in BHE Model
nd Non-Dimensional

p Pipe

pt Property Tax

pp Power Plant

PP Previous Pulse in SBT Model

prod Production Well

r Rock

re f Reference

res Residence
rsv Reservoir

R Pipe Element Radius in SBT Model

s Ground (Soil)
S Start of Pulse in SBT Model
out Outlet; Outer Solution in SBT Model
o Outer in BHE Model
O&M Operation & Maintenance
stim Stimulation
t Year

xix
u Upwards in BHE Model

Utilization in GEOPHIRES
w Water
well Wellfield

xx
 LIST OF TABLES

4.1 Working fluids considered in sub- and supercritical ORC power plant
models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2 Parameter values for ORC utilization efficiency correlation . . . . . . . . . . 40
4.3 Parameter values for flash power plant utilization efficiency correlation . . 40
4.4 Parameter values for double-flash power plant capital cost correlation . . . 46

5.1 Technical and economic parameter values for EGS scenarios in

GEOPHIRES Case-Study 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.2 Resource parameter values for EGS scenarios in GEOPHIRES Case-Study 1 58
5.3 EGS scenarios simulation results for GEOPHIRES Case-Study 1 . . . . . . . 62
5.4 Base case technical and financial parameter values for GEOPHIRES Case-
Study 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.5 Base case simulation results for GEOPHIRES Case-Study 3 . . . . . . . . . . 91

8.1 Component overview and specifications of shelter and HVAC system at

Varna Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
8.2 Geothermal BHE and circulating fluid specifications at Varna Site . . . . . . 129
8.3 List of relevant data logged at Varna Site . . . . . . . . . . . . . . . . . . . . 131
8.4 Published thermophysical properties for shale . . . . . . . . . . . . . . . . . 137
8.5 Model input parameter values and results for Varna Site thermal response
test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

9.1 TRNSYS model financial parameter values . . . . . . . . . . . . . . . . . . . 159

9.2 Overall TRNSYS model validation results . . . . . . . . . . . . . . . . . . . . 160
9.3 TRNSYS base case parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 162
9.4 TRNSYS base case simulation results . . . . . . . . . . . . . . . . . . . . . . 165

B.1 Case-study parameter values for validating BHE heat transfer model . . . . 234

C.1 Effect of total BHE depth on system performance in Case 1 (GSHP only)
in base case scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
C.2 Sensitivity of TRNSYS simulation results to shelter heat generation . . . . . 239
C.3 Sensitivity of TRNSYS simulation results to reservoir thermal conductivity 239
C.4 Sensitivity of TRNSYS simulation results to electricity rate . . . . . . . . . . 240
C.5 Sensitivity of TRNSYS simulation results to net discount rate . . . . . . . . 241
C.6 Sensitivity of TRNSYS simulation results to drilling cost . . . . . . . . . . . 241

D.1 Non-dimensional numbers utilized in equations and decision trees for cal-
culating fi, j,m,n in SBT numerical model . . . . . . . . . . . . . . . . . . . . . 242
D.2 Heat source model equations to calculate fi, j,m,n for selected pipe in SBT
numerical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
D.3 Heat source model equations to calculate fi, j,m,n for neighboring pipes in
SBT numerical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244

xxi
 LIST OF FIGURES

1.1 Historic and projected worldwide primary energy consumption by fuel type 2
1.2 Carbon budget and global energy-related CO2 emissions . . . . . . . . . . . 2
1.3 Schematic diagram of hydrothermal system . . . . . . . . . . . . . . . . . . 4
1.4 Schematic diagram of Enhanced or Engineered Geothermal System . . . . 5
1.5 U.S. thermal energy spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Geothermal well drilling and completion costs . . . . . . . . . . . . . . . . . 11
1.7 Temperature at 5.5 km depth for Continental U.S. . . . . . . . . . . . . . . . 12

3.1 Schematic diagram of Enhanced or Engineered Geothermal System for co-

generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

4.1 GEOPHIRES operating scheme . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.2 ORC and flash power plant utilization efficiency . . . . . . . . . . . . . . . . 41
4.3 ORC and flash power plant geofluid exit temperature . . . . . . . . . . . . . 42
4.4 Subcritical ORC and double-flash power plant capital cost correlations . . . 47

5.1 Levelized costs for EGS scenarios in GEOPHIRES Case-Study 1 . . . . . . . 63

5.2 Capital costs distribution for EGS scenarios in GEOPHIRES Case-Study 1 . 63
5.3 Sensitivity of LCOE and LCOH in base case scenario of GEOPHIRES Case-
Study 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.4 Drilling depth sensitivity of LCOE and LCOH in base case scenario of
GEOPHIRES Case-Study 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.5 Comparison of GEOPHIRES LCOE results with LCOE for other electricity
generating technologies in GEOPHIRES Case-Study 1 . . . . . . . . . . . . 70
5.6 Comparison of GEOPHIRES LCOH results with LCOH for natural gas
boilers in GEOPHIRES Case-Study 1 . . . . . . . . . . . . . . . . . . . . . . 72
5.7 Estimated LCOH and total capacity for deep geothermal district heating
systems in various places in NY and PA in GEOPHIRES Case-Study 2 . . . 79
5.8 Envisioned future energy system for Cornell University campus . . . . . . 82
5.9 Cornell University campus heating, electricity and cooling consumption
from 2000 to 2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.10 Cornell University natural gas turbine with heat recovery steam generat-
ing system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.11 Simplified diagram of proposed hybrid geothermal-biomass cogeneration
system in GEOPHIRES Case-Study 3 . . . . . . . . . . . . . . . . . . . . . . 89
5.12 Monthly geothermal and biomass heat and electricity generation results
in GEOPHIRES Case-Study 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

7.1 Simplified diagram of heat pump and heat engine . . . . . . . . . . . . . . . 109

7.2 Simplified diagram of water-to-air GSHP system in cooling mode . . . . . . 111
7.3 Examples of hybrid water-to-air GSHP systems for cooling-dominated ap-
plications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.4 Examples of hybrid water-to-air GSHP systems for heating-dominated ap-
plications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

xxii
8.1 Photograph of drill rig at Varna Site taken in May 2013 . . . . . . . . . . . . 126
8.2 Photograph of Verizon shelter and monopole at Varna Site taken in
September 2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
8.3 Three-dimensional simplified schematic of Varna Site . . . . . . . . . . . . . 127
8.4 Lay-out of BHE field, equipment shelter and cellular tower of Verizon hy-
brid geothermal heat pump project at Varna Site . . . . . . . . . . . . . . . . 130
8.5 Shelter and ambient temperature at Varna Site from August 5th , 2014 till
January 30th , 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
8.6 Circulating fluid (geofluid) return and supply temperature at Varna Site
from August 29th , 2014 till January 30th , 2016 . . . . . . . . . . . . . . . . . . 133
8.7 Reservoir temperature in borehole D at Varna Site from August 5th , 2014
till January 30th , 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
8.8 Diagram of geothermal reservoir geology and drill cuttings collection
depths at Varna Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
8.9 Thermal response test (TRT) at Varna Site . . . . . . . . . . . . . . . . . . . . 139
8.10 Measured average fluid temperature and linear fit for thermal response
test at Varna Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
8.11 Thermal conductivity measurement apparatus at Southern Methodist
University Geothermal Lab . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
8.12 Schematic diagram of thermal conductivity measurement apparatus . . . . 143

9.1 Example of TRNSYS model of hybrid geothermal heat pump system (Case
3: GSHP + DC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
9.2 ACDC Plant heat generation during December 2014 . . . . . . . . . . . . . 151
9.3 Subcomponents in TRNSYS macro component for ClimateMaster TT 026
heat pump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
9.4 Measured and simulated data for ClimateMaster TT 049 residential heat
pump system in Lansing, NY . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
9.5 Varna Site shelter and ambient temperature measured during April and
May 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
9.6 Effect of total BHE depth on lifetime average heat pump COP and electric-
ity consumption for Case 1 (GSHP only) in base case scenario . . . . . . . . 169
9.7 Effect of total BHE depth on TCO for Case 1 (GSHP only) in base case
scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
9.8 Effect of shelter heat generation on TCO for Cases 1 to 5 . . . . . . . . . . . 171
9.9 Effect of reservoir thermal conductivity on TCO for Cases 1 to 3 . . . . . . . 172
9.10 Effect of electricity rate on TCO for Cases 1 to 5 . . . . . . . . . . . . . . . . 173
9.11 Effect of net discount rate on TCO for Cases 1 to 5 . . . . . . . . . . . . . . . 174
9.12 Effect of drilling cost on TCO for Cases 1 to 3 . . . . . . . . . . . . . . . . . . 174

12.1 SBT heat transfer problem and geometry . . . . . . . . . . . . . . . . . . . . 187

13.1 SBT for transient heat conduction numerical framework . . . . . . . . . . . 202

14.1 SBT numerical model case-studies . . . . . . . . . . . . . . . . . . . . . . . . 209

14.2 Results for SBT numerical model case-studies . . . . . . . . . . . . . . . . . 211

xxiii
A.1 Performance datasheet for ClimateMaster heat pump Tranquility 27 TT
026 in part load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
A.2 Performance datasheet for ClimateMaster heat pump Tranquility 27 TT
026 in full load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
A.3 Airflow correction tables for ClimateMaster heat pump Tranquility 27 TT
026 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
A.4 Air temperature correction tables for ClimateMaster heat pump Tranquil-
ity 27 TT 026 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

B.1 Simplified diagram of single-U BHE . . . . . . . . . . . . . . . . . . . . . . . 225

B.2 Drawing of g-functions as a function of dimensionless time for several
BHE field configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
B.3 Cross-section of BHE illustrating spatial discretization in horizontal plane . 230
B.4 Network of thermal resistances and capacities within 1 horizontal layer of
a discretized BHE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
B.5 Diagram showing layered thermal network within BHE . . . . . . . . . . . 232
B.6 Case-study simulation results for validation of hybrid BHE heat transfer
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235

D.1 Decision trees utilized in SBT numerical model . . . . . . . . . . . . . . . . 245

xxiv
 CHAPTER 1
INTRODUCTION AND MOTIVATION

The world faces a daunting challenge over the next several decades in providing af-
fordable, environmentally-friendly, and reliable energy and other natural resources to
feed a growing population, fuel a growing economy, and supply increasing consump-
tion, while dealing with climate change, geopolitical tensions, and soaring national debts.
The challenge might look even more profound when acknowledging that the average
consumer has little interest in global affairs and elected political officials tend to follow

short-term agendas.

Several agencies and energy companies are projecting a steady increase in global en-

ergy demand in the next few decades with fossil fuels remaining as the dominant en-

ergy source (EIA, 2015; OECD/IEA, 2015; BP, 2015; ExxonMobil, 2015). The International

Energy Agency, for example, predicts a 20% increase in global primary energy demand

between 2013 and 2030, illustrated in Figure 1.1, for the Intended Nationally Determined

Contributions (INDC) scenario which accounts for the measures put forward by several

countries for reducing greenhouse gas emissions. Even with these reduction goals, how-

ever, the remaining carbon budget to limit the average global temperature rise to 2◦ C
by 2100, would already shrink to zero by 2040, as shown in Figure 1.2. A 2◦ C tempera-

ture rise is the threshold put forward by the Intergovernmental Panel on Climate Change

(IPCC, 2014) to limit the impact of climate change on weather, oceans, ecosystems, agri-
culture, etc., with examples such as rising sea levels, regionally diminishing crop yields,
and increased frequency and intensity of severe weather events including droughts and
heatwaves. It is evident that bold leadership, massive investments, and unprecedented
international cooperation are required to transform to a sustainable, low-carbon global
society where clean energy sources such as solar, wind, biomass, geothermal, and poten-
tially nuclear play a prominent role.

1
Chapter 1. Introduction and Motivation 2

Figure 1.1 – Historic and projected worldwide primary energy consumption by fuel type in Million
ton of oil equivalent (Mtoe) for the Intended Nationally Determined Contributions (INDC) scenario by
the International Energy Agency (OECD/IEA, 2015). 15,000 Mtoe corresponds to about 630 EJ. “Other
renewables” includes geothermal, solar, wind and marine.

Figure 1.2 – Global energy-related CO2 emissions in Gt for INDC scenario (red bars; left axis) and
remaining carbon budget to limit global temperature rise to 2◦ C with > 50% probability (blue curve;
right axis) (OECD/IEA, 2015).
Chapter 1. Introduction and Motivation 3

1.1 Geothermal Energy: a Versatile and Reliable Clean Energy Source

Geothermal energy has inherent advantages as a versatile and reliable clean energy source

and should therefore be given serious thought as a key player in the future energy sys-
tem (Tester et al., 2006, 2012; Glassley, 2015). The resource is versatile since it occurs

in many forms, qualities, and quantities, and can be utilized for various end-uses in-
cluding electricity production, direct-use heat and cogeneration of electricity and heat.
Deep geothermal energy is defined as resources located at depths from several 100 m

to several km and are categorized in this work into (1) hydrothermal resources, (2) hot
sedimentary aquifers, and (3) Enhanced or Engineered Geothermal Systems (EGS), also

called hot dry rock (HDR). Other deep geothermal resource types and classifications do

exist such as magma energy and geopressured systems, but are not further discussed.

Hydrothermal resources are the most well-known, with examples as Lardarello in Italy,

Hellisheidarvirkjun in Iceland, and the Geysers in California, the latter being the largest

geothermal field in the world (DiPippo, 2012). These resources are characterized as hav-

ing in-situ hot water, and high reservoir permeability and connectivity (see Figure 1.3).

They are widely developed globally with a total installed electricity generating capacity

of over 12 GWe (Bertani, 2015), but typically only occur in tectonically or volcanically

active regions. The second group, hot sedimentary aquifers, are more geographically dis-
tributed with good permeability and connectivity that require moderate drilling depths

but typically have lower temperatures and are not as extensively developed. The last cat-
egory, EGS, typically refers to the thermal energy stored in hot crystalline basement rocks,
present anywhere in the world (see Figure 1.4). These reservoirs usually require stimula-
tion to enhance the permeability and connectivity, and moderate to deep drilling depths
to reach sufficiently high temperatures. EGS is still in its research phase with few success-
ful pilot field projects that have demonstrated its feasibility at a limited scale (Tester et al.,
2006) such as the Fenton Hill project in New Mexico and the Soultz-sous-Forêts project in
Chapter 1. Introduction and Motivation 4

Figure 1.3 – Schematic diagram of hydrothermal geothermal resource, which are characterized as hav-
ing high reservoir permeability and in-situ hot water (Goldstein et al., 2011).

France. However, due to its massive resource base, developing EGS technology at a com-

mercial scale would be a key factor for geothermal to become a dominant energy player

in the 21st century.

Throughout this work, high-, medium-, and low-grade deep geothermal energy
resources are interpreted as having high (∼70◦ C/km), medium (∼50◦ C/km) and low
(∼30◦ C/km) average geothermal gradients, respectively. Further, high-, medium-, and

low-temperature deep geothermal energy resources are considered >200◦ C, in the range
120-200◦ C, and <120◦ C, respectively. In terms of terminology, they are used interchange-
ably with high, medium, and low geothermal energy supply temperatures. In theory,
for a given resource grade, any supply temperature could be achieved depending on the
drilling depth. However, from an economic point of view, to obtain high supply tem-
Chapter 1. Introduction and Motivation 5

Figure 1.4 – Schematic diagram of Enhanced or Engineered Geothermal System. Heat is extracted
from hot crystalline rocks by circulating a heat exchange fluid in a closed-loop through a man-made
reservoir (Tester et al., 2006).

peratures, typically high-grade resources are required whereas low supply temperatures

could be obtained from any resource grade.

In contrast to deep geothermal energy, shallow geothermal energy is defined as the

large thermal source or sink at constant, low temperatures (10-20◦ C) in the upper 100 to
200 m of the crust. Although not always considered a true geothermal resource, we will
refer to it in this work as low-temperature geothermal energy as well, while recognizing
that the temperatures are usually too low for direct-use heat. Nevertheless, a shallow
geothermal reservoir in combination with a heat pump can efficiently provide space and
water heating and cooling. In fact, millions of geothermal heat pump (GSHP) systems
have been installed throughout the world.
Chapter 1. Introduction and Motivation 6

Geothermal reservoirs have an embedded thermal storage system which allows for re-
liable baseload, load-following, peak, or dispatchable power production without the need
for backup capacity that is typically required for intermittent energy sources such as solar

and wind. Further, the amount of thermal energy stored in the Earth, which is a result of
the formation of the Earth’s core and the decay of radioactive isotopes (Glassley, 2015), is

on the order of 1013 EJ (Rybach et al., 2000), several orders of magnitude larger than the
current worldwide primary energy consumption of about 570 EJ per year (OECD/IEA,
2015). Even when only considering a 1◦ C drawdown of the upper 10 km of the accessible
continental crust, the resource base is still massive, about 2·106 EJ (Armstead and Tester,

1987). These huge amounts for stored thermal energy in accessible rocks in the subsurface

make up the geothermal resource potential. Unfortunately, some have mistakenly based
the geothermal potential on the average steady-state continental heat flux (87 mW/m2

(Glassley, 2015)), a value much smaller than the solar constant (1,354 W/m2 (Tester et al.,

2012)). In practice, the rate of heat extraction from geothermal systems in commercial

use greatly exceeds the steady-state conduction rate, as a result of both convective and

conductive heat transfer from the rock mass surrounding the center reservoir. Moreover,

research by Fox et al. (2013) has shown that even EGS reservoirs that are governed by

heat conduction only can sufficiently recover when deactivated for a period of 2-4 times

their initial thermal extraction period. Hence, when applying heat farming strategies by
rotating between different reservoirs, EGS methods can be utilized to tap sustainably into
the vast geothermal resource base.

Other advantages are that deep geothermal energy systems require no fuel and there-
fore their operation and economics are not susceptible to volatile fuel prices. Further, they
have a relatively low land area footprint and emit few to none greenhouse gas emissions
during operation (Tester et al., 2006, 2012; Glassley, 2015). However, greenhouse gases
are emitted during construction (e.g. for drilling) and a proper Life Cycle Assessment
(LCA) is required to assess the overall environmental impact during the lifetime of the
Chapter 1. Introduction and Motivation 7

plant (Frick et al., 2010; Lacirignola and Blanc, 2013; Bayer et al., 2013; Gerber, 2015).

On the other hand, shallow GSHP systems do require electricity (or a heat source with
absorption heat pumps) and therefore greenhouse gases are emitted during operation
when electricity is derived from fossil fuels. However, when neglecting the electrical grid
transmission losses, and assuming an average power plant conversion efficiency of 50%,
and a heat pump coefficient of performance (COP) of 4, meaning one unit of electricity is
required to supply four units of heat, 50% less energy is consumed in comparison with
on-site heating oil or natural gas heating systems.

1.2 Geothermal Energy: an Underused Heat Supplier

Geothermal energy can be utilized for production of electricity, direct-use heat, or for co-

generation of electricity and heat, also called combined heat and power (CHP). Medium-

and high-grade geothermal resources with supply temperatures of 150◦ C or higher are

required to produce electricity using binary cycle or flash power plants with sufficiently

high conversion efficiencies. Resources with lower supply temperatures are ideal to pro-

duce direct-use heat for district heating systems, fish farming, greenhouses, and indus-

trial processes such as pasteurization and pulp drying. Other uses for the geothermal

brine, not further discussed here, are mineral extraction, and for recreational and med-
ical purposes. Unfortunately, government agencies, utilities, and energy companies of-
ten consider geothermal energy solely for electricity production, despite the fact that the
majority of the resources are of lower-temperature, with great potential for supplying
direct-use heat.

A key improvement in the current energy system would result by properly integrat-
ing low-temperature geothermal energy utilization. This would enable achieving a better
Chapter 1. Introduction and Motivation 8

match between different energy sources and energy end-uses. Natural gas for example
is a high-grade fuel with correspondingly high availability or exergy meaning great po-
tential to do work (Tester et al., 2012). With combustion, temperatures up to 2,000◦ C

are obtained which are ideal for electricity generation in a combined cycle power plant
with utilizing some remaining heat for direct-use in a district heating system or industrial

process. In contrast, burning natural gas to supply residential heat directly at tempera-
tures around 50◦ C is from a thermodynamic point of view rather wasteful and results in
large amounts of exergy destruction. On the other hand, lower-grade geothermal energy
with supply temperatures in the range 50-100◦ C is a perfect match for supplying resi-

dential heat, but is unfortunately often overlooked. In many places around the world,

low-temperature heat is predominantly supplied by high-grade fuels such as natural gas

and heating oil instead of more appropriate energy sources such as direct-use geothermal,

GSHPs, solar thermal and waste heat.

The market for low-temperature heat sources, often forgotten about in national energy

debates, is significant. For the U.S., Fox et al. (2011) developed a distribution spectrum of

the thermal energy demand in the residential, commercial and industrial sector as a func-

tion of its end-use temperature, shown in Figure 1.5. They estimated that a whopping 25

EJ or about 25% of the total U.S. primary energy supply is consumed as low-temperature

heat below 120◦ C, with space and water heating around 50◦ C accounting for more than
half. This compares to about 40% of the primary energy used in the U.S. to generate elec-

tricity (EIA, 2011). In the European Union in 2009 (27 member states), the total demand
for low-temperature heat in the industrial sector (under 100◦ C), and for water heating,
and space heating and cooling in the residential and commercial sector is estimated at 15
EJ (Pardo et al., 2012), about 20% of the total primary energy demand (EC, 2014). Without
a doubt, similar low-temperature heat opportunities are present in other regions around
the world.
Chapter 1. Introduction and Motivation 9

Figure 1.5 – 2008 U.S. thermal energy spectrum. Distribution of thermal energy demand as a function
of end-use temperature (Fox et al., 2011).

1.3 Geothermal Energy: Challenges and Solutions

Despite having favorable properties including reliability, sustainability, and versatility,

the pace of development of geothermal systems, especially deep geothermal, in many

countries including the U.S. is rather low. The compound annual growth rate between
2010 and 2014 for worldwide installed capacity of deep geothermal for electricity pro-
duction is only 3% (10.9 to 12.6 GWe ) (Bertani, 2015), while for GSHPs around 8.5% (33.1
to 49.9 GWth ) (Lund and Boyd, 2015). This is in contrast to an annual growth rate of
13% (198 to 370 GWe ) for wind and 34% (40 to 177 GWe ) for solar photovoltaic (PV)
Chapter 1. Introduction and Motivation 10

(REN21, 2015). Key barriers for widespread deployment of both deep and to a certain
extent shallow geothermal systems are high investment costs in comparison with alter-
natives, currently low prices of oil and natural gas (at least in the U.S.), and lack of public

understanding coupled with an inherent resistance to change. Other challenges specifi-

cally for deep geothermal systems are risks and uncertainty with respect to the resource

quality and reservoir productivity.

For developing deep geothermal systems, drilling is required to depths of typically

several km which is a capital-intensive operation and can constitute over 50% of the to-
tal investment cost of the project. Lukawski et al. (2014) assembled a database of about
150 geothermal wells drilled between 1972 and 2013, and developed an average well

drilling and completion cost curve, shown in Figure 1.6. For example, assuming an av-

erage geothermal gradient of 30◦ C/km, a surface temperature of 20◦ C, and a wellbore

temperature drop of 10◦ C, a doublet system consisting of two wells of 4 km depth would

result in an moderate initial production temperature of about 130◦ C and require a drilling

and completion cost of about $11M per well. This illustrates the high upfront capital

costs, without even considering that exploration, reservoir stimulation and surface plant

costs are incurred before any kWh of heat or electricity are produced. Research is on-

going to develop new and potentially more affordable deep drilling techniques such as

jet-assisted, thermal spallation and laser drilling (Maurer, 1980; Pierce et al., 1996; Hillson
and Tester, 2015).

Another challenge for exploiting deep geothermal energy is identifying and charac-
terizing the local resource. Unlike wind and solar energy which can be easily measured
and quantified, the local geothermal potential can usually not be assessed accurately di-
rectly until one or several costly exploration wells have been drilled, since it depends on
factors such as the geothermal gradient, the rock permeability and connectivity, and the
presence of in-situ hot water. Instead, one utilizes first indirect geological, geochemical
Chapter 1. Introduction and Motivation 11

Figure 1.6 – Geothermal Well Drilling and Completion Costs in U.S. M$ as a function of measured
depth (Lukawski et al., 2014).

and geophysical methods to collect information on the subsurface (Glassley, 2015), a sim-

ilar approach as applied in the oil and gas sector. High-grade resources are often found

close to active or recently-active volcanoes (e.g. Geysers Field in California), calderas (e.g.
Los Humeros Caldera in Mexico), and fault-bounded sedimentary basins (e.g. Imperial
Valley in California) or extensional complexes (e.g. Great Basin in Western U.S.). Fur-
ther, surface manifestations such as geysers, hot springs, and silica sinter and travertine
deposits can be an indication of the presence of underground hydrothermal resources.
Also, the local fluid composition can reveal the temperature of a potential geothermal
Chapter 1. Introduction and Motivation 12

Figure 1.7 – Southern Methodist University (SMU) geothermal resource map for continental U.S.: Tem-
peratures at 5.5 km depth (Blackwell et al., 2013).

reservoir by using Geothermometers (e.g. based on SiO2 ) and isotope analysis. More data

can be collected through geophysical methods including aeromagnetic, resistivity, mag-

netotelluric, gravity, and seismic surveys which can reveal subsurface anomalies such

as hydrothermally-altered rocks, high fracture density zones, and the presence of in-situ

water. Various datasets can be combined to develop favorability maps (e.g. the geother-
mal potential map for the Great Basin by Coolbaugh et al. (2005)), which can help target

promising areas for drilling of exploratory wells. For EGS, one typically creates surface

heat flux and temperature-by-depth maps based on well logs from oil, gas, and geother-
mal exploration and development wells. Examples are the geothermal resource maps for
the continental U.S. by researchers from Southern Methodist University (Blackwell et al.,
2013) (see Figure 1.7) or the regional maps for the states of New York and Pennsylvania
(Stutz et al., 2015).

In order to predict the thermal and hydraulic performance of a potential geothermal

Chapter 1. Introduction and Motivation 13

site, or for field management throughout the lifetime of an existing geothermal system,
proper reservoir modeling and simulation is required (Grant and Bixley, 2013). Various
models can be utilized, from simple analytical to spatially three-dimensional numeri-

cal simulators, with or without incorporating data from well logs or productivity tests.
O’Sullivan et al. (2001) have provided an excellent overview of geothermal simulators,

with TOUGH2 (Pruess, 1991), developed at Lawrence Berkeley Laboratory, probably the
most common one. Other examples are the General Purpose Research Simulator (GPRS)
from Stanford University (Wong et al., 2015) or the thermo-hydraulic reservoir simulator
for fracture-based systems developed by Fox et al. (2014) at Cornell University .

Acid, thermal, or hydraulic stimulation might be necessary to increase the perme-

ability, connectivity and hence productivity of a geothermal reservoir (Tester et al., 2006;

Glassley, 2015; Grant and Bixley, 2013). Acid stimulation can be used to remove an ob-

struction near the wellbore and hereby lower the overall reservoir impedance. Thermal

stimulation is achieved by injecting cold water in a hot reservoir, which causes rocks to

contract and open up existing fractures. Hydraulic stimulation, the main method applied

for creating an EGS reservoir, is based on injecting high-pressure fluids to hydro-shear or

hydro-fracture the rocks. Reservoir stimulation can cause induced seismicity (Majer et al.,

2007), potentially to unacceptable levels. The Basel 1 EGS site is an example where the

occurrence of several microseismic events led to the suspension and later abandonment
of the project (Häring et al., 2008). However, with proper operation and management,
induced seismicity is expected to be manageable and kept within safe limits (Tester et al.,

2006). Research activities (e.g. by McClure (2012)) and field experiments (e.g. the New-
berry Volcano EGS Demonstration Project in Oregon (Cladouhos et al., 2015)) are ongoing
which provide a better understanding on the link between reservoir stimulation and in-
duced seismic events.

Development of shallow geothermal energy systems is less capital-intensive, less

Chapter 1. Introduction and Motivation 14

technically-complex, and typically at a smaller scale (i.e. at the level of an individual

home or business owner). However, they are still the subject of extensive research world-
wide (Yang et al., 2010; Spitler, 2005; Chua et al., 2010), particularly in the fields of reser-

voir simulation (e.g. estimating the long-term thermal performance with influence of
groundwater flow), and overall system design and hybridization (e.g. integration and op-

timization of seasonal thermal storage with a hybrid solar-geothermal system). Another

specific challenge for GSHPs is raising public awareness of the existence and benefits of
this underused technology, e.g. through publicity campaigns (Chua et al., 2010).

1.4 Dissertation Topics and Structure

The dissertation is divided into three parts: Part I: Techno-Economic Analysis of Deep

Geothermal Energy for Direct-Use Heat, Electricity and Cogeneration (Chapters 3 to 6);

Part II: Techno-Economic Analysis of Hybrid Geothermal Heat Pump Systems for

Cooling-Dominated Applications (Chapters 7 to 10); and Part III: Slender-Body Theory

for Transient Heat Conduction (Chapters 11 to 15). Chapter 2 provides the specific dis-

sertation objectives and approach. The key dissertation conclusions and a list of recom-

mendations for future work are given in Chapter 16, the final chapter.

Part I of the dissertation provides a techno-economic analysis of deep geothermal en-

ergy systems, with focus on EGS for either electricity generation, direct-use heat produc-
tion, or cogeneration. Given the high investment costs, and resource and reservoir risks
and uncertainties of deep geothermal energy, it is crucial to properly evaluate the techni-
cal and economic performance of any proposed system. Part I has four chapters: Chapter
3 sets the tone for Part I with reemphasizing the potential of deep geothermal, specifically
EGS, beyond electricity. Chapter 4 presents a computer program to simulate the heat
and/or electricity production of a deep geothermal system, and, combined with capital
Chapter 1. Introduction and Motivation 15

and operation and maintenance (O&M) cost models, to evaluate the performance over
the lifetime of the system. This program is applied in Chapter 5 to different case-studies
including deep geothermal for solely electricity or direct-use heat production, as well as

for cogeneration in a hybrid setting with biomass for the Cornell University campus. The
take-away messages are listed in Chapter 6.

Part II of the dissertation provides a techno-economic analysis of shallow geothermal

energy exploited with hybrid heat pump systems for cooling-dominated, non-residential
applications. While GSHPs for residential heating and cooling purposes is nowadays a

standard, well-understood HVAC (Heating, Ventilating, and Air Conditioning) option, a

better understanding of hybrid systems, as well as of applications beyond the residential

sector, is desired. Chapter 7 provides an introduction to heat pumps, hybridization op-

tions and modeling tools. An experimental full-scale hybrid GSHP set-up providing the

climate control of a cellular tower shelter on the Cornell University campus is introduced

in Chapter 8. Numerical simulations and results of various system configurations are

presented in Chapter 9. The main conclusions are listed in Chapter 10.

Part III of the dissertation presents a novel approach using slender-body theory to

simulate accurately and computationally-fast the transient heat conduction with slender

bodies, e.g. geothermal boreholes and slinky-coil heat exchangers. Chapter 11 in Part III
introduces the slender-body theory and discusses its advantages but also constraints for
modeling transient heat conduction. In Chapter 12, the theoretical framework is devel-

oped and Chapter 13 presents a possible numerical implementation of the model. Two
case-studies to illustrate the applicability and high computational speed of the model are
discussed in Chapter 14. Finally, a summary and conclusions are given in Chapter 15.
Chapter 1. Introduction and Motivation 16

References

Armstead, H. C. H. and Tester, J. W. (1987). Heat Mining. E. & F.N. Spon Ltd., London and
New York.

Bayer, P., Rybach, L., Blum, P., and Brauchler, R. (2013). Review on life cycle environmen-
tal effects of geothermal power generation. Renewable and Sustainable Energy Reviews,
26: 446–463.

Bertani, R. (2015). Geothermal Power Generation in the World 2010-2014 Update Report.

In Proceedings World Geothermal Congress 2015, Melbourne, Australia, 19-25 April 2015.

Blackwell, D. D., Richards, M. C., and Frone, Z. S. (2013). SMU Geothermal Resource Map.

Southern Methodist University (SMU), Dallas, Texas, United States.

BP (2015). BP Energy Outlook 2035.

Chua, K., Chou, S., and Yang, W. (2010). Advances in heat pump systems: A review.

Applied Energy, 87 (12): 3611–3624.

Cladouhos, T. T., Petty, S., Swyer, M. W., Uddenberg, M. E., and Nordin, Y. (2015). Results

from Newberry Volcano EGS Demonstration. In Proceedings World Geothermal Congress

2015, Melbourne, Australia, 19-25 April 2015.

Coolbaugh, M., Zehner, R., Kreemer, C., Blackwell, D., Oppliger, G., Sawatzky, D., Blewitt,

G., Pancha, A., Richards, M., Helm-Clark, C., Shevenell, L., Raines, G., Johnson, G.,
Minor, T., and Boyd, T. (2005). Geothermal potential map of the Great Basin, western United
States.

DiPippo, R. (2012). Geothermal power plants: principles, applications, case studies and environ-

mental impact. Butterworth-Heinemann, 3rd edition.

Chapter 1. Introduction and Motivation 17

EC (2014). EU energy in figures, Statistical Pocketbook 2014. European Commis-

sion. Available at http://ec.europa.eu/energy/sites/ener/files/documents/
2014 pocketbook.pdf.

EIA (2011). Annual Energy Review 2011, U.S. Energy Information Administration.
DOE/EIA-0384.

EIA (2015). International Energy Outlook 2015, U.S. Energy Information Administration.

ExxonMobil (2015). The Outlook for Energy: A View to 2040.

Fox, D. B., Sutter, D., and Tester, J. W. (2011). The thermal spectrum of low-temperature
energy use in the United States. Energy & Environmental Science, 4 (10): 3731–3740.

Fox, D. B., Sutter, D., Beckers, K. F., Lukawski, M. Z., Koch, D. L., Anderson, B. J., and

Tester, J. W. (2013). Sustainable heat farming: Modeling extraction and recovery in

discretely fractured geothermal reservoirs. Geothermics, 46: 42–54.

Fox, D. B., Koch, D. L., and Tester, J. W. (2014). Modeling Discretely Fractured Geothermal

Reservoirs: A Focus on Thermal Recovery, Fracture Intersections, and Fracture Rough-

ness. In GRC Transactions, 38: 271–280.

Frick, S., Kaltschmitt, M., and Schröder, G. (2010). Life cycle assessment of geothermal

binary power plants using enhanced low-temperature reservoirs. Energy, 35 (5): 2281–

2294.

Gerber, L. (2015). Designing Renewable Energy Systems: A Life Cycle Assessment Approach.
EPFL Press.

Glassley, W. E. (2015). Geothermal energy: renewable energy and the environment. CRC Press,
2nd edition.

Goldstein, B., Hiriart, G., Bertani, R., Bromley, C., Gutiérrez-Negrı́n, L., Huenges, E.,
Muraoka, H., Ragnarsson, A., Tester, J., and Zui, V. (2011). Geothermal Energy. In
Chapter 1. Introduction and Motivation 18

Edenhofer, O., Pichs-Madruga, R., Sokona, Y., Seyboth, K., Matschoss, P., Kadner, S.,
Zwickel, T., Eickemeier, P., Hansen, G., Schlömer, S., and von Stechow, C., editors,
IPCC Special Report on Renewable Energy Sources and Climate Change Mitigation. Cam-

bridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Grant, M. and Bixley, P. F. (2013). Geothermal Reservoir Engineering. Elsevier, 2nd edition.

Häring, M. O., Schanz, U., Ladner, F., and Dyer, B. C. (2008). Characterisation of the Basel
1 enhanced geothermal system. Geothermics, 37 (5): 469–495.

Hillson, S. D. and Tester, J. W. (2015). Heat Transfer Properties and Dissolution Behavior

of Barre Granite as Applied to Hydrothermal Jet Drilling with Chemical Enhancement.

In Proceedings Fortieth Workshop on Geothermal Reservoir Engineering, Stanford University,

Stanford, California, January 26 - January 28, 2015, SGP-TR-204.

IPCC (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I,

II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate

Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva,

Switzerland.

Lacirignola, M. and Blanc, I. (2013). Environmental analysis of practical design options for
enhanced geothermal systems (EGS) through life-cycle assessment. Renewable Energy,

50: 901–914.

Lukawski, M. Z., Anderson, B. J., Augustine, C., Capuano, L. E., Beckers, K. F., Livesay, B.,
and Tester, J. W. (2014). Cost analysis of oil, gas, and geothermal well drilling. Journal
of Petroleum Science and Engineering, 118: 1–14.

Lund, J. W. and Boyd, T. L. (2015). Direct Utilization of Geothermal Energy 2015 World-
wide Review. In Proceedings World Geothermal Congress 2015, Melbourne, Australia,
19-25 April 2015.
Chapter 1. Introduction and Motivation 19

Majer, E. L., Baria, R., Stark, M., Oates, S., Bommer, J., Smith, B., and Asanuma, H. (2007).
Induced seismicity associated with enhanced geothermal systems. Geothermics, 36 (3):
185–222.

Maurer, W. C. (1980). Advanced drilling techniques. Penn Well Books, Tulsa, Oklahoma,

United States.

McClure, M. W. (2012). Modeling and characterization of hydraulic stimulation and induced

seismicity in geothermal and shale gas reservoirs. PhD Dissertation, Stanford University,
Stanford, California, United States.

OECD/IEA (2015). World Energy Outlook Special Report. Energy and Climate Change, Organ-

isation for Economic Co-operation and Development / International Energy Agency.

O’Sullivan, M. J., Pruess, K., and Lippmann, M. J. (2001). State of the art of geothermal

reservoir simulation. Geothermics, 30 (4): 395–429.

Pardo, N., Vatopoulos, K., Krook-Riekkola, A., Moya, J. A., and Perez, A. (2012). Heat and

cooling demand and market perspective. European Commission, JRC Scientific and Policy

Reports.

Pierce, K. G., Livesay, B. J., and Finger, J. T. (1996). Advanced drilling systems study (No.

SAND95-0331). Sandia National Laboratories, Albuquerque, New Mexico, United

States.

Pruess, K. (1991). TOUGH2: A general-purpose numerical simulator for multiphase fluid and
heat flow. Lawrence Berkeley Laboratory, Berkeley, California, United States.

REN21 (2015). Renewables 2015, Global Status Report, Key Findings. Renewable Energy
Policy Network for the 21st Century. Available at http://www.ren21.net.

Rybach, L., Mégel, T., and Eugster, W. (2000). At what time scale are geothermal resources
Chapter 1. Introduction and Motivation 20

renewable? In Proceedings World Geothermal Congress 2000, Kyushu - Tohoku, Japan,

May 28 - June 10, 2000, volume 2, pages 867–873.

Spitler, J. (2005). Editorial: Ground-source heat pump system researchpast, present, and
future. HVAC&R Research, 11 (2): 165–167.

Stutz, G. R., Shope, E., Aguirre, G. A., Batir, J., Frone, Z., Williams, M., Reber, T. J., Wheal-

ton, C. A., Smith, J. D., Richards, M. C., Blackwell, D. D., Tester, J. W., Stedinger, J. R.,
and Jordan, T. E. (2015). Geothermal energy characterization in the Appalachian Basin
of New York and Pennsylvania. Geosphere, 11 (5): 1291–1304.

Tester, J. W., Anderson, B., Batchelor, A., Blackwell, D., DiPippo, R., Drake, E., Garnish,

J., Livesay, B., Moore, M. C., Nichols, K., Petty, S., Toksz, M. N., and Veatch Jr., R. W.

(2006). The future of geothermal energy: Impact of enhanced geothermal systems (EGS) on the

United States in the 21st century. MIT.

Tester, J., Drake, E., Driscoll, M., Golay, M., and Peters, W. (2012). Sustainable Energy:

Choosing Among Options. MIT Press, 2nd edition.

Wong, Z. Y., Horne, R., and Voskov, D. (2015). A Geothermal Reservoir Simulator with

AD-GPRS. In Proceedings World Geothermal Congress 2015, Melbourne, Australia, 19-25

April 2015.

Yang, H., Cui, P., and Fang, Z. (2010). Vertical-borehole ground-coupled heat pumps: a
review of models and systems. Applied Energy, 87 (1): 16–27.
 CHAPTER 2
OBJECTIVES AND APPROACH

2.1 Dissertation Objectives

The objectives of this dissertation are twofold:

1. Assess the technical and economic performance of deep geothermal energy systems,

with a focus on low-temperature Enhanced Geothermal Systems (EGS) for electric-

ity, direct-use heat and cogeneration.

2. Assess the technical and economic performance of shallow geothermal energy sys-

tems, with a focus on hybrid geothermal heat pumps for cooling-dominated, non-

residential applications.

2.2 Dissertation Approach

The first objective, the topic of Part I of the dissertation, is addressed by developing a

simulation tool, called GEOPHIRES (Geothermal Energy for Production of Heat and elec-
tricity (“IR”) Economically Simulated). This tool incorporates reservoir, wellbore, and

power plant models, as well as economic models and capital and operation & mainte-
nance (O&M) cost correlations, in order to calculate the heat and/or electricity output
over the lifetime of the plant, and estimate the overall required investment and levelized
cost of electricity and/or heat. Different end-uses, i.e. direct-use heat, electricity, and
cogeneration, are evaluated for different resources grades and different technology as-
sumptions. Further, a comparison is made with other energy technologies to identify

challenges and opportunities for developing deep geothermal energy systems.

21
Chapter 2. Objectives and Approach 22

The second objective is the subject of Part II of the dissertation. It is approached by

developing validated system models using the software packages TRNSYS and MAT-
LAB, in order to conduct transient simulations of hybrid geothermal heat pump systems

and assess their energetic performance and costs over the lifetime of the system. The
models are validated with data from a full-scale experimental set-up of a hybrid shallow

geothermal system providing the climate control of a Verizon cellular tower shelter on
the Cornell University campus, as well as with data from literature and data collected at
a residential heat pump system in Lansing, NY. The Cornell-Verizon hybrid heat pump
system consists of six shallow geothermal boreholes, three heat pump units, a dry-cooler,

and air-economizer. Further, the computer models are used to assess the total cost of

ownership, and lifetime electricity consumption and CO2 emissions of various system
configurations.

For modeling both deep and shallow geothermal systems, accurate and efficient ther-

mal simulation of the subsurface reservoir is required. A novel approach for simulating

transient heat transfer with slender bodies in a conductive medium, using the slender-

body theory, is derived in Part III of the dissertation. Based on the theoretical framework,

a computationally-fast calculation tool is developed, which is validated with two case-

studies. The slender-body theory model allows for simulating complex slender geome-

tries, encountered in many geothermal applications (e.g. deviated geothermal wells and
slinky-coil heat exchangers), for which currently no standard heat transfer model exists

beyond finite element models, which are typically cumbersome and computationally-
intense. Part III contributes both to the first objective of the dissertation, e.g. for esti-
mating wellbore heat transmission with deep deviated geothermal wells which causes a
decrease in fluid production temperature and corresponding decrease in heat and elec-
tricity output, as well as to the second objective, e.g. for efficiently calculating g-functions
utilized in the TRNSYS borehole heat exchanger model.
 Part I

Techno-Economic Analysis of Deep

Geothermal Energy for Direct-Use Heat,

Electricity and Cogeneration

23
 CHAPTER 3
RETHINKING DEEP GEOTHERMAL ENERGY BEYOND ELECTRICITY

Given the benefits and challenges of deep geothermal, a call was made in the introduction
(Chapter 2), to consider this energy source not only for electricity production, but also for

direct-use heat and cogeneration. Iceland provides a successful example of an integrated

geothermal energy infrastructure, where deep geothermal supplies 62% of the country’s
primary energy demand including about 90% of its space heating demand (Tester et al.,
2015b). Nevertheless, only few countries are blessed with high-grade hydrothermal re-

sources like Iceland, and therefore the focus in Part I of the dissertation will be on extract-

ing deep geothermal through Enhanced Geothermal Systems (EGS) for various end-uses,
applicable nearly anywhere in the world.

When utilizing thermal energy of the geothermal fluid for supplying low-temperature

heat for e.g. district heating instead of electricity generation, one avoids low conversion

efficiencies and correspondingly large amounts of waste heat and poor system energetic

performance. In addition, lower geothermal fluid temperatures might be acceptable for

the direct-use application, which results in lower drilling depths, lower drilling costs,

and a decrease in risk and uncertainty. Figure 3.1 shows schematically an EGS scheme

producing both electricity in an Organic Rankine Cycle (ORC) or Kalina cycle (DiPippo,
2012), and direct-use heat for a district heating system.

Direct-use and cogeneration typically increase the techno-economic performance of a

deep geothermal energy system with respect to pure electricity generation. However, to

quantify by how much, careful modeling of the entire system including reservoir, well-
bore and surface plant, along with capital and O&M cost predictions for each system
component are required. Therefore, a software tool fulfilling this function is developed
in Chapter 4 and applied to various case-studies in Chapter 5 to estimate typical power

24
Chapter 3. Rethinking Deep Geothermal Energy Beyond Electricity 25

Figure 3.1 – Schematic diagram of deep geothermal energy extraction through EGS for cogeneration
(Goldstein et al., 2011).
Chapter 3. Rethinking Deep Geothermal Energy Beyond Electricity 26

outputs, required capital investment and cost-competitiveness of deep geothermal not

only for electricity, by also for direct-use heat and cogeneration. Also, a comparison with
alternatives, and take-away messages are provided.

References

DiPippo, R. (2012). Geothermal power plants: principles, applications, case studies and environ-

mental impact. Butterworth-Heinemann, 3rd edition.

Goldstein, B., Hiriart, G., Bertani, R., Bromley, C., Gutiérrez-Negrı́n, L., Huenges, E.,
Muraoka, H., Ragnarsson, A., Tester, J., and Zui, V. (2011). Geothermal Energy. In

Edenhofer, O., Pichs-Madruga, R., Sokona, Y., Seyboth, K., Matschoss, P., Kadner, S.,

Zwickel, T., Eickemeier, P., Hansen, G., Schlömer, S., and von Stechow, C., editors,

IPCC Special Report on Renewable Energy Sources and Climate Change Mitigation. Cam-

bridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Tester, J. W., Reber, T. J., Beckers, K. F., and Lukawski, M. Z. (2015). Deep geothermal

energy for district heating: Lessons learned from the U.S. and beyond. In Advanced

District Heating and Cooling (DHC) Systems. Woodhead Publishing.

 CHAPTER 4
GEOPHIRES SIMULATION TOOL

4.1 Motivation and History

Given the risks, uncertainties and upfront costs associated with deep geothermal projects,

it is necessary to model and simulate the reservoir, wellbores, and surface plant prior to
any site development. Predictions of both technical and economic performance are neces-
sary for comparing with other energy sources, selecting the best location for a new plant,

optimizing a system at a given location, developing an optimal reservoir management

strategy, or for estimating future behavior using historical operating data.

The Geothermal Energy Technology Evaluation Model (GETEM) (DOE, 2012) and

Hot Dry Rock economic (HDRec) (Heidinger et al., 2006) are two examples of techno-

economic models used to simulate deep geothermal systems. However, the focus of these

and other prior models is solely on generation of electricity, which may not be the most

appropriate end-use, at least not for low-grade geothermal. Therefore, a new software

tool is developed, called GEOPHIRES, that not only allows simulating production of

electricity but also production of direct-use heat and combined heat and power (CHP)
or cogeneration. GEOPHIRES is an acronym that stands for GEOthermal energy for the

Production of Heat and electricity (“IR”) Economically Simulated with “IR” representing

electric current and resistance and referring to the electricity mode.

The GEOPHIRES program is built upon previous research and models that date back
to the 1970’s Fenton Hill HDR (Hot Dry Rock) project at Los Alamos National Laboratory.
The work in the Los Alamos project resulted in the thermo-economic HDR model which
was used in the “Heat Mining” book by Armstead and Tester (1987). The HDR model
was upgraded in the late 1980’s culminating in the release of the MIT-HDR model (Tester

27
Chapter 4. GEOPHIRES Simulation Tool 28

and Herzog, 1990). Further upgrades were implemented to produce a Windows version
of the MIT-HDR model (Kitsou et al., 2000). In the 2000’s, the Windows version became
formally known as the “MIT-EGS” model, and was used in the “Future of Geothermal

Energy” study (Tester et al., 2006).

In 2012-2014, the “MIT-EGS” model was modified extensively to develop

GEOPHIRES. In addition to incorporating different end-uses beyond only electricity
production, GEOPHIRES includes new capital and operation&maintenance (O&M) cost
models such as the latest geothermal well drilling and completion cost correlations

(Lukawski et al., 2014), updated power plant models developed in AspenPlus and MAT-
LAB, and a new economic, reservoir, and wellbore model.

GEOPHIRES was first introduced at the Stanford Geothermal Workshop (Beckers

et al., 2013) and later described in detail in an article published in the Journal of Renew-

able and Sustainable Energy (Beckers et al., 2014a,b). This latter publication forms the

basis for most of the material presented in this chapter and the first case-study, utilizing

GEOPHIRES, in Chapter 5 (Section 5.1). The other two case-studies in Chapter 5 (Sections

5.2 and 5.3) are based on the journal paper by Reber et al. (2014) on the potential for deep

geothermal district heating systems in NY and PA, and the Stanford Geothermal Work-

shop paper by Beckers et al. (2015) on a hybrid geothermal-biomass cogeneration system

for Cornell University, respectively.

4.2 GEOPHIRES Model Structure

GEOPHIRES combines reservoir, wellbore, power plant, and economic models to sim-
ulate the electricity and/or direct-use heat production and predict the capital invest-
ment, O&M costs, and Levelized Cost of Electricity (LCOE) and/or Levelized Cost of
Chapter 4. GEOPHIRES Simulation Tool 29

Heat (LCOH). The LCOE and LCOH are expressed in 2012 U.S. ¢/kWhe and 2012 U.S.
¢/MMBTU (1 ¢/ kWhth = 2.931 $/MMBTU), respectively.

GEOPHIRES has 96 input parameters grouped into 7 categories: resource parameters,

engineering parameters, reservoir parameters, financial and operating parameters, cap-
ital cost parameters, O&M cost parameters and optimization parameters. The user can
choose to either perform a simulation of the geothermal system for a fully determined set
of parameters or optimize the system for a minimum LCOE or LCOH with respect to a
subset of parameters. GEOPHIRES can either be used as a stand-alone program through

a Graphical User Interface (GUI) or as a subroutine to be called from a user-developed

master program. The GUI is written in VB.net 9.0 and the underlying calculations are im-

plemented in FORTRAN 77 and 90. Figure 4.1 shows the GEOPHIRES operating scheme.

The GUI components are shown in orange ellipses and the FORTRAN components in

green rectangles. “Input.txt” and “Output.txt” are text files with respectively the input

parameters and output results.

4.3 Correlations and Component Models in GEOPHIRES

The next section provides a detailed description of the GEOPHIRES correlations and com-
ponent models. Section 4.3.1 explains in detail the 4 different available models imple-
mented in GEOPHIRES to simulate the thermal output of a geothermal reservoir. Consec-

utively, the production wellbore heat and pressure losses are calculated using a wellbore
model, outlined in Section 4.3.2. Next, the surface plant is modeled using correlations
provided in Section 4.3.3. The available end-use options are electricity, direct-use heat
or combined heat and power (cogeneration). The electricity and/or direct-use heat pro-
duction over the lifetime of the plant are used in conjunction with capital and O&M cost
predictions (Section 4.3.4) to calculate the levelized cost of energy. Three levelized cost
Chapter 4. GEOPHIRES Simulation Tool 30

Figure 4.1 – GEOPHIRES operating scheme. GUI components are shown with orange ellipses; the
FORTRAN model components are represented by green rectangles.

models, presented in Section 4.3.5, are available to estimate the LCOE and/or LCOH of

the system. In simulation mode, GEOPHIRES calculates the LCOE and/or LCOH in a

single run with a set of specified input parameters; in optimization mode, the calcula-

tions are iterated with updated parameter values using the NAG E04 Numerical Library
(NAG, 2011) until a minimum LCOE or LCOH is obtained.

For some components in GEOPHIRES, various models are available with varying
levels of detail and complexity. This allows the user to perform simulations of a deep
geothermal system with different levels of accuracy ranging from a rough estimate to an
in-depth technical and economic feasibility study of a specific project.
Chapter 4. GEOPHIRES Simulation Tool 31

4.3.1 Reservoir Models

GEOPHIRES has four built-in models to simulate the reservoir thermal drawdown: (1)
the Multiple Parallel Fractures Model, (2) the 1-D Linear Heat Sweep Model, (3) the m/A
Thermal Drawdown Parameter Model, and (4) the Percentage Temperature Drawdown
Model.

Multiple Parallel Fractures Model

The first model is the Multiple Parallel Fractures Model developed by Gringarten et al.

(1975) in which the reservoir is represented by an infinite series of parallel, equidistant

and vertical fractures with uniform aperture. The heat extraction occurs through thermal

convection with 1-D water flow in the fractures and thermal conduction within the homo-

geneous, isotropic, and impermeable rock. The reservoir outlet temperature is calculated

in the Laplace domain and numerically converted back into the time domain. The water

temperature T w and the fracture spacing S f r are non-dimensionalized as follows:

T r,0 − T w
T w,nd = (4.1)
T r,0 − T w,inlet
and
Sfr
S f r,nd = (4.2)
2b
with T r,0 the initial rock temperature at the point of injection (which is equal to the far-
field rock temperature at the same depth), T w,inlet the water inlet temperature, and b the

half-width aperture of the fracture. The solution for the dimensionless water outlet tem-
perature T w,outlet,nd in the Laplace domain is given by:
( )
1 ϕ ϕ ϕ
T̄ w,outlet,nd = 1+ exp(−β) + − (4.3)
s β s sβ
with
( )
S f r,nd − 1 √ √
β = tanh s s, (4.4)
χ
Chapter 4. GEOPHIRES Simulation Tool 32

ωH f r
ϕ= , (4.5)
T r − T w,inlet
2kr H f r
χ= . (4.6)
ρw cw vb
In these equations, s is the Laplace variable, kr is the thermal conductivity of the rock, ρw
is the density of the water, cw is the specific heat capacity of the water, ω is the geothermal
gradient, H f r is the height (distance along the flow path) of the fractures, and v is the
volume flow rate per fracture and per unit depth of the fracture. In order to use this
model, the user has to provide a full geometric description of the reservoir in terms of

number, spacing and one-side surface area of fractures.

1-D Linear Heat Sweep Model

The second model is the 1-D Linear Heat Sweep Model (Hunsbedt et al., 1984). It as-

sumes that the fractured Enhanced Geothermal System (EGS) reservoir can be modeled

as a porous medium with fluid surrounding blocks of rock characterized by a lumped

effective rock radius. The heat transfer occurs through 1-D linear heat sweep of water.

When non-dimensionalizing the water temperature using:

T w − T w,inlet
T w,nd = , (4.7)
T r,0 − T w,inlet
the solution for the water outlet temperature in the Laplace domain becomes:
1( )
T̄ w,outlet,nd = 1 − exp(−ϵ s) (4.8)
s
with
NT U
ϵ =1+ . (4.9)
Λ(s + NT U)
The outlet temperature is numerically converted back into the time domain using the
Stehfest algorithm (Hunsbedt et al., 1984). NT U and Λ are the number of heat transfer

units and the heat storage ratio, respectively. They are calculated as:
tres
NT U = (4.10)
τe f,r
Chapter 4. GEOPHIRES Simulation Tool 33

and
ρw cw Φ
Λ= . (4.11)
ρ r cr 1 − Φ
In these equations, Φ is the rock porosity, tres is the residence time of water in the reservoir
which can be calculated from the reservoir dimensions and fluid flow rate, and τe f,r is the
effective rock time constant. Typical porosities for fractured hard rocks in EGS reservoirs
are 1 to 7% (Tester et al., 2006); in GEOPHIRES the porosity for this equation is set at 4%.
Further, cr and ρr are the specific heat capacity and density of the rock, respectively. The

effective rock time constant is given by:

()
re2 f,r 1
τe f,r = 0.2 + . (4.12)
3α Bi

In this equation, re f,r is the effective rock radius representing the entire collection of rock

blocks, α the thermal diffusivity and Bi the Biot number. The value 0.2 represents the

typical ratio of conduction path length to re f,r for spherical shapes (Hunsbedt et al., 1984).

re f,r and Bi are calculated as:

( )1/3
3V
re f,r = 0.83 (4.13)
4π
and
hre f,r
Bi = (4.14)
kr
with V the mean volume of the rock blocks, and h the rock surface heat transfer coefficient.

This heat transfer coefficient will be relatively large but its exact value is unimportant,

assuming that the heat transfer in the reservoir is dominated by the thermal conduction
in the rock. Therefore in GEOPHIRES, a value of 500 W/(m2 ·K) is heuristically taken

for h. The value 0.83 in Equation (4.13) is the Kuo sphericity and represents the typical
mean sphericity of a distribution of irregularly shaped rock blocks found in geothermal
reservoirs (Hunsbedt et al., 1984).

In order to use the 1-D linear heat sweep model, the user has to provide reservoir
dimensions and properties, similar to the first model.
Chapter 4. GEOPHIRES Simulation Tool 34

m/ A Thermal Drawdown Parameter Model

The third model is the m/A Thermal Drawdown Parameter Model discussed by Armstead
and Tester (1987), which treats the reservoir as a single rectangular fracture of specified
area with a uniform flow of fluid passing over the fracture surface. The user has to pro-
vide the value for the mass loading parameter m/A along with the rock properties kr , ρr ,
and cr . The term m/A is defined as the mass flow rate per unit area A of a single side of the
fracture. The solution for the dimensionless water outlet temperature T w,outlet,nd is given

by:  √ 
T w,outlet − T w,inlet  1 1 kr ρr cr 
= = erf   (4.15)
t 
T w,outlet,nd
T r,0 − T w,inlet m/A cw
with erf the error function and t representing the time in seconds.

Percentage Temperature Drawdown Model

The fourth model is the Percentage Temperature Drawdown Model which expresses the

reservoir thermal drawdown in percentage temperature decrease per year. This model is

similar to the GETEM thermal drawdown model (DOE, 2012). With p being the percent-

age temperature decrease per year, the dimensionless water outlet temperature T w,outlet,nd
after t years is calculated as:

T w,outlet − T w,inlet ( p )t
T w,outlet,nd = = 1− . (4.16)
T r,0 − T w,inlet 100

Since the Percentage Temperature Drawdown Model is non-dimensionalized in

GEOPHIRES, the percentage temperature decrease per year p is independent of the units
of the outlet temperature (◦ C or K).

Which reservoir thermal model to select depends on how much information is avail-
able about the subsurface. The Multiple Parallel Fractures Model and the 1-D Linear Heat
Sweep Model are the most advanced models, but they also require detailed knowledge
Chapter 4. GEOPHIRES Simulation Tool 35

of the fracture and reservoir dimensions which is not always available. Also, a geome-
try of parallel and equidistant fractures could be too idealistic for a real EGS reservoir,
while modeling the heat extraction through heat sweep around spheres might be a better

fit for highly permeable and porous hydrothermal systems. If the user wants to imple-
ment a known temperature drawdown profile obtained e.g. through measurements of

an existing EGS or from a detailed subsurface reservoir model, the Percentage Tempera-
ture Drawdown Model may be the best suited. In cases with no prior knowledge of EGS
reservoir characteristics or only limited experimental data available, analyzing the ex-
pected drawdown for a range of m/A values with the m/A Thermal Drawdown Parameter

Model might be the best strategy.

Reservoir Pressure Drop Model

The user inputs in GEOPHIRES the reservoir impedance per well-pair in units of

GPa/(m3 /s) in order to calculate the reservoir pressure drop. A typical value would

be 0.15 GPa/(m3 /s) (Tester et al., 2006). To calculate the pumping power, this pressure

drop is added to the wellbore pressure losses and buoyancy effect (Section 4.3.2). No hy-

draulic drawdown is considered in GEOPHIRES, only thermal drawdown. This means

the mass flow rate per production well remains constant over the lifetime of the plant.

However, the circulation pump power will increase with thermal drawdown to offset the

decreasing buoyancy effect and maintain a constant mass flow rate.

4.3.2 Wellbore Models

There are two approaches available in GEOPHIRES to calculate the temperature drop
from bottom to top in the production wells. Either the user provides a constant tem-
Chapter 4. GEOPHIRES Simulation Tool 36

perature drop, or the temperature drop over time is predicted with Ramey’s wellbore
transient heat transmission model (1962). In Ramey’s model, the water temperature drop
∆T w is given by:
( ) ( ) ( −z )
rsv
∆T w = T r,0 − T w − ω (zrsv − Γ) + T w − ωΓ − T r,0 exp (4.17)
Γ
with T r,0 being the initial average rock temperature in the reservoir, zrsv the average depth
of the reservoir, and Γ a parameter defined as (assuming that the thermal resistances of
the well casing and cement are negligible):
mcw f (t)
Γ= . (4.18)
2πkr
In Equation (4.18), f (t) is the time function which for a line heat source is given by:
( )
rcas
f (t) = −ln √ − 0.29 (4.19)
2 αt
with rcas being the outside radius of the casing. For an outside casing diameter of 0.2 m or

less, the line source time function is expected to give accurate results after approximately

1 week of operation.

For geologic formations with multiple geothermal gradients, Equation (4.18) has to

be applied for each gradient. In such case, zrsv becomes the thickness of each gradient

segment, T w stands for the water outlet temperature from the previous segment and T r,0
is the initial or far-field rock temperature at the bottom of the considered segment.

The temperature increase in the injection well is neglected because the reservoir mod-
els in Section 4.3.1 assume a constant reservoir inlet temperature over time. Nevertheless,
a small change in reservoir inlet temperature causes only a negligible change in reservoir
outlet temperature. Therefore, this simplification has no impact on the final result.

The frictional pressure drop in the injection and production wells is calculated using
the Darcy-Weisbach equation (Fox et al., 2004):
L ρw U 2
∆p = f (4.20)
d 2
Chapter 4. GEOPHIRES Simulation Tool 37

with f being the Darcy friction factor, L the borehole length, d the borehole inner diameter
and U the average water velocity in the borehole. The Darcy friction factor is estimated it-
eratively with the Colebrook-White equation for turbulent flow in pipes (Fox et al., 2004):
 
 e/d 
√ 
1 2.51
√ = −2log10  + (4.21)
f 3.7 Re f

with e being the borehole pipe surface roughness and Re representing the Reynolds num-
ber. In GEOPHIRES, a surface roughness of well casing is set at 0.0001 m.

The buoyancy effect on the pressure drop is calculated as:

( )
∆P = gL ρw,prod − ρw,in j (4.22)

with g being the gravitational acceleration, L the wellbore length, and ρw,prod and ρw,in j the

average fluid density in the production and injection wellbores, respectively. The total

pressure drop the pump has to overcome is calculated by adding the reservoir pressure

drop to the injection wellbore, production wellbores and buoyancy correction pressure

drops.

4.3.3 Power Plant Models

The user can choose from three different end-use options: electricity, direct-use heat, and

combined heat and power (CHP).

Electricity Generation Model

In electricity mode, all the heat extracted from the geothermal fluid is utilized in a power
plant to generate electricity. This power plant can be either an Organic Rankine Cycle
(ORC) unit or a flash power plant (DiPippo, 2012). The electricity output W is calculated
Chapter 4. GEOPHIRES Simulation Tool 38

as:
W = ηu B (4.23)

where ηu is the utilization efficiency and B the exergy or availability of the geothermal
fluid. The exergy of geothermal fluid is calculated as:

( )
B = m h prod − h0 − T 0 (s prod − s0 ) (4.24)

with m, h, s and T 0 the geothermal fluid mass flow rate, its specific enthalpy, specific

entropy, and dead state (ambient) temperature, respectively.

The net electricity production is calculated by subtracting the geothermal fluid pump-

ing power from the electricity production calculated using Equation (4.23). The pumping
power is calculated as:
m∆P
W pump = (4.25)
ρw η pump
where ∆P is the total pressure drop as discussed in Section 4.3.2 and η pump is the efficiency

of the pump.

The utilization efficiency of binary cycle power plants is evaluated using models of a

conventional subcritical as well as a supercritical ORC created in Aspen Plus V7.0 soft-

ware. These models were modified from work done by Randall Field and Chad Augus-
tine (Augustine, 2009) and are continually being updated in our group (Lukawski, 2016).

Thermodynamic properties of working fluids are calculated based on the Benedict-Webb-

Rubin-Starling (BWRS) equation of state. The design specifications related to the heat
exchangers, turbines, and pumps are adopted from Augustine (2009). The model opti-
mization procedure maximized the cycle utilization efficiency and converged to a unique
solution by varying the turbine inlet and outlet pressures. In the supercritical cycle, which
has an additional degree of freedom, the working fluid mass flow rate is adjusted indi-
rectly by varying the temperature difference at the hot end of the primary heat exchanger.
In both ORC models, a heat recuperator is used, if necessary, to recover the heat from tur-
Chapter 4. GEOPHIRES Simulation Tool 39

bine exhaust fluid before it is passed to air-cooled condensers. In addition, the turbine
exhaust steam quality is limited to ≥0.9 to avoid erosion of the turbine blades. The drop
in efficiency due to wet expansion is calculated using the Baumann rule (DiPippo, 2012).

Additionally, the condensing pressure in the air cooled condensers is maintained above 1
bar to avoid contamination of the working fluid and its deterioration.

Both the power plant’s utilization efficiency ηu and the temperature of reinjected
geothermal fluid T w,in j are dependent on the temperatures of heat source T prod and heat
sink T 0 . Using the Aspen Plus models, ηu and T w,in j are evaluated for geothermal fluid

wellhead temperatures of 100-200◦ C and dry-bulb ambient temperatures of 0-30◦ C. For

each resource temperature, a working fluid maximizing the utilization efficiency is cho-

sen from 25 fluids listed in Table 4.1. Figure 4.2 shows the utilization efficiency of optimal

configurations of sub- and supercritical ORC as a function of geothermal fluid wellhead

temperature. They are implemented in GEOPHIRES using the following correlation:

ηu = K2 · T prod
2
+ K1 · T prod + K0 (4.26)

with the coefficients K provided in Table 4.2.

Single- and double-flash power plants are simulated in MATLAB using the methodol-

ogy described by DiPippo (2012). A dry-turbine isentropic efficiency of 85% is assumed.

The efficiency is corrected using the Baumann Rule if wet expansion occurred. The flash

pressures are chosen using the “equal-split temperature rule”. The condenser temper-

ature is heuristically set at 25◦ C above the ambient dry-bulb temperature. The models
assume all the geothermal fluid, including the condensed fluid after the turbine, is rein-
jected to obtain a closed system as desired for EGS. This explains why reinjection temper-
atures below 100◦ C can be obtained for flash power plants. The results for the utilization

efficiency (ηu ) for a single- or double-flash power plant as a function of the geothermal
production temperature (T prod ) and ambient temperature (T 0 ) are shown in Figure 4.2.
The correlation is given by Equation (4.26) with the coefficients K provided in Table 4.3.
Chapter 4. GEOPHIRES Simulation Tool 40

Table 4.1 – Working fluids considered in sub- and supercritical ORC power plant models. Molecular
masses (M) of these fluids are listed together with their critical temperatures (T cr ) and pressures (Pcr ).

Fluid Subcritical Supercritical M T cr (K) Pcr (MPa)

Propane No Yes 44.1 369.9 4.25
Isobutene Yes Yes 56.2 418.1 4.01
Isobutane Yes Yes 58.1 407.8 3.63
Butene Yes Yes 56.1 419.3 4.01
Butane Yes Yes 58.1 425.1 3.80
Isopentane Yes Yes 72.2 460.4 3.38
Pentane Yes Yes 72.2 469.7 3.37
Isohexane Yes Yes 86.2 497.7 3.04
Hexane Yes Yes 86.2 507.8 3.03
Cyclopropane Yes Yes 42.1 398.3 5.58
Cyclopentane Yes No 70.1 511.8 4.50
Cyclohexane Yes No 84.2 553.6 4.08
R32 No Yes 52.0 351.3 5.78
R152a Yes Yes 66.1 386.4 4.52
R134a Yes Yes 102.0 374.2 4.06
R245ca Yes Yes 134.1 447.6 3.93
R236ea Yes Yes 152.0 412.4 3.50
R227ea Yes Yes 170.0 376.0 2.93
R218 No Yes 188.0 345.0 2.64
R142b Yes Yes 100.5 410.3 4.06
R141b Yes No 117.0 477.5 4.21
Ammonia Yes No 17.0 405.4 11.33
Benzene Yes No 78.1 562.1 4.89
Toluene Yes No 92.1 591.8 4.13

Table 4.2 – Parameter values for ORC utilization efficiency correlation.

Subcritical Power Plants Supercritical Power Plants

T0 K2 K1 K0 K2 K1 K0
5 ◦C 0 2.746 e-3 -8.38 e-2 -1.550 e-5 7.604 e-3 -3.78 e-1
15 ◦ C 0 2.713 e-3 -9.18 e-2 -1.499 e-5 7.427 e-3 -3.79 e-1
25 ◦ C 0 2.676 e-3 -1.01 e-1 -1.550 e-5 7.550 e-3 -4.04 e-1

Table 4.3 – Parameter values for flash power plant utilization efficiency correlation.

Single-Flash Power Plants Double-Flash Power Plants

T0 K2 K1 K0 K2 K1 K0
5 ◦C -0.650e-6 0.969e-3 1.673e-1 -1.431e-6 1.334e-3 2.149e-1
15 ◦ C -0.837e-6 1.085e-3 1.449e-1 -1.680e-6 1.490e-3 1.883e-1
25 ◦ C -1.057e-6 1.221e-3 1.192e-1 -1.975e-6 1.673e-3 1.533e-1
Chapter 4. GEOPHIRES Simulation Tool 41

Figure 4.2 – Utilization efficiency ηu for different power plant types as a function of wellhead temper-
ature and dry-bulb ambient temperature.

These utilization efficiency correlations presented are implemented in GEOPHIRES.

Linear interpolation and extrapolation is used for different dry-bulb ambient tempera-

tures (T 0 ).

The utilization efficiency of ORC and flash power plants has been validated using ηu of

existing geothermal power plants (DiPippo, 2012, 2004), shown in Figure 4.2 with square
markers. Due to lack of data for supercritical ORC, the performance of the supercritical

ORC model was compared with advanced subcritical ORC plants (dual-level and dou-
ble pressure U.S. Geothermal Raft River plant and original dual-fluid Magmamax plants
(DiPippo, 2012)). The utilization efficiency of power plants in GEOPHIRES correlates
well with performance of existing plants. The slightly higher ηu in GEOPHIRES can be
Chapter 4. GEOPHIRES Simulation Tool 42

Figure 4.3 – Geothermal fluid exit temperature for different power plant types as a function of wellhead
temperature and dry-bulb ambient temperature.

explained by: (1) not including parasitic power losses for downhole pumps in ηu of our

models as it is subtracted later by the program, (2) incorporating heat recuperators in all
Aspen Plus ORC models, (3) no non-condensable gases present in geothermal fluid in

GEOPHIRES, (4) many existing power plants have been designed decades ago, and (5)

no constraint on the minimum temperature of geothermal fluid leaving the power plant
was assumed in GEOPHIRES.

The temperature of the geothermal fluid leaving the power plant as a function of
the ambient and production temperature is shown in Figure 4.3. As mentioned in Sec-
tion 4.3.2, a constant reinjection temperature is required for the reservoir models. The
user provides this temperature but GEOPHIRES verifies it is lower than the geothermal
fluid temperature leaving the power plant. In the case-studies conducted in the next
Chapter 4. GEOPHIRES Simulation Tool 43

chapter, only subcritical ORC and double-flash power plants are considered because the
capital cost correlations are developed for these types of power plants only. The levelized
cost is calculated as levelized cost of electricity (LCOE) in 2012 U.S. ¢/kWhe .

Direct-Use Heat Generation Model

In direct-use heat mode, all the heat extracted from the geothermal fluid is used in a
direct-use heat application such as an industrial process or a district heating system. No
specific process models are built-in into GEOPHIRES but the user can develop its own
heat exchanger models, coupled to GEOPHIRES. An example of using GEOPHIRES in a

geothermal district heating study is done in the work by Reber (2013), further discussed

in Chapter 5.

The pumping power electricity consumption is modeled as an operating expense with

user-defined electricity cost. The economic feasibility of direct-use geothermal projects is

evaluated in terms of levelized cost of heat (LCOH) and expressed in 2012 $/MMBTU.

Cogeneration or Combined Heat and Power (CHP) Models

In CHP mode, three different configurations are available: a topping cycle, a bottoming

cycle and a parallel cycle. In a topping cycle, a high temperature heat-to-power conver-
sion cycle is in series with a low-temperature direct-use heat process. In less common
bottoming cycles the geothermal fluid first goes through a high temperature direct-use
heat process and consecutively a low-temperature heat-to-power conversion process. In
a parallel cycle, the geothermal fluid splits into two streams serving a heat-to-power and
direct-use heat process at the same temperature.

In a topping cycle, the temperature of the geothermal fluid leaving the power plant
Chapter 4. GEOPHIRES Simulation Tool 44

is assumed to be the input temperature for the direct-use heat process. In a bottoming
cycle, the user provides the temperature of the geothermal fluid leaving the direct-use
heat process, which is assumed to be the fluid temperature entering the power plant. In a

parallel cycle, the user provides the mass fraction of geothermal fluid used for electricity
production. In each configuration, the geothermal fluid pump power is subtracted from

the gross electricity production to calculate the net electricity production.

The levelized cost in CHP mode can be calculated in GEOPHIRES either as LCOE

using the heat sales as operating income with a user-defined heat price ($/MMBTU), or as
LCOH using the electricity sales as operating income with a user-defined electricity price
(¢/kWhe ). The program can also evaluate both the LCOE and LCOH with the fraction of

the shared costs attributed to the electricity and direct-use heat production based on their

respective use of thermal energy contained in the geothermal fluid.

4.3.4 Cost Correlations

GEOPHIRES has built-in correlations to estimate the capital costs and operation & main-

tenance (O&M) costs of EGS projects. The user can use these cost correlations directly,

multiply them by a certain factor or provide their own costs.

Capital Cost Correlations

The capital costs (Ccap ) are calculated as the sum of the geothermal well drilling and
completion costs (Ccap,well ), the surface plant costs (Ccap,pp ), the reservoir stimulation costs
(Ccap,stim ), the fluid distribution costs (Ccap,distr ) and the resource exploration costs (Ccap,expl ):

Ccap = Ccap,well + Ccap,pp + Ccap,stim + Ccap,distr + Ccap,expl . (4.27)

Chapter 4. GEOPHIRES Simulation Tool 45

The geothermal well drilling and completion costs are calculated as (in 2012 U.S. M$)
(Lukawski et al., 2014):
( )
Ccap,well = N · Ccap,1well = N · 1.72 · 10−7 · MD2 + 2.3 · 10−3 · MD − 0.62 (4.28)

with N the number of wells drilled (assuming all identical), and MD the measured depth
of one well (i.e. length of the wellbore along path) in meters. In comparison with the 2004

drilling cost correlation in the Future of Geothermal Energy report (Tester et al., 2006),
the drilling costs have been increased 30 to 80% as a function of depth with an average of
about 50%.

The power plant capital cost correlations are presented in Figure 4.4. The correlations
for Organic Rankine Cycle (ORC) power plants are based on the revised cost correla-

tion by Mines (2008), the correlations used in the Job and Economic Development Impact

(JEDI) Model (Johnson et al., 2012) and the correlations used in the Future of Geothermal

Energy report (Tester et al., 2006). The cost data is brought to 2012 U.S. $ using the IHS

CERA Power Capital Costs Index (PCCI) (IHS, 2013). The correlation for a 15 MWe ORC

is given by (in $):

 ( )


 ◦
 W · K3 · T prod + K2 · T prod + K1 · T prod + K0 if T prod < 170 C
 3 2
Ccap,pp = 
 ( ( )) (4.29)


 W · 2,212.5 − 0.514 · T prod − 170◦ C if T prod ≥ 170◦ C

with 





 K3 = 1.45833 e-3







 K2 = 7.6875 e-1



(4.30)


 K1 = −1.34792 e2







 K0 = 1.0075 e4
and W the nominal power output (installed capacity) expressed in kWe . For a power plant
with output different than 15 MWe , the capital cost is calculated as the cost for a 15 MWe
ORC multiplied with a correction factor:
( )( W
)−0.06
Ccap,pp = Ccap,pp for 15,000 kWe (4.31)
15,000 kWe
Chapter 4. GEOPHIRES Simulation Tool 46

with W expressed in kWe . For a double-flash power plant, the capital costs correlations
are derived from 2012 GETEM capital cost data (DOE, 2012). GEOPHIRES calculates the
costs as:

K2 · T prod
2
+ K1 · T prod + K0 (4.32)

with parameters K specified in Table 4.4.

Table 4.4 – Parameter values for double-flash power plant capital cost correlation.

Wre f K2 K1 K0
5 MWe 4.847 e-2 -35.219 8.447 e3
10 MWe 4.060 e-2 -29.382 6.991 e3
25 MWe 3.277 e-2 -23.552 5.526 e3
50 MWe 3.472 e-2 -23.814 5.179 e3
75 MWe 3.527 e-2 -24.396 5.197 e3
100 MWe 3.391 e-2 -23.489 5.024 e3

For installed capacities different than those given above, the capital costs are estimated

using interpolation/extrapolation with a power law curve. The power law curve is fitted

to the lower and upper bound of the bin containing the actual power output W for inter-

polation or to the two closest values (5 and 10 MWe or 75 and 100 MWe ) for extrapolation.

In direct-use heat mode, the surface plant equipment cost strongly depends on the

application. The built-in correlation in GEOPHIRES is (in $):

Ccap,pp = 150 · W (4.33)

with W the nominal thermal output (installed capacity) in kWth . In CHP mode, the power

plant capital investment is the sum of the electric power plant and direct-use heat plant
costs, which are calculated using the equations above.

For the reservoir stimulation costs, it is assumed that only the injection well is stimu-
lated with a flat rate of $2.5M per injection well (Mines and Nathwani, 2013). The maxi-
mum number of production wells per injection well is set at four.
Chapter 4. GEOPHIRES Simulation Tool 47

Figure 4.4 – Capital cost correlations for subcritical ORC and double-flash power plants as a function
of initial wellhead temperature (production temperature) and installed generating capacity.

The fluid distribution costs are the capital costs related to the surface brine gathering

system and are estimated as (in $):

Ccap,distr = 50 · W (4.34)

where W is the thermal energy extracted from geothermal fluid in kWth . These expen-

ditures cover only the piping between production and injection wells and the plant. In
the case of a geothermal district heating system, the user has to provide additional cost
figures to account for these expenses.

The resource exploration cost correlation in GEOPHIRES is loosely based on the cor-
relation used in GETEM (DOE, 2012) and is given by:

Ccap,expl = 1.12 · ($1M + 0.6 · C1 well ) (4.35)

Chapter 4. GEOPHIRES Simulation Tool 48

which assumes one slim-hole well is drilled at 60% of the cost of a regular well (pro-
duction or injection well), $1M are the expenses to cover non-drilling activities such as
geophysical surveys and field work, and the factor 1.12 accounts for the technical and of-

fice support expenses. This correlation is arbitrary and will not represent the exploration
cost of every EGS project. Depending on the success rate of the exploration wells and

whether the site is in a “green field” or “brown field”, different projects require different
non-drilling related work and a different number of exploration wells. Further, the cost of
each exploration well depends on whether it is a core-hole, slim-hole or production-size
well. Finally, some projects require drilling of confirmation wells.

O&M Cost Correlations

The annual operation and maintenance (O&M) costs are calculated as the sum of the

power plant O&M costs (CO&M,pp ), wellfield O&M costs (CO&M,w f ) and make-up water costs

(CO&M,w ):

CO&M = CO&M,pp + CO&M,w f + CO&M,w . (4.36)

To estimate the power plant O&M costs, it is assumed they consist of 75% of the total

labor costs and 1.5% of the upfront power plant capital costs, a similar approach as used

in the GETEM program (DOE, 2012) (in $/year):

CO&M,pp = 0.75 · Clabor + 1.5% · Ccap,pp (4.37)

where Clabor is the annual labor cost, calculated in electricity mode using the following
correlation based on the GETEM correlation(DOE, 2012) (in k$):





 236 if W < 2.5 MWe
Clabor = 
 (4.38)


 589 · ln(W) − 304 if W ≥ 2.5 MWe
Chapter 4. GEOPHIRES Simulation Tool 49

where W is the nominal output of the plant in MWe and ln the natural logarithm. In
direct-use heat and CHP mode, the correlation for the labor cost is (in k$):





 236 if W < 12.5 MWth
Clabor = 
 (4.39)


 589 · ln (W/5) − 304 if W ≥ 12.5 MWth

where W is the amount of heat withdrawn from the geothermal fluid in MWth .

The wellfield O&M costs consist of the remaining 25% of the labor costs and 1% of the
total well capital costs, again similar to the correlation in GETEM (DOE, 2012) (in $/year):

CO&M,w f = 0.25 · Clabor + 1% · Ccap,well . (4.40)

In the case of reservoir temperature drawdown beyond a user-defined threshold,

redrilling is required. The capital cost for these wells are modeled as O&M costs by

dividing their cost by the lifetime of the plant and adding them to the wellfield O&M

costs.

The user has to input the water loss rate occurring in the EGS reservoir. The annual

expenses to pay for the make-up water are calculated using a water rate of $660/Ml ($2.5

per 1000 gallons).

All the capital and O&M cost correlations provided in this section come with inherent
uncertainties. To address these uncertainties, a sensitivity analysis is performed when

evaluating the levelized costs for various EGS scenarios in Section 5.1.

4.3.5 Levelized Cost Models

Three models are available in GEOPHIRES to calculate the levelized cost of electricity
(LCOE) or heat (LCOH): (1) the Fixed Charge Rate Model, (2) the Standard Levelized
Cost Model, and (3) the BICYCLE Levelized Life Cycle Cost Model.
Chapter 4. GEOPHIRES Simulation Tool 50

Fixed Charge Rate (FCR) Model

The first model is the Fixed Charge Rate (FCR) Model in which the LCOE/LCOH is cal-
culated as (Armstead and Tester, 1987):
FCR · Ccap + CO&M
LCOE or LCOH = (4.41)
E

with Ccap the total capital investment, CO&M the yearly operation and maintenance (O&M)
cost and E the average annual electricity output in kWhe or direct-use heat generation in
MMBTU. The FCR is a single number, e.g. 12.8% in the National Energy Modeling System
(Tester et al., 2006), which is multiplied by the total capital cost to obtain the annual cost

of invested capital. The FCR is estimated based on several financial parameters, e.g. debt

interest rate, rates of return on equity capital, and tax rates (Armstead and Tester, 1987):

FCR = CRF + f (CRF(i, n), tax credits) (4.42)

with CRF the capital recovery factor. The CRF is the fraction of capital investment that

must be paid back every year to fully repay a loan within n years at an interest rate i. It is

calculated as (Armstead and Tester, 1987):

i
CRF(i, n) = . (4.43)
1 − (1 + i)−n

Standard Levelized Cost Model

The second model is the Standard Levelized Cost Model which discounts both expendi-
tures and revenues to current day dollars (OECD/IEA, 2010):
∑ C −It
Ccap + ltt=1 O&M,t
(1+i)t
LCOE or LCOH = ∑n E t (4.44)
t=1 (1+i)t

with lt being the lifetime of the plant, Ccap the capital investment, and CO&M,t , It and Wt
representing the O&M cost, revenue from heat or electricity sales in CHP mode, and
electricity or direct-use heat generation in year t, respectively.
Chapter 4. GEOPHIRES Simulation Tool 51

BICYCLE Levelized Cost Model

The third and most advanced economic model is the BICYCLE Levelized Life Cycle Cost
Model developed at Los Alamos National Laboratory (Hardie, 1981). This model assumes
that the financing of the EGS project occurs through debt (bonds) and equity and allows
for variable debt/equity return rates. The ratio of outstanding debt to outstanding equity
remains constant while paying off over the lifetime of the plant. Furthermore, the model
takes into account inflation, tax rates and tax credits. The levelized cost is calculated as:

NPV
LCOE or LCOH = ∑ Wt ·(1+iin f )t
(4.45)
lt
t=1 (1+iave )t

where NPV is the net present value, iin f the inflation rate and iave the average return on

investment (tax and inflation adjusted), calculated as:

iave = db · idb · (1 − iit ) + eq · ieq (4.46)

with db, eq, idb , iit and ieq the fraction of investment through debt, the fraction of invest-

ment through equity (eq = 1 − db), the inflated debt interest rate, income tax rate, and the

inflated equity return rate, respectively.

The NPV in Equation (4.45) is calculated as:

NPV = NPVcap + NPVO&M + NPV f c + NPVit + NPVgrt − NPVitc (4.47)

with NPVcap , NPVO&M , NPV f c , NPVit , NPVgrt and NPVitc representing the net present value
contribution due to the capital costs, O&M costs, fixed charges, income taxes, gross rev-

enue taxes, and investment tax credits. They are calculated as follows:
lt (
∑ )
Ccap · CRF(iave , n)
NPVcap = , (4.48)
t=1
(1 + iave )t

lt (
∑ )
CO&M,t · (1 + iin f )t
NPVO&M = , (4.49)
t=1
(1 + iave )t
Chapter 4. GEOPHIRES Simulation Tool 52

lt (
∑ )
Ccap · i pt · (1 + iin f )t
NPV f c = , (4.50)
t=1
(1 + iave )t
( ) ( )
∑
lt  iit
 1−iit Ccap · CRF(iave , lt) − Ccap lt 
1
NPVit =   , (4.51)
t=1
(1 + iave )t
( )
igrt
NPVgrt = (NPVcap + NPVO&M + NPV f c + NPVit − NPVitc ) , (4.52)
1 − igrt
Ccap · iitc
NPVitc = (4.53)
1 − iit
with iit , igrt , i pt , iin f and iitc representing the income tax rate, gross revenue tax rate, property
tax rate, inflation rate and investment tax credit rate, respectively. In Equation (4.51),

the term (Ccap /lt) represents linear depreciation. Further, the term Ccap represents the

overnight capital cost which can be adjusted for inflation during construction.

Which economic model to use depends on how rigorous the economic analysis should

be. The BICYCLE model is believed to more closely represent market conditions, but

requires a good understanding and knowledge of the different economic parameters. The

FCR and Standard Levelized Cost Model on the other hand are easier to understand and

to apply but may be less realistic.

References

Armstead, H. C. H. and Tester, J. W. (1987). Heat Mining. E. & F.N. Spon Ltd., London and

New York.

Augustine, C. R. (2009). Hydrothermal spallation drilling and advanced energy conversion

technologies for Engineered Geothermal Systems. PhD Dissertation, Massachusetts Institute
of Technology, Cambridge, Massachusetts, United States.

Beckers, K. F., Lukawski, M. Z., Reber, T. J., Anderson, B. J., Moore, M. C., and Tester,
J. W. (2013). Introducing GEOPHIRES v1.0: Software Package for Estimating Levelized
Chapter 4. GEOPHIRES Simulation Tool 53

Cost of Electricity and/or Heat from Enhanced Geothermal Systems. In Proceedings

Thirty-Eighth Workshop on Geothermal Reservoir Engineering, Stanford University, Stan-
ford, California, February 11 - February 13, 2013, SGP-TR-198.

Beckers, K. F., Lukawski, M. Z., Anderson, B. J., Moore, M. C., and Tester, J. W. (2014a).

Levelized costs of electricity and direct-use heat from Enhanced Geothermal Systems.
Journal of Renewable and Sustainable Energy, 6, 013141.

Beckers, K. F., Lukawski, M. Z., Anderson, B. J., Moore, M. C., and Tester, J. W. (2014b).
Erratum: “Levelized costs of electricity and direct-use heat from Enhanced Geothermal

Systems” [J. Renewable Sustainable Energy 6, 013141 (2014)]. Journal of Renewable and
Sustainable Energy, 6, 059902.

Beckers, K. F., Lukawski, M. Z., Aguirre, G. A., Hillson, S. D., and Tester, J. W. (2015). Hy-

brid Low-Grade Geothermal-Biomass Systems for Direct-Use and Co-Generation: from

Campus Demonstration to Nationwide Energy Player. In Proceedings Fortieth Workshop

on Geothermal Reservoir Engineering, Stanford University, Stanford, California, January

26 - January 28, 2015, SGP-TR-204.

DiPippo, R. (2004). Second law assessment of binary plants generating power from low-

temperature geothermal fluids. Geothermics, 33 (5): 565–586.

DiPippo, R. (2012). Geothermal power plants: principles, applications, case studies and environ-
mental impact. Butterworth-Heinemann, 3rd edition.

DOE (2012). Geothermal Electricity Technology Evaluation Model (GETEM), Version Au-
gust 2012. U.S. Department of Energy Geothermal Technologies Program. Available
at http://www1.eere.energy.gov/geothermal/getem.html.

Fox, R. W., Pritchard, P. J., and McDonald, A. T. (2004). Introduction to Fluid Mechanics.
John Wiley & Sons, 6th edition.
Chapter 4. GEOPHIRES Simulation Tool 54

Gringarten, A. C., Witherspoon, P. A., and Ohnishi, Y. (1975). Theory of heat extraction
from fractured hot dry rock. Journal of Geophysical Research, 80 (8): 1120–1124.

Hardie, R. W. (1981). BICYCLE II: A Computer Code for Calculating Levelized Life-Cycle Costs,
LA-89089. Los Alamos National Laboratory, Los Alamos, New Mexico, United States.

Heidinger, P., Dornstädter, J., and Fabritius, A. (2006). HDR economic modelling: HDRec

software. Geothermics, 35 (5): 683–710.

Hunsbedt, A., Lam, S. T.-F., and Kruger, P. (1984). User’s manual for the one-dimensional
linear heat sweep model. Stanford Geothermal Program, Interdisciplinary Research in En-

gineering and Earth Sciences, Stanford University, Stanford, California, United States.

IHS (2013). IHS CERA Power Capital cost Index (PCCI). Available at http://

www.ihs.com/info/cera/ihsindexes/index.aspx.

Johnson, C., Augustine, C., and Goldberg, M. (2012). Jobs and Economic Development Impact

(JEDI) Model Geothermal User Reference Guide, Technical Report NREL/TP-6A20-55781.

National Renewable Energy Laboratory (NREL), Golden, Colorado, United States.

Kitsou, O. I., Herzog, H. J., and Tester, J. W. (2000). Economic modeling of HDR enhanced

geothermal systems. In Proceedings World Geothermal Congress 2000, Kyushu - Tohoku,

Japan, May 28 - June 10, 2000, pages 3779–3784.

Lukawski, M. Z., Anderson, B. J., Augustine, C., Capuano, L. E., Beckers, K. F., Livesay, B.,
and Tester, J. W. (2014). Cost analysis of oil, gas, and geothermal well drilling. Journal
of Petroleum Science and Engineering, 118: 1–14.

Lukawski, M. Z. (2016). Comparison of Subcritical and Supercritical Organic Rankine Cycles

(ORC) for Efficient Conversion of Low- and Medium-Temperature Heat to Electricity. Internal
Report, Cornell Energy Institute, Cornell University, Ithaca, New York, United States.
Chapter 4. GEOPHIRES Simulation Tool 55

Mines, G. (2008). Geothermal Electricity Technologies Evaluation Model DOE tool for
Assessing Impact of Research on Cost of Power. In Thirty-Third Workshop on Geothermal
Reservoir Engineering, Stanford University, Stanford, California, January 28 - January 30,

2008.

Mines, G. and Nathwani, J. (2013). Estimated power generation costs for EGS. In Pro-
ceedings Thirty-Eighth Workshop on Geothermal Reservoir Engineering, Stanford University,
Stanford, California, February 11 - February 13, 2013, SGP-TR-198.

NAG (2011). NAG Fortran Library Manual - Mark 23. The Numerical Algorithms Group

Limited.

OECD/IEA (2010). Projected Costs of Generating Electricity: 2010 Edition. OECD NEA/IEA,

Organisation for Economic Co-operation and Development Nuclear Energy Agency /

International Energy Agency.

Ramey Jr, H. J. (1962). Wellbore heat transmission. Journal of Petroleum Technology, 14 (04):

427–435.

Reber, T. J. (2013). Evaluating Opportunities for Enhanced Geothermal System-Based District

Heating in New York and Pennsylvania. Master Thesis, Cornell University, Ithaca, New
York, United States.

Reber, T. J., Beckers, K. F., and Tester, J. W. (2014). The transformative potential of geother-
mal heating in the US energy market: A regional study of New York and Pennsylvania.

Energy Policy, 70: 30–44.

Tester, J. W. and Herzog, H. J. (1990). Economic Predictions for Heat Mining: A Review and
Analysis of Hot Dry Rock (HDR) Geothermal Energy Technology, Massachusetts Institute
of Technology Energy Laboratory Report MIT-EL 90-001. U.S. Department of Energy,
Geothermal Technology Division.
Chapter 4. GEOPHIRES Simulation Tool 56

Tester, J. W., Anderson, B., Batchelor, A., Blackwell, D., DiPippo, R., Drake, E., Garnish,
J., Livesay, B., Moore, M. C., Nichols, K., Petty, S., Toksz, M. N., and Veatch Jr., R. W.
(2006). The future of geothermal energy: Impact of enhanced geothermal systems (EGS) on the

United States in the 21st century. MIT.

 CHAPTER 5
GEOPHIRES CASE-STUDIES

This chapter presents three case-studies utilizing GEOPHIRES to investigate the tech-
nical and economic performance of deep geothermal energy using Enhanced Geother-

mal Systems (EGS) for electricity and direct-use heat (Section 5.1), district heating sys-
tems (Section 5.2), and cogeneration (Section 5.3). The sections are based on material
published in (Beckers et al., 2014a,b), (Reber et al., 2014), and (Beckers et al., 2015), re-

spectively. Other published works utilizing GEOPHIRES, not further discussed here, are
(Tester et al., 2015a) and (Tester et al., 2015b).

5.1 Case-Study 1: Deep Geothermal for Electricity and Direct-Use Heat

5.1.1 Parameter Values for EGS Scenarios

Three different technology maturity levels (today’s, mid-term, and commercially ma-

ture technology), are combined with three different resource quality levels (low-grade,

medium-grade and high-grade) to estimate ranges of levelized costs, capital investment

and operation & maintenance (O&M) costs of EGS plants. All 9 proposed cases are eval-
uated for both electricity and direct-use heat resulting in 18 EGS scenarios in total. Tables
5.1 and 5.2 contain the values of technology, economic, and resource parameters used in
this study.

The main differences between the three considered technology maturity levels are

reservoir productivity rates scaled by the geothermal water mass flow rate per produc-
tion well and the reservoir temperature decline or drawdown rate. The today’s technol-
ogy case assumes a well productivity of 30 kg/s and a 2% temperature drop per year (see

57
Chapter 5. GEOPHIRES Case-Studies 58

Table 5.1 – Technical and economic parameter values for EGS Scenarios.

Commercially
Today’s Mid-Term
Parameter Mature
Technology Technology
Technology
Reservoir Thermal Model Percentage Temperature Drawdown
Reservoir Temp. Drawdown 2%/year 1.5%/year 1%/year
Fluid Temp. Drawdown Threshold 19% 21% 14%
Geofluid Flow Rate per producer 30 kg/s 50 kg/s 70 kg/s
Pump Electr. Rate in Direct-Use 7 ¢/kWhe 7 ¢/kWhe 7 ¢/kWhe
Drilling Costs Multiplier (Eq. (4.28)) 100% 90% 80%
Hydraulic Impedance per Well-Pair 0.15 MPa·s/L 0.15 MPa·s/L 0.15 MPa·s/L
Wellbore Thermal Model Ramey Ramey Ramey
◦ ◦
Maximum Borehole Temp. 350 C 375 C 400◦ C
Water Loss Rate 2% 2% 2%
Geofluid Pump Efficiency 80% 80% 85%
Well Casing Inner Diameter 0.2 m 0.2 m 0.22 m
Levelized Cost Model BICYCLE BICYCLE BICYCLE
Discount Rate 7% 7% 7%
Inflation Rate 2% 2% 2%
Income Tax Rate 39.2% 39.2% 39.2%
Plant Lifetime 30 years 30 years 30 years
Capacity Factor 90% 90% 95%

Table 5.2 – Resource parameter values for EGS scenarios with E and D-U referring to electricity and
direct-use, respectively.

Parameter Low-Grade Medium-Grade High-Grade

Ambient Temp. 15◦ C 15◦ C 15◦ C
Geothermal Grad. 30◦ C /km 50◦ C /km 70◦ C /km
E D-U E D-U E D-U
Well Depth 6 km 4 km 5 km 2.5 km 4 km 1.8 km
◦
Bottom Hole Temp. 195 C 135◦ C ◦
265 C 140◦ C ◦
295 C 141◦ C
Power Plant Type ORC N/A Flash N/A Flash N/A
◦
Injection Temp. 70 C 50◦ C ◦
75 C 50◦ C ◦
80 C 50◦ C
No. of Inj. Wells 1 1 1 1 1 1
No. of Prod. Wells 2 2 4 2 4 2
Chapter 5. GEOPHIRES Case-Studies 59

Section 4.3.1 for more information on the Percentage Temperature Drawdown Model).
The mid-term technology and commercially mature technology cases assume a produc-
tion well flow rate of 50 and 70 kg/s and a percentage temperature drop per year of 1.5%

and 1%, respectively. To obtain both a lower annual temperature drop and an increase in
well flow rate with improving technology requires that also the surface area of the frac-

tures increases (larger reservoir). Other differences between the three technology cases
include well drilling and completion costs, circulation pump efficiency, and maximum
allowable bottom hole temperature. The latter is determined by the maximum allowable
operating conditions for downhole tools and completion equipment and is assumed to

improve from 350◦ C for today’s technology to 400◦ C for commercially mature technology

(Tester et al., 2006). Further, in contrast to the production temperature decline, no decline
in production well flow rate (hydraulic drawdown) is assumed for EGS in GEOPHIRES.

However, pump power requirements do change over time due to changes in buoyancy

effects (see Section 4.3.1). Ramey’s model is used to estimate the temperature drop due

to heat losses in the production wells (Section 4.3.2). The “Fluid Temperature Drawdown

Threshold” refers to the thermal drawdown rate that triggers redrilling of all the wells.

In every scenario, both for electricity and direct-use heat, this parameter is chosen so

that redrilling is performed once at the mid-life point of the plant, which would make

most economic sense. It is expected that the required improvements between today’s and
mid-term technology and between mid-term and commercially mature technology are
evolutionary and do not rely on any fundamental technological breakthroughs.

The low-grade, medium-grade and high-grade resource cases have average geother-
mal gradients of 30◦ C/km, 50◦ C/km and 70◦ C/km, respectively. While temperature gra-
dients of up to 30◦ C/km are typical for the Eastern U.S., 50◦ C/km is commonly available
in the Western U.S. (Tester et al., 2006). Gradients of 70◦ C/km or more can be found in
some of the “hot spots”, for example at existing hydrothermal sites in the Western U.S.
(e.g. the Geysers Field in California) or areas of relatively recent volcanism (e.g. the New-
Chapter 5. GEOPHIRES Case-Studies 60

berry Caldera in Oregon). For the electricity scenarios, an Organic Rankine Cycle (ORC)
power plant with one injection and two production wells and a well depth of 6 km (195◦ C)
is chosen for the low-grade resource. For the medium- and high-grade resources, double-

flash power plants with one injection and four production wells are used and the drilling
depths are respectively 5 km (265◦ C) and 4 km (295◦ C). For the direct-use heat scenarios,

the bottom hole temperature is set at approximately 140◦ C in each case which translates
into drilling depths of 4, 2.5 and 1.8 km for the low-, medium- and high-grade resources,
respectively.

The BICYCLE levelized cost model is used for the calculation of the levelized cost of
electricity (LCOE) or levelized cost of heat (LCOH) in each scenario. A discount rate of

7%, lifetime of 30 years and income tax rate of 39.2% is assumed for each case in order

to allow comparison with reported values for other energy technologies from the OpenEI

Transparent Cost Database (U.S. DOE, 2013). In addition, capacity factors of 90 to 95%

are assumed depending on the technology case, for both the electricity and direct-use

heat scenarios. The capacity factor is defined as the percentage of time the plant runs in

baseload operation which is very similar to the percentage of the actual net electricity gen-

eration to the maximum possible net electricity generation within e.g. one year. With this

definition, the capacity factor is equivalent to the availability of the plant. In GEOPHIRES,

the EGS plant is either on in baseload or off for maintenance; variable geothermal fluid
flow rates for load-following operation are not considered. Other definitions of capacity
factor are found in the literature; applying those to these cases would not necessarily lead

to the same values for capacity factor as those in Table 5.1. The relatively high capac-
ity factors for direct-use heat applications would apply only for industrial or commercial
processes that can operate continuously and constantly use the delivered heat. Some ex-
amples are paper processing, food pasteurization, and biomass drying; other examples
are provided in Section 5.1.3. In contrast, geothermal district heating systems for residen-
tial and commercial buildings operate at lower capacity factors because heat is required
Chapter 5. GEOPHIRES Case-Studies 61

only on cold days. Reber (2013) conducted a separate study using GEOPHIRES focusing
on these systems using a capacity factor of 50% for heating only, and developed a geother-
mal district heating supply curve for the states of New York and Pennsylvania. Some of

his key findings are included in the discussion of results (Section 5.1.3) and further pre-
sented in Section 5.2.

The capital investment and levelized costs calculated in this paper for both the elec-
tricity and direct-use heat cases do not include transmission or distribution costs, only
energy extraction and conversion costs. In the energy conversion process, it is assumed

that all the enthalpy available between the production and injection temperature is uti-
lized either to generate electricity (with a certain utilization efficiency) in the electricity

cases or directly as heat delivered to a process or series of processes in the direct-use heat

cases. An example of a series of processes is a high-temperature industrial process cas-

caded with a low-temperature industrial process, water heating or space heating. The

reinjection temperature in the direct-use heat cases is set at 50◦ C. The reinjection tem-

perature in the electricity scenarios is determined by the type of power plant, ambient

temperature and production temperature as discussed in Section 4.3.3.

5.1.2 GEOPHIRES Simulation Results

Power Output, and Capital, O&M, and Levelized Costs for EGS Scenarios

The generating capacity (power output), capital costs, O&M cost and levelized costs for
the 18 EGS scenarios are presented in Table 5.3. Figure 5.1 shows the levelized cost of
electricity and direct-use heat for the 18 EGS scenarios. Further, Figure 5.2 shows the cap-
ital cost distribution for the 18 EGS scenarios. The bottom bars represent the geothermal
resource exploration costs. The middle bars represent the geothermal well drilling and
Chapter 5. GEOPHIRES Case-Studies 62

reservoir stimulation costs. Finally, the top bars represent the surface equipment costs
which consist of the power plant costs and the fluid distribution costs.

Table 5.3 – EGS scenarios GEOPHIRES simulation results with E and D-U referring to electricity and
direct-use, respectively.

Low-Grade Medium-Grade High-Grade

E D-U E D-U E D-U
Today’s Technology
Power (MWe or MWth ) 2.5 16.6 11.4 17.4 15.0 17.0
Drilling Costs (M$/well) 19.5 10.1 14.5 4.8 10.1 2.8
Surface Plant Cost (M$) 8.1 3.5 32.3 3.7 35.3 3.7
Stimulation Cost (M$) 2.5 2.5 2.5 2.5 2.5 2.5
Exploration Cost (M$) 14.2 7.9 10.9 4.3 7.9 3.0
Total Capital Cost (M$) 83.2 44.4 118.2 24.9 96.4 17.6
Capital Cost ($/kW) 27,600 2,200 9,410 1,180 5,900 860
O&M Cost (M$/year) 5.1 3.2 7.4 2.0 5.9 1.5
LCOE (¢/kWhe ) or
56.9 14.4 17.8 8.1 10.9 6.1
LCOH ($/MMBTU)
Mid-Term Technology
Power (MWe or MWth ) 3.7 28.2 16.9 29.2 24.4 28.4
Drilling Costs (M$/well) 17.5 9.1 13.1 4.3 9.1 2.5
Surface Plant Cost (M$) 15.0 6.2 49 6.4 53.2 6.3
Stimulation Cost (M$) 2.5 2.5 2.5 2.5 2.5 2.5
Exploration Cost (M$) 12.9 7.3 9.9 4.0 7.3 2.8
Total Capital Cost (M$) 82.9 43.3 126.7 25.8 108.6 19.2
Capital Cost ($/kW) 18,920 1,280 6,910 750 4,150 580
O&M Cost (M$/year) 3.3 2.8 5.2 2.1 4.5 1.8
LCOE (¢/kWhe ) or
31.3 7.8 10.6 5.1 6.4 4.2
LCOH ($/MMBTU)
Commercially Mature Technology
Power (MWe or MWth ) 5.5 41.4 25.1 42.6 34.5 41.4
Drilling Costs (M$/well) 15.6 8.1 11.6 3.8 8.1 2.2
Surface Plant Cost (M$) 21.7 8.8 64.2 9.1 69.5 8.9
Stimulation Cost (M$) 2.5 2.5 2.5 2.5 2.5 2.5
Exploration Cost (M$) 11.6 6.6 8.9 3.7 6.6 2.6
Total Capital Cost (M$) 82.5 42.2 133.7 26.7 119.1 20.8
Capital Cost ($/kW) 12,890 860 4,970 540 3,260 440
O&M Cost (M$/year) 3.4 3.5 5.3 2.9 4.7 2.6
LCOE (¢/kWhe ) or
20.1 5.5 7.1 4.0 4.6 3.5
LCOH ($/MMBTU)
Chapter 5. GEOPHIRES Case-Studies 63

Figure 5.1 – Levelized costs for 18 EGS scenarios. The blue bars represent the levelized cost of electricity
(LCOE) in 2012 U.S. ¢/kWhe for the electricity scenarios (left axis); the red bars represent the levelized
cost of heat (LCOH) in 2012 U.S. $/MMBTU for the direct-use heat scenarios (right axis).

Figure 5.2 – Fraction of capital cost (initial investment) associated with resource exploration (bottom
bars), drilling and reservoir stimulation (middle bars), and surface equipment (top bars) for the 18 EGS
scenarios.
Chapter 5. GEOPHIRES Case-Studies 64

LCOE and LCOH Sensitivity Analysis for Medium-Grade Resource and Mid-Term
Technology Case

A sensitivity analysis is performed on the LCOE and LCOH for the mid-term technol-
ogy case in combination with the medium-grade geothermal resource. The base case
LCOE and LCOH are 10.6 ¢/kWhe and 5.1 $/MMBTU (1.7 ¢/kWhth ), respectively (see
Table 5.3). The parameters considered in the analysis are the capital cost of drilling, reser-
voir stimulation, surface plant, fluid distribution system and resource exploration, the

O&M costs of surface plant and wellfield, and the thermal drawdown rate, discount rate,
reservoir or plant lifetime, and drilling depth. Figure 5.3 shows the effect of all the cost
parameters, the drawdown rate, and the reservoir or plant lifetime on the levelized cost

of end-product. For each datapoint, one parameter value is decreased or increased by 10,

20 or 30%. When changing the lifetime of the system, the thermal drawdown threshold

for drilling new production wells is modified to keep the moment for redrilling at the

mid-life point.

Figure 5.4 shows the effect of drilling depth on the LCOE and LCOH for the medium-

grade resource and mid-term technology case. The drilling depth in the LCOE and LCOH

base case scenario is 5 km and 2.5 km, respectively. When drilling deeper, the increase
in production of direct-use heat and especially of electricity still outweigh the increase

in drilling and surface plant costs, explaining the decrease in LCOH and LCOE. More

discussion on the sensitivity analysis results presented in Figures 5.3 and 5.4 is provided
in the next section.
Chapter 5. GEOPHIRES Case-Studies 65

Figure 5.3 – Sensitivity of LCOE (left figures) and LCOH (right figures) to various parameters for
medium-grade resource and mid-term technology case. High-sensitive and low-sensitive parameters
are shown in top and bottom figures, respectively. The base case LCOE and LCOH are 10.6 ¢/kWhe
and 5.1 $/MMBTU, respectively.
Chapter 5. GEOPHIRES Case-Studies 66

Figure 5.4 – Effect of drilling depth on the LCOE and LCOH for medium-grade resource and mid-term
technology case. The geothermal gradient is constant at 50◦ C/km. For the LCOE, the power plant type
is a subcritical Organic Rankine Cycle for wells shallower than 3.7 km, and a double-flash power plant
for wells deeper than 3.7 km.

5.1.3 Discussion of Results

Deep Geothermal with EGS for Electricity Generation

As shown in Figure 5.1, the Levelized Cost of Electricity (LCOE) from EGS as estimated

with GEOPHIRES varies widely depending on the resource grade and maturity of EGS

technology. For geothermal gradients of 30◦ C/km, which are commonly available in the

Eastern United States (Tester et al., 2006), the LCOE is estimated at almost 60 ¢/kWhe
with today’s technology and could drop to about 20 ¢/kWhe in the future with antici-
pated technology improvements. For a gradient of 50◦ C/km, which is widespread in the
Western United States and can be found in some other regions such as West Virginia and
Texas (Tester et al., 2006), the LCOE with today’s technology is about 18 ¢/kWhe and is
estimated to drop below 10 ¢/kWhe in the future. For the geothermal “hot spots” in the
U.S. with a geothermal gradient of 70◦ C/km or more (Tester et al., 2006), the LCOE with
Chapter 5. GEOPHIRES Case-Studies 67

today’s technology is on the order of 10 ¢/kWhe or less and is predicted to fall below 5
¢/kWhe in the future.

The sensitivity analysis (Figure 5.3) shows that the drilling costs have the highest
impact on the LCOE, followed by the discount rate, plant lifetime, power plant capital
costs, power plant O&M costs and the thermal drawdown rate of the reservoir. Break-
throughs in new drilling techniques might therefore have the highest potential to signif-
icantly lower the LCOE, especially in the Eastern United States. Lower interest rates are
another way to significantly lower the LCOE. They might be obtained by financing the

EGS project through state or federal partnerships or by mitigating risks and uncertainties
by targeting areas with prior knowledge of the subsurface (“brown fields” rather than

“green fields”). The capacity factor is not included in the sensitivity analysis because

it does not have the same level as uncertainty as the considered parameters. Neverthe-

less, when looking at e.g. Equation (4.41), one can find that, to a first approximation, the

LCOE is inversely proportional to the capacity factor. This strong dependence highlights

the importance of designing an EGS for operation at high capacity factors.

Figure 5.2 shows the capital cost distribution of the EGS scenarios. The large contri-

bution of the drilling costs (about 40% for high-grade resources and up to 60% or above

for low-grade resources) in combination with large capital costs in general (3,000 to 6,000
$/kWe for high-grade resources and above 10,000 $/kWe for low-grade resources (see
Table 5.3)) is the main reason for the high sensitivity of LCOE to the drilling costs. Fur-

ther, large upfront capital costs in combination with a long time period until any revenue
is generated and the uncertainty on how big this revenue stream will be, form the main
drawbacks for investors.

Figure 5.5 compares the EGS LCOE estimates from GEOPHIRES with the LCOE for
other electricity generating technologies, obtained from the OpenEI Transparent Cost
Database (U.S. DOE, 2013) developed by the U.S. DOE (Department of Energy) and NREL
Chapter 5. GEOPHIRES Case-Studies 68

(National Renewable Energy Laboratory). The LCOE values taken from this database are
the median values from an extensive list of reported or calculated values for each elec-
tricity generating technology. All LCOE data in Figure 5.5 are calculated for the same

economic parameters (7% discount rate, 30 years lifetime and 39.2% income tax rate).
Cost of transmission lines, subsidies and tax incentives are not included. The LCOE from

EGS at their early stage of development can only compete with concentrating and pho-
tovoltaic solar energy in medium- and high-grade geothermal resources. However, for
commercially mature EGS technology, which is expected to be available by 2030 (assum-
ing proper support for field demonstrations in the next several years), systems utilizing

medium-grade resources are predicted to become economically competitive with other re-

newable and non-renewable energy sources including wind, nuclear, hydro, coal IGCC,
and natural gas turbines. For high-grade resources, EGS is projected to even compete

with combined cycle natural gas and pulverized & scrubbed coal power plants. In com-

parison with the Energy Information Administration (EIA) 2013 Annual Energy Outlook

(EIA, 2013), the prospects for EGS also look favorable. The EIA predicts a generation elec-

tricity price by 2030 of 6 ¢/kWhe in 2011 U.S. $, well within the range of the GEOPHIRES

estimates for medium-grade and high-grade resources, 7.1 and 4.6 ¢/kWhe , respectively.

The LCOE numbers for EGS from the OpenEI Transparent Cost Database are based on

data from the Market Allocation (MARKAL) model (Fishbone and Abilock, 1981), NREL
(Logan et al., 2009), EEE (2010), and Lazard (2013). These studies do not reveal all ma-

jor assumptions for e.g. geothermal gradient or well productivities, and therefore do
not allow for an accurate comparison with the GEOPHIRES results. Nonetheless, other
geothermal economic assessment studies are available with full disclosure of the assump-
tions. For example, Mines and Nathwani (2013) reports LCOE values using GETEM from
76 ¢/kWhe to 13 ¢/kWhe for geothermal gradients ranging from 40 to 70◦ C/km, bottom
hole temperatures ranging from 100 to 325◦ C, and 40 kg/s mass flow rate per production
well. Although these EGS scenarios slightly differ from the scenarios in this study, the
Chapter 5. GEOPHIRES Case-Studies 69

values for the LCOE presented by Mines and Nathwani are comparable to the projections
with GEOPHIRES. Good agreement also exists with the values reported in the “Future of
Geothermal Energy” report (Tester et al., 2006). For a resource with 50◦ C/km gradient,

the “MIT-EGS” model estimates an LCOE of 17.5 ¢/kWhe in the today’s technology case
(20 kg/s per production well) and 5.2 ¢/kWhe in the commercially mature technology

case (80 kg/s per production well). The values projected by GEOPHIRES for a 50◦ C/km
gradient are 17.8 ¢/kWhe in the today’s technology case (30 kg/s per production well)
and 7.1 ¢/kWhe in the commercially mature technology case (70 kg/s per production
well). The GEOPHIRES estimates are higher but the increase in LCOE can be explained

by the higher drilling costs used in comparison with the earlier study which used 2004

drilling costs (Tester et al., 2006; Lukawski et al., 2014).

GEOPHIRES calculations and the comparison with other energy technologies should

be interpreted with care. The LCOE for the non-renewable energy sources depends on the

consumption of the market commodity fuels natural gas, oil and coal, and therefore they

are inherently uncertain with pending fluctuations in prices and availability. The LCOE

for EGS projects will look less promising when including the costs of permits, federal land

leases and other infrastructure requirements, for example, transmission lines and water

supply pipelines. Moreover, in real EGS projects, the exploration costs might be higher

than assumed in GEOPHIRES. The reason is that several exploration and confirmation
wells might be required instead of only one, especially in areas with no prior knowledge

of the subsurface (“green fields”). On the other hand, CO2 credits, subsidies and tax in-
centives will have a positive impact on the EGS costs. Also, learning effects, especially
related to well drilling and stimulation within a certain field, will have a positive effect
on the levelized and capital costs. Furthermore, the reservoir drawdown rate, even for
the commercially mature technology case, might be overestimated. Mines and Nathwani
(2013) consider a drawdown rate of 0.5%/year and Augustine et al. (2010) a target rate
of only 0.3%/year, in comparison with the 1%/year for the commercially mature case in
Chapter 5. GEOPHIRES Case-Studies 70

GEOPHIRES. Finally, the LCOE in the different scenarios has not been optimized with
respect to the drilling depth. For example, Figure 5.4 shows that for the medium-grade
resource and mid-term technology case, the LCOE is not yet minimized within the con-

sidered drilling depth range of 2.5 to 6.5 km. When drilling deeper, the combination of
higher geothermal fluid exergy and higher power plant utilization efficiency can still out-

weigh the increase in drilling costs. Nevertheless, an increased drilling trouble probabil-
ity and higher uncertainties associated with ultra-deep EGS might increase the LCOE for
such projects. Therefore, for this particular EGS scenario, the drilling depth was limited
to 5 km.

Figure 5.5 – Comparison of LCOE expressed in 2012 U.S. ¢/kWhe for EGS obtained using GEOPHIRES
(red bars) with LCOE for different electricity generating technologies (blue bars). LCOE values for
other energy technologies are taken from the OpenEI Transparent Cost Database (U.S. DOE, 2013). The
solid bars and striped bars represent the current LCOE and projected LCOE for 2030, respectively. The
GEOPHIRES commercially mature technology EGS scenarios are used to model the projected system
performance in the year 2030.
Chapter 5. GEOPHIRES Case-Studies 71

Deep Geothermal with EGS for Direct-Use Heat Production

The GEOPHIRES model estimates the LCOH for direct-use applications at 14.4
$/MMBTU (4.9 ¢/kWhth ) for today’s technology and 5.5 $/MMBTU (1.9 ¢/kWhth ) for
commercially mature technology (see Figure 5.1) in areas where a geothermal gradient
of 30◦ C/km is available. For a geothermal gradient of 50◦ C/km, widely found in the
Western U.S., these numbers drop to 8.1 $/MMBTU (2.8 ¢/kWhth ) and 4.0 $/MMBTU
(1.4 ¢/kWhth ), respectively. For a high-grade geothermal resource with a gradient of

70◦ C/km, the LCOH is only 6.1 $/MMBTU (2.1 ¢/kWhth ) for today’s technology and
only 3.5 $/MMBTU (1.2 ¢/kWhth ) for future technology.

Figure 5.6 compares these numbers with the LCOH from natural gas boilers. The lat-

ter LCOH was calculated using the same discount rate of 7%, a lifetime of 30 years and

a capacity factor of 90-95%. The natural gas boiler price was taken at 50 $/kWth, found

to be a good estimate for boilers with thermal output on the order of 100 to 1,000 kWth .

The delivered natural gas prices for industrial and residential sectors in 2011 (5.0 and 11.0

$/MMBTU) and projections for 2030 (6.7 and 13.7 $/MMBTU) were taken from the EIA

2013 Annual Energy Outlook (EIA, 2013) and assuming a Higher Heating Value (HHV)

of 1,020 BTU per cubic feet. The HHV efficiency of the natural gas boiler was taken at

90%. Figure 5.6 also includes the LCOH for EGS district heating systems as calculated by
Reber (2013) using GEOPHIRES. Reber assumed a discount rate of 4% (instead of 7%), a

lifetime of 30 years and a capacity factor of 50% (instead of 90%), and included the capital
and O&M costs for the district heating system. The flow rate per producer was 30 kg/s
for the today’s technology case and 80 kg/s (instead of 70 kg/s) for the commercially
mature technology case. The discount rate is lower because it is assumed the EGS project
will be executed by a utility or public agency, which can borrow money at lower interest
rates. Also, the capacity factor is lower because the EGS district heating system is only
used for space and water heating whose demand varies significantly throughout the year.
Chapter 5. GEOPHIRES Case-Studies 72

Figure 5.6 – Comparison of LCOH (2012 U.S. $/MMBTU) from EGS estimated with GEOPHIRES for
industrial direct-use heat processes (light red bars) and district heating systems (dark red bars) with
LCOH from natural gas boilers (blue bars). The current and projected residential and industrial gas
prices are taken from the 2013 EIA Annual Energy Outlook (EIA, 2013). It is assumed that the today’s
technology EGS scenarios in GEOPHIRES represent the 2012 costs. The commercially mature tech-
nology EGS scenarios are used to model the projected system performance in the year 2030. Low-,
medium-, and high-grade resource refer a geothermal gradient of 30, 50, and 70◦ C/km, respectively.

The results were presented as a supply curve for New York State and Pennsylvania. The
systems with lowest LCOH on this supply curve correspond to geothermal gradients of
around 40-50◦ C/km, which are labeled as “Medium-grade resource” in Figure 5.6. This
figure shows that with current low industrial natural gas prices, no EGS scenario from this
study (light solid red bars) is cost-competitive. However, by 2030, all considered geother-
mal scenarios are predicted to be economically viable. Furthermore, in comparison with
Chapter 5. GEOPHIRES Case-Studies 73

residential natural gas prices, EGS district heating systems with medium-grade resource
are currently not cost competitive but are predicted to be so by 2030. The LCOH for nat-
ural gas boilers are presented in Figure 5.6 for a capacity factor of 90% though, not 50%.

Nevertheless, in comparison with the annual fuel costs, the capital costs of a boiler are so
insignificant over the lifetime of the system that the capacity factor has barely any impact

on the LCOH. Therefore, for a capacity factor of 50%, the LCOH would only increase by
about 1% in these cases. The LCOH for natural gas boilers to a first approximation is
only governed by the natural gas price. This means the large difference in natural gas
LCOH between the residential and industrial case as shown in Figure 5.6 is caused by

the large difference in residential and industrial natural gas price. The higher LCOH for

EGS district heating systems in comparison with the LCOH for EGS direct-use heat cases
is caused mainly by the lower capacity factor and higher capital and O&M costs for EGS

district heating systems.

Both the projections for the natural gas price and the GEOPHIRES estimates for the

LCOH in 2030 should be interpreted with care. No model can accurately predict future

natural gas prices. Over the last 10 years for example, the annual average Henry Hub

spot price for natural gas rose from around 4 $/MMBTU (1.4 ¢/kWth ) in 2002 to peak

around 10 $/MMBTU (3.4 ¢/kWth ) in 2005 and drop below 3 $/MMBTU (1.0 ¢/kWth ) in

2012 as a result of the unconventional gas boom (EIA, 2013).

Similar to the LCOE in the previous section, several factors which could affect the
LCOH are not taken into account in the GEOPHIRES model. These include, among oth-
ers, permits and federal leases, subsidies and tax incentives, learning effects in drilling,
and the cost of transmission pipelines to the end-consumer.

The assumed parameter values for the different technology cases presented in Ta-
ble 5.1 also have their inherent uncertainties. Figure 5.3 shows the LCOH sensitivity to
several of these parameters. The LCOH is again most sensitive to the drilling costs but
Chapter 5. GEOPHIRES Case-Studies 74

the effect is not as strong as with the LCOE. The reason is that the drilling costs are again
the major cost for a direct-use heat system; however, its dominance is not as high as in the
case of electricity production (see Table 5.3 for exact cost numbers). Moreover, a better

learning curve and an overall average drilling cost decline on a system-wide basis can be
expected for direct-use heat systems in comparison with electricity systems, since more

wells at a lower completion cost can be drilled for e.g. a large industrial process or large
district heating system. Other high impact parameters include discount rate, followed by
power plant O&M costs and plant lifetime.

Similar to the electricity case, the capacity factor is not included in the sensitivity anal-
ysis because it is not an uncertain parameter; a high capacity factor can be designed for in

an industrial process. Nevertheless, the capacity factor has a large impact on the LCOH

as can been seen from the inversely proportional relationship between the two (Equa-

tion (4.41)). This explains directly why EGS district heating systems suffer from lower

capacity factors.

Finally, Figure 5.4 shows that the LCOH is much less dependent on the drilling depth

that the LCOE. The reason is that the LCOE, unlike the LCOH, is highly sensitive to the

geofluid temperature because a change in temperature affects the heat-to-power conver-

sion twice. Both the geofluid exergy and the utilization efficiency increase with increasing
geofluid temperature resulting in a stronger than linear dependence of the electricity pro-
duction on the drilling depth. The direct-use heat production for the LCOH changes only

linearly with depth. Also, for a direct-use heat application, one might not be interested to
drill as deep as possible because the production temperature might eventually no longer
suit the intended application.

Even though EGS, and in general geothermal energy, is usually associated with elec-
tricity production, its potential for direct-use heat should be considered given their po-
tential for having projected costs that would compete favorably with natural gas. Even
Chapter 5. GEOPHIRES Case-Studies 75

for low-grade geothermal resources, direct-use heat, unlike electricity, is expected to be-
come cost-competitive with natural gas by 2030, with only a minor increase in natural gas
prices assumed (from 5.0 to 6.7 $/MMBTU in industrial sector and 11.0 to 13.7 $/MMBTU

in residential sector). Furthermore, the required well depths are shallower and the cap-
ital costs are lower, which means lower risk and probably lower discount rates. Also,

the potential is large in the U.S. Recently, Fox et al. (2011) calculated that about 25% of
the total U.S. primary energy demand (25 EJ out of 100 EJ per year) is consumed as heat
below 120◦ C, in comparison to 40% of total U.S. primary energy being converted to elec-
tricity (EIA, 2011). Unfortunately, most U.S. homes, businesses and companies meet their

heating demand for space and water heating and low-temperature industrial processes

by using a combustion furnace fueled by natural gas or oil. To change the current energy
supply system for heat would require significant new infrastructure to deliver thermal

energy from where it is produced in a geothermal field to where it is consumed. Given

that the U.S. has to deal with its aging infrastructure problems in the next few decades, a

real opportunity of such transformational change exists.

Geothermal energy in many cases is ideal to provide heat or both heat and electricity in

cascaded, integrated systems. Examples of low-temperature heat processes with geother-

mal potential (Lund, 2010; Gudmundsson et al., 1985) are space and water heating, space

and water cooling (through absorption chillers), heating greenhouses, fish farming, food
processes such as pasteurization, lumber drying, paper processing including pulping and

drying, biofuels production, and snow melting. The Lindal diagram (Gudmundsson
et al., 1985) ranks several of these applications according to their process temperature.
Iceland’s utilization of its geothermal resources in a fully integrated fashion for electric-
ity and direct-use heat provides a poignant example of what can be done to utilize the
potential value of geothermal resources (Ragnarsson, 2000, 2005, 2010).
Chapter 5. GEOPHIRES Case-Studies 76

5.1.4 Conclusions

In the first case-study, GEOPHIRES was used to simulate 18 EGS scenarios using pa-
rameter values and assumptions described in Tables 5.1 and 5.2. Both electricity and
direct-use heat were considered as end-products for 3 different resource grades (low-,
medium-, and high-grade resource, represented by average geothermal gradients of 30,
50, and 70◦ C/km, respectively) and 3 different levels of technological maturity (today’s,
mid-term, and commercially mature technology represented by production rate improve-

ments from 30 to 70 kg/s per well). Over the entire range of scenarios studied, the lev-
elized cost of electricity ranged from 4.6 to 57 ¢/kWhe and was 11 ¢/kWhe for the base
case scenario (medium-grade resource and mid-term technology). The levelized cost of

heat for the direct-use heat scenarios ranged from 3.5 to 14 $/MMBTU (1.2 to 4.8 ¢/kWhth )

and was 5.1 $/MMBTU (1.7 ¢/kWhth ) for the base case scenario. A sensitivity analysis on

uncertain parameter values in the base case scenarios showed that the drilling costs have

the biggest impact on the LCOE and LCOH. The importance of designing the system for

high capacity factors was also emphasized in the discussion.

GEOPHIRES LCOE estimates are comparable with values found in literature for sim-

ilar EGS conditions, as for example in some of the cases studied by Mines and colleagues
using GETEM. When comparing EGS with other renewable (including hydrothermal)

and non-renewable energy technologies for electricity generation, EGS is not expected

to be cost-competitive currently except with photovoltaic and concentrated solar energy.

However, with moderate technological improvements that would emulate the produc-
tion characteristics of commercial hydrothermal systems in operation today, EGS utilizing
medium- and high-grade geothermal resources is predicted to become cost-competitive
with all renewables, including hydrothermal, and most non-renewable energy tech-
nologies. Because of the low natural gas prices that exist in the U.S. today (around 5
$/MMBTU (1.7 ¢/kWth ) in 2011 as delivered to industrial consumers), no technology
Chapter 5. GEOPHIRES Case-Studies 77

can compete with high efficiency natural gas fired combined cycles for electricity gener-
ation. In addition, even for industrial direct-use heat, EGS using today’s technology is
not economically competitive with natural gas boilers with today’s wholesale industrial

gas prices. Also in comparison with natural gas boilers for space and water heating us-
ing residential gas prices (11 $/MMBTU (3.8 ¢/kWth )), EGS district heating systems are

currently not cost-competitive. For commercially mature technology though, it is pre-

dicted that even for low-grade geothermal resources, EGS for industrial direct-use heat
would be cost-competitive with projected 2030 industrial natural gas prices (7 $/MMBTU
(2.4 ¢/kWth )). Further, EGS district heating systems for medium-grade resources are ex-

pected to become cost-competitive with natural gas boilers at projected 2030 residential

natural gas prices (14 $/MMBTU (4.8 ¢/kWth )).

The results of the GEOPHIRES simulations reported in this case-study should be in-

terpreted with care. Several factors that have not been taken into account may affect the

LCOE or LCOH. Examples of factors that may increase the levelized cost are the cost

for transmission lines, confirmation wells and royalties, and the potential delays due to

unsuccessful wells and permitting issues. On the other hand, subsidies, incentives, and

greenhouse gas credits would have a beneficial impact on the levelized cost. The lev-

elized cost is also not the only criterion to evaluate an EGS project. When assessing the

feasibility of the project or comparing with other energy technologies, EGS will benefit
from its high capacity factors, low environmental impact, and its versatility as energy

source. On the contrary, large upfront capital costs for power plant components and well
drilling and uncertainties and risks associated with the resource itself and the ability to
stimulate reservoir production to commercial levels are currently the main barriers for
investors. Taking all into consideration, the prospect for EGS as a major energy source in
the 21st century looks promising. In the next 10-20 years, with proper Research, Devel-
opment & Demonstration (RD&D) support for field demonstrations to lower risks and
prove the scalability of the technology, the potential of EGS for electricity generation in
Chapter 5. GEOPHIRES Case-Studies 78

medium- and high-grade regions and for direct-use heat, even in lower-grade regions, is
significant.

5.2 Case-Study 2: Deep Geothermal for District Heating in NY and PA

GEOPHIRES has been used by Reber in his graduate work on assessing the potential for
deep geothermal district heating systems in the states of New York and Pennsylvania (Re-
ber, 2013). Based on climate data, Residential and Commercial Buildings Energy Survey

data from the Energy Information Administration, and building and economic data from
the U.S. Census Bureau, Reber estimated first the thermal energy demand for space and

water heating for each town in NY and PA (2894 places in total) (Beckers et al., 2013).

Combined with an updated geothermal gradient map for the region, and estimates of the

local capital and O&M costs for a district heating system based on the total street length,

Reber calculated the LCOH for deep geothermal district heating systems at each place,

using a MATLAB program incorporating the GEOPHIRES tool. He considered three dif-

ferent cases of technological maturity, a similar approach as in Case-Study 1. A map with

the result for the commercially mature case is shown in Figure 5.7. A key take-away mes-

sage from his work is that for a moderate LCOH of 15 $/MMBTU (5.1 ¢/kWhth ), about
20 GWth in development across NY and PA becomes economically attractive in the mid-

term. A full description on the methodology, and an in-depth discussion of all results can

be found in (Reber, 2013) and (Reber et al., 2014).

Chapter 5. GEOPHIRES Case-Studies 79

Figure 5.7 – Estimated LCOH ($/MMBTU) and total capacity (MWth ) for deep geothermal district
heating systems in various places in New York State and Pennsylvania for commercially mature case.
Adapted from (Reber, 2013).

5.3 Case-Study 3: Hybrid Low-Grade Geothermal-Biomass Cogenera-

tion System for Cornell University Campus

5.3.1 Introduction and Background Information

In 2009, students, faculty and staff of Cornell University developed the Climate Action
Plan (CAP) in response to the university’s commitment to become carbon-neutral by 2050

(Cornell University, 2014a). The CAP consists of 62 actions aimed to cut carbon emissions
to zero while setting an example for other campuses and creating a living laboratory for
education and research on climate, energy and sustainability. Key actions include imple-
menting building standards and energy conservation initiatives, producing heat and elec-
Chapter 5. GEOPHIRES Case-Studies 80

tricity from sustainable energy sources, and offsetting unavoidable CO2 emissions with
forest management and community projects. In 2013, accelerated action was called upon
by the Faculty Senate in order to become carbon neutral by 2035. A report on the updated

action plan, labeled the Accelerated Climate Action Plan (ACAP), has been released in
2015 (Cornell University, 2015).

The existing energy system on campus already consists of several energy-efficient and
low-carbon-intense components such as two 15 MWe natural gas turbines with heat re-
covery steam generators co-producing electricity and heat, a campus-wide district heat-

ing and cooling network, a lake-source cooling system, a 1.1 MWe hydroelectric power
plant, and a 1.8 MWe solar photovoltaic (PV) array. Under the CAP, an extension of the

solar PV array is suggested to 10 MWe , as well as electricity generation from the 11.9

MWe Black Oak Wind Farm, improvements in the district heating network by switching

from steam to hot water, and a hybrid geothermal-biomass cogeneration system (Cornell

University, 2014a). The envisioned future energy system for the campus is illustrated in

Figure 5.8.

Integrating energy sources in a hybrid configuration can allow us to take advantage of

synergetic effects by obtaining a higher combined energy output and efficiency or lower

overall levelized cost, or by offsetting weaknesses of one energy source with strengths
of another. In literature, geothermal energy is most often combined with solar energy
when investigating its hybrid potential (Astolfi et al., 2011; Ghasemi et al., 2014). Exam-

ples of regional case-studies are a hybrid geothermal-solar system for Mexico (Lentz and
Almanza, 2006), Australia (Zhou et al., 2013), and Nevada, USA (Greenhut et al., 2010).
In 2012, Enel Green Power developed the first commercial-scale hybrid geothermal-solar
PV plant in Nevada and recently expanded it with a concentrated solar thermal system
(DiMarzio et al., 2015). Hybrid geothermal-biomass plants have also been investigated.
Examples are the case-study for hybridizing an existing geothermal plant in New Zealand
Chapter 5. GEOPHIRES Case-Studies 81

(Thain and DiPippo, 2015), and in Italy (Srinivas et al., 2014). Another case-study of in-
tegrating geothermal energy with biomass for generating electricity, liquefying gas (air),
and providing drying, heating and cooling was conducted by Malik et al. (2015). Recently,

Enel Green Power announced plans to develop the first hybrid geothermal-biomass plant
in Tuscany, Italy (Enel Green Power, 2014).

Given Cornell University’s location in Upstate New York characterized by cold win-
ters and correspondingly large heating demands, a hybrid geothermal-biomass cogener-
ation system is recommended by the CAP to cover most of the heating load while cut-

ting CO2 emissions by up to 40%. With the exception of a few isolated hot springs, no
known hydrothermal resources have been identified in Upstate New York. Therefore,

the geothermal system envisioned for Cornell in this study is an Enhanced Geothermal

System (EGS), although a hot sedimentary aquifer might be possible as well. The overall

heat flow and average geothermal temperature gradients, while locally higher than the

averages for the Northeast region, are still below the higher-grade systems in the Western

U.S. Nevertheless, indigenous geothermal energy and biomass are clear choices because:

(1) even a lower-grade geothermal gradient of around 25◦ C/km is still acceptable when

focusing on direct-use heat applications, (2) Cornell’s agriculture and forest land hold-

ings are extensive. Cornell University owns 4,000 acres (1.6·107 m2 ) of idle pasture or

crop land where bio-energy crops could be cultivated, and about 8,000 acres (3.2·107 m2 )
of forested land that could be harvested sustainably. Additional biomass resources might
include campus food and solid waste, and manure from Cornell’s two dairy farms. Fur-

ther (3), alternative renewable heating sources, such as solar thermal, are lower-grade and
intermittent. A small modular nuclear reactor for cogeneration could be another option
but permitting and gaining public acceptance in Tompkins County would be extremely
difficult and hence is not considered here.

Geothermal energy and biomass can be integrated in several ways, with different con-
Chapter 5. GEOPHIRES Case-Studies 82

Figure 5.8 – Future energy system on the Cornell University campus as envisioned by the Climate
Action Plan (Cornell University, 2014a). When fully converted to a carbon-neutral campus, a large
part of the heating and electricity consumption would be supplied by a hybrid geothermal-biomass
cogeneration system.

figurations shown in the works by Thain and DiPippo (2015) and Srinivas et al. (2014). In

the configuration proposed by the CAP and analyzed in this study, the geothermal system

covers the majority of the heating load during the cold months and generates electricity

during the summer. During very cold winter days when geothermal energy alone is not

sufficient, pretreated biomass is converted in a gasifier to syngas and combusted in the

existing natural gas units to produce additional heat with electricity as co-product. The

objective of this study was to investigate a base case scenario of the proposed hybrid

geothermal-biomass system at Cornell. Using a geothermal simulation tool, cost correla-

tions and resource maps for the area, all recently developed in our research group, new
results on the performance of EGS in Ithaca are reported. This study continued the analy-
sis in our team by providing an update and expansion of previous studies by Tester et al.

(2010) and Lukawski et al. (2013). The results in this study are for a base case scenario
only. Future work will look into variations on this base case to explore the sensitivity of
the technical and economic system performance to several parameters including drilling
Chapter 5. GEOPHIRES Case-Studies 83

costs, discount rate, and reservoir performance.

The existing Cornell energy system is described in more detail in Section 5.3.2 where
data on current energy production, consumption, efficiency, and CO2 emissions are pro-
vided. The proposed geothermal-biomass cogeneration system for Cornell is presented in
Section 5.3.3. Simulation parameters and results are given in Section 5.4.4. The simulation
model is based on GEOPHIRES in combination with capital and O&M costs correlations,
as well as energy conversion efficiencies for the biomass gasifier from literature. The over-
all capital and levelized cost, energy output, and avoided CO2 emissions of the hybrid

system are estimated for two cases (small-scale pilot project and large-scale full conver-
sion system). The focus of this study is on the generation of heat and electricity, not its

distribution. In Section 5.3.5, the results are put in perspective by comparing them with

results from previous studies on utilizing geothermal energy for Cornell (Tester et al.,

2010; Lukawski et al., 2013) and with a traditional natural gas boiler system. Finally, con-

clusions of this case-study are given in Section 5.3.6.

5.3.2 Existing Energy System at Cornell University

Cornell University’s main campus in Ithaca, New York represents a mid-size community
in the Northeastern United States with around 31,000 students, faculty and staff (Cornell

University, 2014a,b). In 2013, the campus gross area was around 14,000,000 ft2 (1,300,000
m2 ), with almost 10,000,000 ft2 (930,000 m2 ) of net building area and 200,000 ft2 (19,000 m2 )
of greenhouse area. The campus energy consumption in 2013 for heating, electricity and
cooling was around 1 trillion BTU (290 GWh), 0.75 trillion BTU (220 GWh), and 0.5 trillion
BTU (145 GWh), respectively (Cornell University, 2014a). Figure 5.9 shows that, despite a
significant growth in campus buildings gross area, campus energy needs have remained
steady over the last several years, thanks to the recently implemented energy conserva-
Chapter 5. GEOPHIRES Case-Studies 84

Figure 5.9 – Campus heating, electricity and cooling consumption (in 1,000 BTU/year) and campus
gross area (in ft2 ) from 2000 to 2013. Despite campus area growth, total energy consumption has re-
mained steady due to energy conservation initiatives (Cornell University, 2014a).

tion initiatives. Cornell’s location in Upstate New York characterized by long and cold

winters explains the large heating demand with 7,182 annual standard heating degree-

days in contrast to only 315 cooling degree-days (Lukawski et al., 2013). The monthly

heating demand ranges between 11 and 53 GWh for campus buildings and 0.45 and 2.3

GWh for the greenhouses. This data was utilized as input data for the simulations.

A state-of-the-art natural gas combined heat and power (CHP) system was commis-

sioned at Cornell in 2009 and covers the majority of the campus heating and electricity

demand. This system consists of two identical natural gas turbines with generators rated
at 15 MWe each, and a heat recovery steam generator producing up to 22,500 pounds per
hour (2.8 kg/s) of low-pressure stream and 136,000 pounds per hour (17.1 kg/s) of high-
pressure steam (see Figure 5.10). After combining the steam outputs from both units, the
low-pressure steam is directly fed to the campus district heating system, while the high-
pressure steam passes first through two back-pressure steam turbines, installed in 1986,
with a total nominal output of 7.5 MWe before eventually feeding into the district heating
network as well. The steam supply temperature and pressure to the district heating sys-
Chapter 5. GEOPHIRES Case-Studies 85

Figure 5.10 – Diagram illustrating one of two identical natural gas combustion turbines with exhaust
gas heat recovery system for generation of low- and high-pressure steam (Cornell Utilities, 2009).

tem range between 320 and 475◦ F (160 and 246◦ C) and between 75 and 90 psia (5.2 and
6.2 bara), respectively. For cold days, additional steam can be generated using natural gas
boilers. The total length of the district heating network is about 28.5 miles (46 km) and
approximately 17% of the distributed heat is lost. In 2013, the CHP system generated 215
GWh of electricity and 1.25 trillion BTU (366 GWh) of steam. The total amount of natural
gas input was 1.98 trillion BTU (580 GWh) resulting in an overall energetic efficiency of
77%. An additional 4,000 MWh of electricity were generated by a 1.1 MWe hydroelec-
Chapter 5. GEOPHIRES Case-Studies 86

tric plant, and a net amount of 15,700 MWh of electricity was purchased from the grid.
The recent installation of a 1.8 MWe solar PV array is expected to produce 2,500 MWh of
electricity annually. Further, 99% of the cooling demand was covered by the lake source

cooling system, commissioned in 2000, and operating with a coefficient of performance

(COP) of 23.

In 2012, the total campus carbon footprint was 218,000 metric tons of CO2 -eq with
on-site combustion, commuting and air travel, and purchased electricity responsible for
142,000 tons, 59,000 tons, and 17,000 tons, respectively (Cornell University, 2014a). No-

tably, the 2012 carbon emissions were 32% lower than the 2008 emissions, amounting to
a reduction of 101,000 metric tons per year. This decrease was mainly due to switching

from coal-fired power to the natural gas-based CHP plant in 2009 and from that point on

eliminating all on-site coal combustion.

Seven key actions were identified in the CAP to further cut the CO2 emissions by 85%

with respect to 2012 levels: (1) develop the hybrid geothermal-biomass system (38%), (2)

offset emissions by carbon management strategies such as forest management and com-

munity projects (27%), (3) purchase around 42 GWh of electricity expected to be produced

by the Black Oak Wind Farm (5%), (4) implement all campus conservation initiatives (5%),

(5) fully integrate building energy standards on campus (5%), (6) optimize the heat distri-
bution network by switching from steam to hot water (3%), and (7) develop the Cornell
University Renewable Bioenergy Initiative (CURBI) to convert campus waste streams to

biogas (1%). The remaining 15% of emissions would be eliminated by an additional 18

action steps such as expansion of the solar PV system and efficiency improvements of
the hydroelectric power plant. With an anticipated 38% reduction in emissions compared
to 2012 levels, corresponding to 82,000 metric tons of CO2 -eq, the hybrid geothermal-
biomass system acts as a core player in the CAP. In the following sections, the technical
and economic performance and required capital investment of such a system are studied.
Chapter 5. GEOPHIRES Case-Studies 87

5.3.3 Proposed Hybrid Geothermal-Biomass Cogeneration System for

Cornell University

Cornell has several options available for sustainable electricity generation, including
wind, solar PV and to a small extent hydro energy; however, the options for renewable
heat production are limited to geothermal and biomass. As discussed in Section 5.3.5,
only a geothermal system can realistically be scaled up on Cornell lands to provide ap-
proximately 1 trillion BTU (293 GWh) of heat per year to meet the current annual heating

demand. Therefore, the aim of the proposed hybrid geothermal-biomass cogeneration

system is to supply the bulk of the campus heating load with EGS as a base-load heat
provider, and biomass as an auxiliary heating source for cold winter days. Technically,

the geothermal system could be sized to cover all daily heat loads even in winter without

the need for a biomass back-up system. However, such design would not be optimal from

an economic point of view. An oversized geothermal system would be expensive because

it would operate at its maximum thermal output for only a few weeks a year, while the

required capital investment scales according to maximum power output. Furthermore, to

avoid having the geothermal system stand idle in summer, electricity will be generated

using an Organic Rankine Cycle (ORC), hereby increasing the overall capacity factor. The

proposed biomass operation consists of sustainable cultivation of bio-crops on Cornell

lands, drying them using geothermal heat, converting the wood chips in a gasifier to syn-

gas (CO + H2 ), and burning this fuel in the existing CHP plant to generate supplemental
heat and electricity. The biomass could be utilized in different ways, e.g. in torrefac-
tion and wood powder combustion. However, since the investment in the state-of-the-art
CHP plant, described in Section 5.3.2, has been made, gasifying the biomass and burning
the syngas in this plant is the more attractive choice even though it would require some
derating of the plant’s generation capacity. A simplified diagram of the integrated hybrid
geothermal-biomass cogeneration system is shown in Figure 5.11.
Chapter 5. GEOPHIRES Case-Studies 88

As stated earlier, the geothermal system envisioned for the Cornell campus is an En-
hanced or Engineered Geothermal System (EGS). The anticipated geothermal gradient in
the area is approximately 25◦ C/km (Stutz et al., 2015), typical for an above-average gra-

dient in the Eastern U.S. but much lower than the gradients seen in the Western U.S. We
assumed 25◦ C/km as an average value for our study albeit accepting that there is inher-

ent uncertainty in that value until the gradient is measured in a deep exploration hole
at the site. The focus is on direct-use heat and cogeneration rather than just electricity
production in order to obtain acceptable levelized costs for these lower-grade resources
(see also Case-Study 1 and 2). The EGS would consist of several injection and produc-

tion wells intersecting a hydraulically stimulated reservoir of “hot dry rocks” used to

circulate water in a closed loop. The well depths are expected to be on the order of 5
km (16,400 ft.) to obtain average geofluid production temperatures around 125◦ C. This

temperature would be sufficient for campus district heating assuming that the Cornell

heat distribution system is switched from steam to pressurized liquid water. A geofluid

temperature of 125◦ C would also still provide an acceptable utilization efficiency of the

ORC power plant. Lower-temperature water from the lake could be used to provide effi-

cient heat rejection from the ORC. After a fraction of heat carried by the geothermal fluid

is extracted by the district heating system or ORC plant, the remaining low-temperature

heat will be used for drying of the biomass. An alternative to the proposed geothermal
option, a campus-scale geothermal heat pump system, is technically possible but would
significantly increase the electricity consumption and is not considered here.

As mentioned earlier, several biomass feedstocks are available at Cornell to be con-

verted to bio-gas, including food waste, manure, harvested forest products and bio-
energy crops including willow and switchgrass. The performance and potential of many
of these feedstocks for Cornell are analyzed in the CURBI project, one of the key actions in
the CAP (see Section 5.3.2). To simplify our preliminary feasibility study, willow, sustain-
ably cultivated on Cornell lands, was selected as the biomass source. The U.S. 1-billion-
Chapter 5. GEOPHIRES Case-Studies 89

Figure 5.11 – Simplified diagram of the proposed hybrid geothermal-biomass cogeneration system.
Red streams refer to geothermal fluid, green streams refer to biomass.

ton study (Perlack and Stokes, 2011) identified willow as an appealing bio-crop for the

Northeastern U.S. because of: (1) its high yields, (2) ease of breeding, (3) decades of exper-

imental research being available on willow in Europe and North America, and (4) willow

having a high positive net-energy ratio of 1:50, and low carbon intensity when consid-

ering the production, harvest, transportation and conversion. Field tests in central New

York (CURBI, 2014) produced willow with yields between 3.7 and 5.2 dry tons/acre/year

(1 ton equals 907 kg). For this study, an average value of 4.45 dry tons/acre/year was

assumed, close to the value reported by the U.S. DOE study (5.1 dry tons/acre/year).

The maximum amount of idle pasture and crop land available for willow cultivation at
Cornell is about 4,000 acres (1.6·107 m2 ) Cornell University (2014a).

5.3.4 Simulation Parameters and Results

Two hybrid geothermal-biomass cases are investigated in this study: Case 1 represents
a pilot or demonstration project with only one doublet (one production well and one in-
Chapter 5. GEOPHIRES Case-Studies 90

Table 5.4 – Base case technical and financial parameter values.

Case 1 - Pilot Case 2 - Full

Parameter
Project (20%) Conversion (98%)
EGS Engineering and Financial Parameters
Geothermal Gradient 25◦ C/km
Well Depth 5 km
Well configuration Doublet Triplet
Number of producers 1 6
Number of injectors 1 3
◦
Ambient Temperature 15 C
Bottom Hole Temperature 140◦ C
Producer Flow Rate 50 kg/s per Producer
Reservoir Thermal Model Percentage Thermal Drawdown
Reservoir Temperature Drawdown Rate 1%/year
Reservoir Impedance per Well-Pair 0.10 MPa·s/L
Reservoir Water Loss Rate 2%
Production Wellbore Temperature Drop Ramey’s Model
Average Well Casing Inner Diameter 0.254 m (10”)
Reservoir Capacity Factor 90%
Geothermal Fluid Pump Efficiency 85%
Fluid Temperature Drawdown Threshold 14%
Geothermal Power Plant Type Subcritical Organic Rankine Cycle (ORC)
EGS Capital costs GEOPHIRES Correlations
EGS O&M costs GEOPHIRES Correlations
Electricity Rate 7 ¢/kWhe
Reinjection temperature 60◦ C
Biomass System Engineering and Financial Parameters
Biomass Cultivating Acreage 500 acres 2,500 acres
Biomass Crop Willow
Annual biomass Yield 4.45 dry tons/acre
Lower Heating Value (LHV) 17.6 MJ/kg
Gasifier efficiency (based on LHV) 85%
Gasifier Capital Cost 3,250 $/kWth
Gasifier O&M Cost 4.5% of Installed Cost + 0.0037 $/kWhth
Biomass Establishment Cost 1,120 $/acre
Biomass Feedstock Price 50 $/dry metric ton
General Financial Parameters
System Lifetime 30 years
Construction Time 2 years
Discount Rate 5%
Inflation Rate 2%
Levelized Cost Model Standard Discounting LCOH model
Chapter 5. GEOPHIRES Case-Studies 91

Table 5.5 – Base case simulation results.

Case 1 Case 2
EGS Power Output Results
EGS Peak Heat Generating Capacity (MWth ) 14.2 85.4
EGS Peak Net Electricity Generating Capacity (MWe ) 0.7 4.1
EGS Average Annual Heat Production (GWhth /year) 72.9 362.8
EGS Average Annual Net Electricity Prod. (GWhe /year) 0.5 5.7
Biomass Power Output Results
Biomass Peak Heat Generating Capacity (MWth ) 0.8 4.5
Biomass Peak Electricity Generating Capacity (MWe ) 0.5 2.6
Biomass Average Annual Heat Prod. (GWhth /year) 3.3 16.3
Biomass Average Annual Electricity Prod. (GWhe /year) 1.9 9.5
EGS Financial Results
EGS Total Capital Investment (M$) 39.5 176.6
of which: Well Drilling and Completion Cost (M$) 30.4 136.6
Piping, Heat Exchangers and ORC Cost (M$) 5.5 31.4
Reservoir Stimulation Cost (M$) 2.5 7.5
Exploration Cost (M$) 1.1 1.1
EGS O&M Costs (M$/year) 0.7 3.3
Biomass Financial Results
Biomass Total Capital Investment (M$) 7.4 40.2
Biomass O&M Cost (M$/year) 0.4 2.3
Overall Financial and CO2 Emissions Results
Total Capital Cost (M$) 46.9 216.8
Total O&M Cost (M$/year) 1.1 5.6
Levelized Cost of Heat (LCOH) ($/MMBTU) 17.3 15.9
Net Avoided CO2 emissions (metric tons/year) 19,000 94,000

jection well) for the geothermal system and a small-scale biomass operation of 500 acres
(2.0·106 m2 ), which together supply about 20% of the campus heating load. Case 2 repre-

sents a full-scale ( 98%) conversion project with six production and three injection wells

and 2,500 acres (1.0·107 m2 ) for willow cultivation. A MATLAB code (MathWorks, 2012)
has been developed to simulate the technical and economic performance of the hybrid
system. This program incorporates (1) a standard discounted levelized cost model, (2)
performance and cost data for biomass cultivation and gasifier conversion as reported in
literature, and (3) reservoir, wellbore, and ORC performance and cost data for the geother-
mal system from GEOPHIRES. The discount rate in the levelized cost model was set at 5%
Chapter 5. GEOPHIRES Case-Studies 92

Figure 5.12 – Monthly geothermal and biomass heat and electricity generation for Case 1 (left) and
Case 2 (right).

- a low value reflecting Cornell’s long-term time horizon of this infrastructure investment.

All input parameters for the two cases are listed in Table 5.4.

All geothermal parameters and models related to the design and performance of the

EGS reservoir and surface plant are described in detail in Chapter 4 and to a certain extent

Section 5.1 (Case-Study 1). Most parameter values such as a 50 kg/s flow rate per pro-

ducer or 1%/year drawdown rate are taken from the “mid-term technology” and “com-

mercially mature technology” cases (Section 5.1), which we feel is a reasonable assump-

tion given that the time frame for the implementation of Cornell’s system is around 2030.
At a thermal drawdown of 14%, occurring after 15 years (mid-point lifetime), all wells
are replaced with newly drilled wells. A high overall capacity factor of 90% is assumed
for the geothermal system because during times of lower heating demand, i.e. summer
months, the geothermal system will produce both electricity and hot water for use in Cor-
nell’s buildings and laboratories. Further, it is assumed the first well drilled serves both
as exploration well and eventual production well. Potentially, more exploration wells are

required, which would increase the total investment costs and LCOH.
Chapter 5. GEOPHIRES Case-Studies 93

Performance and cost data on willow cultivating and gasifying are taken from litera-
ture. Annual willow biomass yields for central New York have been reported by a CURBI
feasibility study at Cornell (CURBI, 2014). A report by the International Renewable En-

ergy Agency (IRENA, 2012) provides lower heating values (LHV) for willow feedstocks,
and average gasifier capital and O&M costs. The gasifier type selected for this study is

a fixed-bed gasifier. Further, it is assumed that the feedstock cost, which includes cost
for harvesting, chipping, transportation, drying, storage, and maintaining the field is 50
$/dry ton, which is in the range of values reported by CURBI (2014) and IRENA (2012).
Also, the “1-billion ton study” reports a willow field establishment cost of 1,120 $/acre

(Perlack and Stokes, 2011). A gasifier conversion efficiency of 85% has been reported

by Ptasinski (2008). Switching from natural gas to syngas poses technical challenges in-
cluding flame stability issues, component overheating, and increased corrosion potential

(Gibbons and Wright, 2009). Various research has been conducted, e.g. by Chacartegui

et al. (2012b; 2012a; 2013), on the performance and potential of running natural gas tur-

bines on syngas. This research shows that off-design operation without retrofitting is

possible but a higher performance and durability can be obtained by optimizing certain

components such as the turbine cooling system and combustor. Identifying the specific

retrofitting requirements for the Cornell system is out of the scope of this study. How-

ever, a drop of 20% in CHP conversion efficiency to heat and electricity is considered as a
penalty for switching from natural gas to syngas. Retrofitting costs are neglected as they
are assumed as being part of a regular major overhaul investment.

The results for both cases for the heat and electricity generating capacity and annual
production, as well as the capital and O&M costs, LCOH (levelized cost of heat), and
avoided CO2 emissions are listed in Table 5.5. Since the main product of the cogener-
ation system is heat, it is opted to express the levelized costs as LCOH with electricity
sales incorporated into the calculation as positive income cash flow at a rate of 7 ¢/kWhe .
Net avoided CO2 emissions are based on decreased natural gas consumption in the CHP
Chapter 5. GEOPHIRES Case-Studies 94

plant for heat generation with lost electricity purchased from the grid with CO2 emis-
sions factor of 501 lbs/MWhe (227 g/kWhe ) (Cornell University, 2014b). No embedded
CO2 emissions in the geothermal plant or biomass operation are included. The average

monthly heat and electricity generation by the geothermal and biomass systems for both
cases is shown in Figure 5.12. For Case 2, the monthly campus heat demand includes the

heat demand for buildings and greenhouses and assumes a 5% thermal energy distribu-
tion loss.

5.3.5 Discussion of Results

The results show that an investment of around $47M in a demonstration-scale hybrid

geothermal-biomass system can supply around 20% of the campus heating demand and

produce an additional 2.4 GWh of electricity per year while saving annually 19,000 met-

ric tons of CO2 emissions. A full conversion to cover 98% of the heating load requires

around $217M of initial investment, and generates annually 15.2 GWh of electricity while

saving 94,000 metric tons of CO2 emissions. The 15.2 GWh of electricity corresponds to

about 7% of the current campus electricity demand. The LCOH is slightly lower in Case

2 due to economies of scale for the EGS (three triplets instead of one doublet), as reflected

by a lower EGS capital cost per heat generating capacity (2,070 $/kWth vs. 2,780 $/kWth ).
One could have designed a hybrid system to cover 100% of the heating load by increasing

either the EGS reservoir capacity (e.g. four instead of three triplets), or the biomass out-
put (more acres and larger gasifier). However, in both cases, the LCOH would be higher
because either the geothermal or biomass system would be significantly oversized. The
remaining 2% could be covered by geothermal heat pumps or natural gas, whose emis-

sions could then be offset by other means. For the pilot project (Case 1), it would be
possible to operate the EGS reservoir at nearly full capacity for 12 months without requir-
ing an ORC. The LCOH would be lower, however, the performance of the cogeneration
Chapter 5. GEOPHIRES Case-Studies 95

aspect in this demonstration project could then not have been assessed.

An LCOH around 16 to 17 $/MMBTU (5.5 to 5.8 ¢/kWhth ) is of the same order as the
LCOH found in Case-Study 1 (Section 5.1) and Case-Study 2 (Section 5.2). In a 2010 study
on EGS for Cornell, Tester et al. reported LCOH values that are 2-3 times lower; this can
be explained by their assumptions of a higher flow rate per producer, a higher gradient,
drilling costs that are not current, a negligible drawdown rate, and no cogeneration. Also
in an earlier hybrid EGS-biomass study for Cornell by Lukawski et al. (2013), different pa-
rameter values were assumed such as a higher flow rate per producer, higher geothermal

gradient, lower drawdown rate and shorter system lifetime. Further, they considered tor-
refied biomass boilers instead of gasification and expressed the results as levelized cost of

electricity (LCOE) instead of heat (LCOH). Nevertheless, they calculated a $34M invest-

ment for a hybrid geothermal biomass cogeneration system to supply 25% of the campus

heat demand, 4% of the campus electricity demand (9 GWh) and save 25,500 metric tons

of CO2 emissions per year.

Due to low natural gas prices, the geothermal-biomass system is currently not

economically-competitive with natural gas boilers. In today’s energy markets with record

low prices for natural gas, the LCOH for heat generated with natural gas boilers is on the

order of 6 to 12 $/MMBTU (2.0 to 4.1 ¢/kWhth ) for industrial and residential rates, re-
spectively (see Section 5.1), considerably lower than the LCOH of around 16 $/MMBTU
(5.5 ¢/kWhth ) calculated in this case-study. Nevertheless, natural gas heating systems are

exposed to price volatility and are subject to CO2 emissions. Gas price increases and some
valuation of “environmental damage”costs for emitted carbon are likely to occur in the
coming decades.

The results presented in Figure 5.12 and Table 5.5 should be interpreted with care. The
required capital investment and resulting LCOH are highly sensitive to drilling costs,
geothermal gradient, capacity factor, flow rate per producer, surface plant O&M costs,
Chapter 5. GEOPHIRES Case-Studies 96

and drawdown rate (Section 5.1). Moreover, the LCOH also depends significantly on
plant lifetime and discount rates. Furthermore, no costs for permitting, transmission
lines, and water supply pipes are included. On the other hand, no subsidies, tax in-

centives, CO2 credits, and learning effects are incorporated, and a drawdown rate of 1%
might be too high (see Section 5.1). In addition, a lower LCOH could most likely be ob-

tained by performing a full system optimization and considering other components for
biomass utilization such as boilers instead of only the CHP plant. Nevertheless, we be-
lieve the base case results cited in this preliminary study provide a measure of the techni-
cal and economic feasibility of a hybrid geothermal-biomass plant for Cornell. One could

envision starting with a demonstration-scale project employing a small geothermal sys-

tem using an EGS well doublet and small biomass operation and gradually adding more
geothermal wells and biomass cultivation area to scale up to full capacity over a period of

20 years. Eventually saving 94,000 metric tons per year of CO2 emissions, covering 98%

of the campus heating load and supplying an additional 15.2 GWh of electricity, could be

a viable key strategy in the CAP.

5.3.6 Conclusions

This case-study analyzed the technical and economic feasibility of a hybrid geothermal-
biomass cogeneration system for the Cornell University campus as a key action of the
CAP to cut carbon emissions to zero by 2050 or sooner. Cornell University represents

an average-sized community, with an existing district heating and cooling system, high
heating load, easy access to biomass resources, and still acceptable geothermal gradient
within the Northeast, and therefore forms the ideal setting to demonstrate the feasibility

of these types of systems before scaling up and replicating at other communities. With
about 25% of the U.S. primary energy demand consumed as low-temperature heat, and
low to medium-grade EGS and biomass resources widely available throughout the coun-
Chapter 5. GEOPHIRES Case-Studies 97

try, a hybrid geothermal-biomass cogeneration system could become a nationwide energy

player. This scenario appears more likely if one also considers the likely future environ-
ment of higher fossil fuel prices, the need to replace aging infrastructure and the urgency

to cut CO2 emissions.

The geothermal system at Cornell would use EGS technology for engineering the
reservoir. This system would produce heat at times of high heating demand and both
heat and electricity at other times. Peak heating demand during winter would be met by
gasifying biomass, and combusting the syngas in the existing CHP plant. In this prelimi-

nary study willow was used as the biomass resource. As a starting point for this study, we
assumed a set of base case conditions to specify required resource, technical performance

and economic parameters for both the geothermal and biomass components. A MAT-

LAB computer model was developed to model the system by incorporating output from

GEOPHIRES for the specified geothermal system, and relying on cost and performance

data published in literature for the biomass system. Two case-studies were investigated

for the Cornell campus. Case 1 represents a pilot project with one EGS doublet, and 500

acres (2.0·106 m2 ) for cultivated willow, together capable of covering 20% of the heating

demand. Case 2 represents a full conversion system, covering about 98% of the campus

heating load, supplied by three EGS triplets and 2,500 acres (1.0·107 m2 ) of willow culti-

vation.

Key results from the simulations are that the pilot project (Case 1) requires $47M of

capital investment, and produces 76.2 GWh of heat and 2.4 GWh of electricity per year.
The CO2 emissions are cut by 19,000 metric tons per year, and the levelized cost of heat
(LCOH) is 17.3 $/MMBTU (5.9 ¢/kWhth ). The full-conversion project (Case 2) requires
$217M, and generates annually 379 GWh of heat (98% of campus supply) and 15.2 GWh
of electricity (7% of campus consumption). The net avoided CO2 emissions are 94,000
metric tons per year (43% of 2012 campus emissions) and the LCOH is 15.9 $/MMBTU
Chapter 5. GEOPHIRES Case-Studies 98

(5.4 ¢/kWhth ). These results are for a base case scenario; to address the uncertainty in
this analysis, future work will look into the sensitivity of the required capital investment,
LCOH, and heat and electricity output to many of the financial and technical parameters

including drilling costs, discount rate and reservoir performance.

References

Aguirre, G. A. (2014). Geothermal Resource Assessment: A Case Study of Spatial Variability and
Uncertainty Analysis for the State of New York and Pennsylvania. Master Thesis, Cornell
University, Ithaca, New York, United States.

Astolfi, M., Xodo, L., Romano, M. C., and Macchi, E. (2011). Technical and economi-

cal analysis of a solar–geothermal hybrid plant based on an Organic Rankine Cycle.

Geothermics, 40 (1): 58–68.

Augustine, C., Young, K. R., and Anderson, A. (2010). Updated U.S. Geothermal Supply

Curve. In Proceedings Thirty-Fifth Workshop on Geothermal Reservoir Engineering, Stanford

University, Stanford, California, February 1 - February 3, 2010, SGP-TR-191.

Beckers, K. F., Lukawski, M. Z., Reber, T. J., Anderson, B. J., Moore, M. C., and Tester,

J. W. (2013). Introducing GEOPHIRES v1.0: Software Package for Estimating Levelized

Cost of Electricity and/or Heat from Enhanced Geothermal Systems. In Proceedings

Thirty-Eighth Workshop on Geothermal Reservoir Engineering, Stanford University, Stan-

ford, California, February 11 - February 13, 2013, SGP-TR-198.

Beckers, K. F., Lukawski, M. Z., Anderson, B. J., Moore, M. C., and Tester, J. W. (2014a).
Levelized costs of electricity and direct-use heat from Enhanced Geothermal Systems.
Journal of Renewable and Sustainable Energy, 6, 013141.

Beckers, K. F., Lukawski, M. Z., Anderson, B. J., Moore, M. C., and Tester, J. W. (2014b).
Chapter 5. GEOPHIRES Case-Studies 99

Erratum: “Levelized costs of electricity and direct-use heat from Enhanced Geothermal
Systems” [J. Renewable Sustainable Energy 6, 013141 (2014)]. Journal of Renewable and
Sustainable Energy, 6, 059902.

Beckers, K. F., Lukawski, M. Z., Aguirre, G. A., Hillson, S. D., and Tester, J. W. (2015). Hy-
brid Low-Grade Geothermal-Biomass Systems for Direct-Use and Co-Generation: from
Campus Demonstration to Nationwide Energy Player. In Proceedings Fortieth Workshop
on Geothermal Reservoir Engineering, Stanford University, Stanford, California, January
26 - January 28, 2015, SGP-TR-204.

Chacartegui, R., Sánchez, D., de Escalona, J. M., Jimenez-Espadafor, F., Munoz, A., and

Sánchez, T. (2012a). SPHERA project: Assessing the use of syngas fuels in gas turbines
and combined cycles from a global perspective. Fuel Processing Technology, 103: 134–145.

Chacartegui, R., Sánchez, D., de Escalona, J. M., Monje, B., and Sánchez, T. (2012b). On the

effects of running existing combined cycle power plants on syngas fuel. Fuel Processing

Technology, 103: 97–109.

Chacartegui, R., Sánchez, D., de Escalona, J. M., Muñoz, A., and Sánchez, T. (2013). Gas

and steam combined cycles for low calorific syngas fuels utilisation. Applied energy, 101:

81–92.

Cornell University (2014a). Climate Action Plan Update 2013 & Roadmap 2014 - 2015. Avail-

able at http://www.sustainablecampus.cornell.edu/initiatives/climate-action-plan.

Cornell University (2014b). Fiscal Year 2013 Cornell University Central Energy Plant
(CEP) Fast Facts. Available at http://energyandsustainability.fs.cornell.edu/file/
FY 2013 DRF CU Energy Fast Facts Master.pdf.

Cornell University (2015). Cornell Leadership for Climate Neutrality - Strategies & 12-Month
Milestones to Accelerate the Climate Action Plan. Climate Action Plan Acceleration Work-
ing Group, Cornell University.
Chapter 5. GEOPHIRES Case-Studies 100

Cornell Utilities (2009). Combined Heat and Power Plant. Available at http:// energyand-
sustainability.fs.cornell.edu/util/heating/production/cep.cfm.

CURBI (2014). CURBI (Cornell University Renewable Bioenergy Initiative) Feasibility Study.
Available at http://cuaes.cals.cornell.edu/sustainability/curbi/.

DiMarzio, G., Angelini, L., Price, W., Chin, C., and Harris, S. (2015). The Stillwater Triple

Hybrid Power Plant: Integrating Geothermal, Solar Photovoltaic and Solar Thermal
Power Generation. In Proceedings World Geothermal Congress 2015, Melbourne, Aus-
tralia, 19-25 April 2015.

EEE (2010). Capital Cost Recommendations for 2009 TEPPC Study, EEE, Energy and

Environmental Economics, Inc. Available at http://www.wecc.biz/committees/

BOD/TEPPC/Versions/ 100106 TEPPC E3 CapitalCosts.ppt 2.0.ppt.

EIA (2011). Annual Energy Review 2011, U.S. Energy Information Administration.

DOE/EIA-0384.

EIA (2013). Annual Energy Outlook 2013, U.S. Energy Information Administration.

EIA (2015). International Energy Outlook 2015, U.S. Energy Information Administration.

Enel Green Power (2014). Press Release: In Tuscany, first plant in the world to inte-

grate geothermal and biomass. Available at http://www.enelgreenpower.com/eWCM/

salastampa/comunicati eng/1662787-1 PDF-1.pdf.

Fishbone, L. G. and Abilock, H. (1981). Markal, a linear-programming model for energy

systems analysis: Technical description of the bnl version. International journal of Energy
research, 5 (4): 353–375.

Fox, D. B., Sutter, D., and Tester, J. W. (2011). The thermal spectrum of low-temperature
energy use in the United States. Energy & Environmental Science, 4 (10): 3731–3740.
Chapter 5. GEOPHIRES Case-Studies 101

Ghasemi, H., Sheu, E., Tizzanini, A., Paci, M., and Mitsos, A. (2014). Hybrid solar–
geothermal power generation: Optimal retrofitting. Applied Energy, 131: 158–170.

Gibbons, T. B. and Wright, I. G. (2009). A review of materials for gas turbines firing syngas
fuels, ORNL/TM-2009/137. Oak Ridge National Laboratory, Oak Ridge, Tennessee,
United States.

Greenhut, A. D., Tester, J. W., DiPippo, R., Field, R., Love, C., Nichols, K., Augustine,
C., Batini, F., Price, B., Gigliucci, G., and Fastelli, I. (2010). Solar-geothermal hybrid
cycle analysis for low enthalpy solar and geothermal resources. In Proceedings World

Geothermal Congress 2010, Bali, Indonesia.

Gudmundsson, J. S., Freeston, D. H., and Lienau, P. J. (1985). The Lindal Diagram. In

GRC Transactions, 9 (1): 15–19.

IRENA (2012). (International Renewable Energy Agency): Renewable Energy Technolo-

gies: Cost Analysis Series: Volume 1: Power Sector, Issue 1/5, Biomass for Power Gen-

eration, (2012). Available at http://www.irena.org/DocumentDownloads/

Publications/RE Technologies Cost Analysis-BIOMASS.pdf.

Lazard (2013). Levelized Cost of Energy Analysis - Version 4.0. Available at

http://www.dpuc.state.ct.us/DEEPEnergy.nsf/c6c6d525f7cdd1168525797d0047c5bf/
8525797c00471adb852579ea00731d74/$FILE/Ex 13 - Lazard 2010 Levelized Cost of

Energy - v 4.0.pdf.

Lentz, Á. and Almanza, R. (2006). Solar–geothermal hybrid system. Applied Thermal
Engineering, 26 (14): 1537–1544.

Logan, J., Sullivan, P., Short, W., Bird, L., James, T. L., and Shah, M. R. (2009). Evaluat-

ing a Proposed 20% National Renewable Portfolio Standard, Technical Report, NREL/TP-
6A2-45161. National Renewable Energy Laboratory (NREL), Golden, Colorado, United
States.
Chapter 5. GEOPHIRES Case-Studies 102

Lukawski, M. Z., Vilaetis, K., Gkogka, L., Beckers, K. F., Anderson, B. J., and Tester,
J. W. (2013). A proposed hybrid geothermal-natural gas-biomass energy system for
cornell university. technical and economic assessment of retrofitting a low-temperature

geothermal district heating system and heat cascading solutions. In Proceedings Thirty-
Eighth Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, Cal-

ifornia, February 11 - February 13, 2013, SGP-TR-198.

Lukawski, M. Z., Anderson, B. J., Augustine, C., Capuano, L. E., Beckers, K. F., Livesay, B.,

and Tester, J. W. (2014). Cost analysis of oil, gas, and geothermal well drilling. Journal
of Petroleum Science and Engineering, 118: 1–14.

Lund, J. W. (2010). Direct utilization of geothermal energy. Energies, 3 (8): 1443–1471.

Malik, M., Dincer, I., and Rosen, M. A. (2015). Development and analysis of a new renew-

able energy-based multi-generation system. Energy, 79: 90–99.

MathWorks (2012). MATLAB Release 2012a, The Mathworks, Inc., Natick, Massachusetts,

United States.

Mines, G. and Nathwani, J. (2013). Estimated power generation costs for EGS. In Pro-

ceedings Thirty-Eighth Workshop on Geothermal Reservoir Engineering, Stanford University,

Stanford, California, February 11 - February 13, 2013, SGP-TR-198.

U.S. DOE (2013). Transparent Cost Database. National Renewable Energy Labora-
tory, Golden, Colorado, United States. U.S. Department of Energy. Available at

http://en.openei.org/wiki/ Transparent Cost Database.

Perlack, R. and Stokes, B. (2011). U.S. Billion-Ton Update: Biomass Supply for a Bioenergy
and Bioproducts Industry, ORNL/TM-2011/224. Oak Ridge National Laboratory, Oak
Ridge, Tennessee, United States.
Chapter 5. GEOPHIRES Case-Studies 103

Ptasinski, K. J. (2008). Thermodynamic efficiency of biomass gasification and biofuels

conversion. Biofuels, Bioproducts and Biorefining, 2 (3): 239–253.

Ragnarsson, A. (2000). Geothermal development in Iceland 1995-1999. In Proceedings

World Geothermal Congress 2000, Kyushu - Tohoku, Japan, May 28 - June 10, 2000, vol-

ume 1, pages 363–375.

Ragnarsson, A. (2005). Geothermal development in Iceland 2000-2004. In Proceedings

World Geothermal Congress 2005, Antalya, Turkey.

Ragnarsson, A. (2010). Geothermal development in Iceland 2005-2009. In Proceedings

World Geothermal Congress 2010, Bali, Indonesia.

Reber, T. J. (2013). Evaluating Opportunities for Enhanced Geothermal System-Based District

Heating in New York and Pennsylvania. Master Thesis, Cornell University, Ithaca, New

York, United States.

Reber, T. J., Beckers, K. F., and Tester, J. W. (2014). The transformative potential of geother-

mal heating in the US energy market: A regional study of New York and Pennsylvania.

Energy Policy, 70: 30–44.

Srinivas, S., Eisenberg, D., Seifkar, N., Leoni, P., Paci, M., and Field, R. P. (2014).

Simulation-Based Study of a Novel Integration: Geothermal–Biomass Power Plant. En-

ergy & Fuels, 28 (12): 7632–7642.

Stutz, G. R., Shope, E., Aguirre, G. A., Batir, J., Frone, Z., Williams, M., Reber, T. J., Wheal-
ton, C. A., Smith, J. D., Richards, M. C., Blackwell, D. D., Tester, J. W., Stedinger, J. R.,
and Jordan, T. E. (2015). Geothermal energy characterization in the Appalachian Basin
of New York and Pennsylvania. Geosphere, 11 (5): 1291–1304.

Tester, J. W., Anderson, B., Batchelor, A., Blackwell, D., DiPippo, R., Drake, E., Garnish,
J., Livesay, B., Moore, M. C., Nichols, K., Petty, S., Toksz, M. N., and Veatch Jr., R. W.
Chapter 5. GEOPHIRES Case-Studies 104

(2006). The future of geothermal energy: Impact of enhanced geothermal systems (EGS) on the
United States in the 21st century. MIT.

Tester, J. W., Joyce, W. S., Brown, L., Bland, B., Clark, A., Jordan, T., Andronicos, C.,
Allmedinger, R., Beyers, S., Blackwell, D., Richards, M., Frone, Z., and Anderson, B.

(2010). Co-Generation Opportunities for Lower Grade Geothermal Resources in the

Northeast - A Case Study of the Cornell Site in Ithaca, NY. In GRC Transactions, 34:
440–448.

Tester, J., Reber, T., Beckers, K., Lukawski, M., Camp, E., Aguirre, G. A., Jordan, T., and

Horowitz, F. (2015a). Integrating geothermal energy use into re-building american in-
frastructure. In Proceedings World Geothermal Congress 2015, Melbourne, Australia, 19-25

April 2015.

Tester, J. W., Reber, T. J., Beckers, K. F., and Lukawski, M. Z. (2015b). Deep geothermal

energy for district heating: Lessons learned from the U.S. and beyond. In Advanced

District Heating and Cooling (DHC) Systems. Woodhead Publishing.

Thain, I. and DiPippo, R. (2015). Hybrid geothermal-biomass power plants: applications,

designs and performance analysis. In Proceedings World Geothermal Congress 2015, Mel-

bourne, Australia, 19-25 April 2015.

Zhou, C., Doroodchi, E., and Moghtaderi, B. (2013). An in-depth assessment of hybrid
solar–geothermal power generation. Energy Conversion and Management, 74: 88–101.
 CHAPTER 6
TECHNO-ECONOMIC ANALYSIS OF DEEP GEOTHERMAL ENERGY SYSTEMS:
KEY CONCLUSIONS

In order to evaluate the technical and economic performance of deep geothermal en-
ergy extraction through Enhanced Geothermal Systems (EGS) for various end-uses, the
simulation tool GEOPHIRES has been developed, which was presented in Chapter 4.
GEOPHIRES incorporates reservoir, wellbore and surface plant models in combination
with capital and operation & maintenance (O&M) cost correlations to estimate the power

output, and investment and levelized costs of the geothermal plant. GEOPHIRES has

been applied to three case-studies in Chapter 5, from which the following key conclu-
sions can be drawn:

• Deep geothermal with EGS for solely electricity generation is only attractive for

medium- and high-grade resources, defined as having geothermal gradients of 50

and 70◦ C/km, respectively. For these resources, the levelized cost of electricity

(LCOE) is in the range 5 to 18 ¢/kWhe with lower values for higher gradients and

more mature technology. In contrast, for low-grade resources (geothermal gradi-

ents of 30◦ C/km), the LCOE varies from 20 to almost 60 ¢/kWhe , depending on the
level of technology maturity. These high values indicate that subsidies or financial

incentives would be required to make low-grade EGS electricity production cost-

competitive with traditional fossil fuel and other renewable energy technologies,
which typically have LCOE’s between 5 and 10 ¢/kWhe .

• Deep geothermal with EGS for direct-use industrial applications (excluding district
heating systems) has values for levelized cost of heat (LCOH) between 6 and 14
$/MMBTU (2.0 and 4.8 ¢/kWth ) for low-grade resources and 4 and 8 $/MMBTU
(1.4 and 2.7 ¢/kWth ) for medium- and high-grade resources. Even for low-grade

105
Chapter 6. Techno-Economic Analysis of Deep Geothermal Energy Systems:
Key Conclusions 106

resources, these values compare favorably with the end-use LCOH for natural gas
boilers, which ranges between 6 and 16 $/MMBTU (2.0 and 5.5 ¢/kWth ), depending
on industrial or residential and current or projected future natural gas prices.

• Deep geothermal with EGS for district heating typically has a higher LCOH because
of the additional capital cost of a district heating system, and a lower capacity factor.
Nevertheless, the values for LCOH for medium-grade resources are between 10 to
19 $/MMBTU (3.4 and 6.5 ¢/kWth ), which are still attractive.

• Hybridization and cogeneration can increase the performance of the plant. In a

case-study for the Cornell University campus (which has an existing district heat-
ing system), a hybrid geothermal-biomass cogeneration plant is proposed where

hybridizing with biomass peak boilers allows for slightly undersizing the geother-

mal system and hereby increasing its capacity factor. In addition, by cogenerating

electricity and heat during summer, the capacity factor of the geothermal system is

increased further. Despite the relatively low assumed average geothermal gradient

of 25◦ C/km, the LCOH of the system is only about 16 $/MMBTU (5.5 ¢/kWth ).

• Given that low-grade geothermal resources are much more common than medium-

or high-grade resources, and the market for low-temperature heat is significant, the

levelized energy costs provided above convincingly indicate that deep geothermal

energy for direct-use and cogeneration could play a major role in the future energy

system for the U.S. and many other countries.

 Part II

Techno-Economic Analysis of Hybrid

Geothermal Heat Pump Systems for

Cooling-Dominated Applications

107
 CHAPTER 7
INTRODUCTION TO HYBRID GEOTHERMAL HEAT PUMP SYSTEMS

7.1 Heat Pump Basics

A heat pump is a device that transfers heat from a heat source reservoir to a heat sink
reservoir using external power input, as illustrated in the left diagram of Figure 7.1
(Çengel and Boles, 2007; Staffell et al., 2012; Tester et al., 2012). The heat source tem-
perature T c is typically colder than the heat sink temperature T h , which means the heat

transfer is opposite to the direction of spontaneous heat flow. A heat pump is in essence

the reverse of a heat engine which generates work using heat flow from a hot reservoir to

a cold reservoir (right diagram in Figure 7.1).

A heat pump can efficiently provide heating, cooling, or both. The efficiency is ex-

pressed as a dimensionless coefficient of performance (COP) and is calculated for heating

and cooling as:

qh
COPheating = , (7.1)
W
qc
COPcooling = (7.2)
W
with W the external power input (in Watt), and qc and qh the heat exchange (in Watt) with
the heat source (cooling) and heat sink (heating), respectively. The theoretical efficiency

(Carnot efficiency), is given by (Tester et al., 2012):

Th
COPheating,Carnot = , (7.3)
Th − Tc
Tc
COPcooling,Carnot = (7.4)
Th − Tc
with T h and T c expressed in Kelvin. Theoretical COP’s of 10 or higher are not uncommon,
meaning 10 units of heating or cooling are supplied for 1 unit of work input. However,
actual heat pumps COP’s are substantially lower, typically in the range 3-6. In contrast,

108
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 109

Figure 7.1 – Simplified diagram of heat pump and heat engine.

the COP of electrical resistance heaters is 1. In the U.S., the cooling efficiency is sometimes

expressed as the Energy Efficiency Ratio (EER) with qc in BTU/h instead of Watt, resulting

in EER = 3.41213 · COP, since 1 W = 3.41213 BTU/h (Staffell et al., 2012). Also, the COP

is a steady-state measure which does not capture changes in performance throughout the

year. One can use instead the Seasonal Performance Factor (SPF), which represents the

average annual performance of the system at a certain location (Staffell et al., 2012), and

can account for the energy consumption and production of other components such as

auxiliary heaters.

The fluid exchanging heat between the heat pump working fluid and the source or
sink is typically air or water (possibly with antifreeze). Depending on the configuration,

one uses the terms air-to-air, air-to-water (or water-to-air), or water-to-water heat pump
system. Heat pumps exchanging heat with the air environment are also called air-source
heat pumps (ASHPs), of which a traditional air-conditioning unit is an example. In the
case of a ground-loop system exchanging heat with the underground soil and rock, one
speaks of a ground-source or geothermal heat pump system (GSHP).

The most common heat pump operation mechanisms are based on the vapor-
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 110

compression and vapor-absorption cycle. In a simple vapor-compression cycle, a refrig-

erant (e.g. R-410A) undergoes in a closed cycle: (1) compression, (2) condensation, (3)
expansion, and (4) evaporation (Çengel and Boles, 2007; Staffell et al., 2012). The work in-
put (W) is the electricity required for the compressor and the circulation pump. Figure 7.2
shows a simplified diagram of the different components in a water-to-air GSHP system
in cooling mode based on the vapor-compression cycle. In heating mode, the evaporator
is coupled to the ground-loop system and the condenser to the air supply. In a vapor-
absorption cycle, an absorber replaces the compressor where the refrigerant is absorbed

into an liquid. The liquid solution is pumped to higher pressures and eventually the
refrigerant is released again as vapor, typically using an external heat source.

Depending on the heat pump source and sink temperatures, as well as the power

plant conversion and electrical grid transmission efficiencies, heat pumps can have ex-

cellent energetic performance in comparison with natural gas, heating oil, and electrical

resistance heating systems. In addition, there are no on-site emissions and the heating

and/or cooling is carbon-free in case the electricity is generated from nuclear or renew-

able resources. Furthermore, heat pumps are highly reliable devices, partially because

of the use of standard, low-tech components, no occurrence of high temperatures due to

combustion, and, specifically for ground-source heat pumps, no direct contact with the
outside environment. For residential and commercial space heating and cooling, ground-
source heat pumps usually perform slightly better (i.e. higher COP) than ASHPs since

the ground temperature is typically higher than the outside temperature during heating
season (winter), and lower during cooling seasons (summer).
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 111

Figure 7.2 – Simplified diagram of water-to-air GSHP system in cooling mode. Red and blue colors
represent arbitrarily warm and cold fluid temperatures, respectively. Drawing not too scale.

7.2 Opportunities, Challenges, and Examples of Hybrid Heat Pump

Systems

In cooling- (or heating-) dominated applications, utilizing regular heat pump systems will
result in thermal imbalance in the ground heat exchange, meaning heat is predominantly
injected (or extracted) from the reservoir. This can lead to long-term ground tempera-
ture changes and degrading system performance over time. One solution lies in oversiz-
ing the geothermal reservoir, which leads to higher capital costs and does not decrease
the thermal imbalance, but rather diminishes its negative impact on the system perfor-
mance. Another solution is hybridizing the system by accompanying the GSHP with
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 112

supplemental heat rejection or heat generation systems, in the case of cooling-dominated

or heating-dominated applications, respectively. These extra components do require ad-
ditional investment costs but might lead to a smaller geothermal reservoir and corre-
spondingly lower geothermal borehole heat exchanger (BHE) field costs. Even in some
cases without thermal imbalance occurring, hybridizing the system can be beneficial as
it can lead to lower overall costs over the lifetime. Challenges with hybrid systems are
more complicated designs, more complicated operation control schemes, and potentially
higher maintenance costs.

Figure 7.3 provides a few simplified examples of hybrid water-to-air GSHP systems

for cooling-dominated applications. Case (a) is the base case non-hybridized water-to-
air heat pump system with HP, CP and SA referring to heat pump, circulation pump

and supply air, respectively. In case (b), the system is hybridized with an air-economizer

(AE) which provides cooling using cold outside air directly and allows the heat pump

system to be turned off. In case (c), supplemental heat rejection is obtained using radiative

cooling (Man et al., 2011). In case (d), a dry-cooler (DC) (also called dry-fluid cooler)

removes heat by forcing ambient air to flow through the DC and cool down the circulating

fluid flowing through a tube bank. In case (e), supplemental heat rejection is provided

using evaporative water cooling in a cooling tower. Some examples of case-studies and
research on GSHPs hybridized with a cooling tower are the works by Xu (2007), Yi et al.
(2008), and Hackel and Pertzborn (2011).

In cases (c), (d), and (e), the valve system allows for various operation modes. With
the heat pump running, the supplemental heat rejection can be operated as “pre-cooler”
in series with the BHE field, or as sole system for heat rejection by bypassing the BHE
field. Alternatively, the heat pump can be bypassed and the supplemental heat rejection
system can actively cool down or recharge the geothermal field such as for seasonal ther-
mal storage (Lhendup et al., 2014). Other valve systems could allow for other operation
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 113

Figure 7.3 – Examples of hybrid water-to-air GSHP systems for cooling-dominated applications. Base
case is shown in case (a) with HP, SA, and CP referring to heat pump, supply air, and circulation pump,
respectively. In case (b), an air-economizer (AE) supplies cool air in parallel to the GSHP system.
Supplemental heat rejection in cases (c), (d), and (e) can be provided in various configurations by
radiative cooling, dry-cooler (DC) and cooling tower (CT), respectively. In case (e), HE refers to heat
exchanger.
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 114

Figure 7.4 – Examples of hybrid water-to-air GSHP systems for heating-dominated applications. Base
case is shown in case (a) with HP, SA, and CP referring to heat pump, supply air, and circulation pump,
respectively. In case (b), a heater (e.g. natural gas heater) supplies warm air in parallel to the GSHP
system. In case (c), a solar thermal heater can provide supplemental heating in various configurations.

modes, e.g. a fraction of the fluid could be cooled by the BHE field and a fraction by the

supplemental heat rejection system.

A few simplified examples of hybrid water-to-air GSHP systems for heating-

dominated applications are illustrated in Figure 7.4. Again, case (a) is the base case with
the same component labeling as in Figure 7.3. In case (b), supplemental heat is provided

in parallel to the heat pump by a heater, e.g. a natural gas or heating oil heater (Hackel
and Pertzborn, 2011). In case (c), depending on the operation mode, solar thermal energy
can be used (1) for thermal recharging of the geothermal field, (2) as sole system for heat
supply to the heat pump by bypassing the geothermal field, or (3) as supplemental heat-
ing in series with the geothermal field. Several studies have been conducted on GSHPs
hybridized with solar thermal energy such as the works by Chua et al. (2010), Wang et al.
(2012), Rad et al. (2013), and Chiasson and Yavuzturk (2014). Other components have
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 115

been integrated in a hybrid set-up such as solar photovoltaic panels or water thermal
energy storage tanks.

Various, sometimes complicated control schemes are available for optimal operation
of a hybrid GSHP system. For example, for a water-to-air GSHP hybridized with a cool-

ing tower for a cooling-dominated application, Wang et al. (2015) and Yavuzturk and
Spitler (2000) have investigated several control schemes: e.g. activation of the heat pump
and/or cooling tower can be dependent on the heat pump entering (EWT) or exiting wa-
ter temperature (ExWT), the difference between EWT or ExWT and ambient dry-bulb or

wet-bulb temperature, or based on a fixed time schedule such as nightly winter operation

for ground recharge during the cold season.

7.3 Design and Modeling of Hybrid Geothermal Heat Pump Systems

Optimal design of hybrid GSHP systems is a rather difficult task because of the large

number of system parameters, and available configurations and control schemes. A rule

of thumb often applied when hybridizing a GSHP is to size the supplemental heat sup-

ply or rejection component such that annual balance in ground heat exchange is obtained

(Yavuzturk and Spitler, 2000). This approach will prevent any system performance degra-
dation caused by a long-term rise (cooling-dominated applications) or decline (heating-
dominated applications) in circulating fluid temperature. However, it might not lead to

the optimal configuration in terms of total cost of ownership or net present value over the
lifetime of the system (Hackel, 2008). Hackel (2008) and Hackel et al. (2009) developed
guidelines for optimizing hybrid systems utilizing TRNSYS (SEL, 2014a), a simulation
tool discussed below. For example, they argue that for a cooling-dominated application,
the optimal size of the BHE is for it to meet the peak heating demand while the sup-
plemental heat rejection component should be sized so that it meets the peak cooling
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 116

demand that is not met by the geothermal system. Other research on optimal design of
hybrid systems is the work by Chiasson (2007), Alavy et al. (2013), and the many studies
by Kavanaugh, the most recent one being (Kavanaugh and Rafferty, 2014).

For an in-depth analysis of hybrid GSHP systems, computer modeling and simulating

are required, for which various software tools are available. One example is utilizing the
open-source, equation-based language Modelica (Mattsson et al., 1998), originally devel-
oped at Lund University in Sweden, for modeling complex physical systems. Huchte-
mann and Müller (2009) present a case-study using Modelica libraries for simulating a

hybrid GSHP with boiler and water storage tank. Another approach is using Simulink

(MathWorks, 2012), e.g. in combination with HAMBASE, a building thermal load model
developed at University of Eindhoven in the Netherlands (De Wit, 2006). Gaspredes et al.

(2014) followed this method to conduct a short and long time-scale analysis of a GSHP

system. Other popular software is EnergyPlus, a whole building simulation tool devel-

oped at the U.S. Department of Energy (Crawley et al., 2001). Sankaranarayanan (2005)

investigated hybrid GSHP systems in EnergyPlus for his master thesis. Another tool is

HVACSIM+, originally developed at the National Bureau of Standards to study HVAC

systems for buildings (Park et al., 1985). Cui et al. (2008) have used HVACSIM+ for a hy-

brid ground-source heat pump case-study with hot water storage tank. Finally, TRNSYS
(TRaNsient SYstem Simulation program) has been developed at the Solar Energy Labo-
ratory at the University of Wisconsin - Madison (SEL, 2014a), and is probably the most

widely used software program for analyzing hybrid GSHP systems. TRNSYS is intro-
duced in more detail below and is applied in this dissertation for analysis of a hybrid
GSHP for a cooling-dominated application (Chapter 9). More background information
on some of these software tools including the underground heat transfer model they have
incorporated is provided by (Spitler et al., 2009) and (Yang et al., 2010).

TRNSYS is a graphically-based software tool for analysis of transient system behav-

Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 117

ior, originally developed to study solar hot water systems but now with applications in
various fields, mostly related to energy, such as fuel cells, GSHPs, wind turbines, etc. It
uses a component-based modeling approach, in a sense similar to the tools HVACSIM+
and Simulink, where the system is represented by an interconnected set of components
(called Types in TRNSYS), such as pumps, heat exchangers, controllers, loads, etc. TRN-
SYS comes delivered with a built-in library of Types, however, also third party Types are
available, for example the TESS (Thermal Energy System Specialists) GHP library (TESS,
2014) for modeling BHEs (see Chapter 9).

Dozens if not hundreds of studies on GSHP and hybrid GSHP utilizing TRNSYS have

been published in literature. A few examples relevant to this work are the studies by
Wang et al. (2012) and Kjellsson et al. (2010) on hybrid geothermal solar thermal systems,

the optimization study by (Cui et al., 2015) on a cooling tower either in series or in par-

allel with a GSHP, the study by Chiasson and Yavuzturk (2009) on the development of

a spreadsheet tool for design of hybrid GSHP for cooling-dominated applications, based

on 90+ TRNSYS simulations, and the work by Hackel (2008) mentioned earlier, on the

design of guidelines for hybrid GSHP for cooling-dominated applications.

References

Alavy, M., Nguyen, H. V., Leong, W. H., and Dworkin, S. B. (2013). A methodology
and computerized approach for optimizing hybrid ground source heat pump system
design. Renewable energy, 57 : 404–412.

Çengel, Y. and Boles, M. (2007). Thermodynamics: An Engineering Approach. McGraw Hill,

6th edition.

Chiasson, A. D. (2007). Simulation and Design of Hybrid Geothermal Heat Pump Systems.
PhD Dissertation, University of Wyoming, Laramie, Wyoming, United States.
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 118

Chiasson, A. D. and Yavuzturk, C. (2009). A design tool for hybrid geothermal heat pump
systems in cooling-dominated buildings. ASHRAE Transactions, 115 (2).

Chiasson, A. and Yavuzturk, C. (2014). Simulation of hybrid solar-geothermal heat pump

systems. In Proceedings Thirty-Ninth Workshop on Geothermal Reservoir Engineering, Stan-

ford University, Stanford, California, February 24 - February 26, 2014, SGP-TR-201.

Chua, K., Chou, S., and Yang, W. (2010). Advances in heat pump systems: A review.

Applied Energy, 87 (12): 3611–3624.

Crawley, D. B., Lawrie, L. K., Winkelmann, F. C., Buhl, W. F., Huang, Y. J., Pedersen, C. O.,
Strand, R. K., Liesen, R. J., Fisher, D. E., Witte, M. J., and Glazer, J. (2001). EnergyPlus:

creating a new-generation building energy simulation program. Energy and buildings,

33 (4): 319–331.

Cui, P., Yang, H., Spitler, J. D., and Fang, Z. (2008). Simulation of hybrid ground-coupled

heat pump with domestic hot water heating systems using HVACSIM+. Energy and

Buildings, 40 (9): 1731–1736.

Cui, W., Zhou, S., and Liu, X. (2015). Optimization of design and operation parameters

for hybrid ground-source heat pump assisted with cooling tower. Energy and Buildings,
99: 253–262.

De Wit, M. (2006). HAMBASE: Heat Air and Moisture model for Building and Systems Evalu-
ation.

Gaspredes, J. L., Masada, G. Y., and Moon, T. J. (2014). A Simulink⃝

R
-Based Building Load-
Ground Source Heat Pump Model Used to Assess Short-and Long-Term Heat Pump
and Ground Loop Performance. Journal of Thermal Science and Engineering Applications,
6 (2): 021013.
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 119

Hackel, S. P. (2008). Development of Design Guidelines for Hybrid Ground-Coupled Heat

Pump Systems. Master Thesis, University of Wisconsin - Madison, Madison, Wiscon-
sin, United States.

Hackel, S., Nellis, G., and Klein, S. (2009). Optimization of cooling-dominated hybrid
ground-coupled heat pump systems. ASHRAE Transactions, 115 (1).

Hackel, S. and Pertzborn, A. (2011). Effective design and operation of hybrid ground-

source heat pumps: three case studies. Energy and Buildings, 43 (12): 3497–3504.

Huchtemann, K. and Müller, D. (2009). Advanced simulation methods for heat pump
systems. In 7th Modelica Conference, Como, Italy, September 20 - September 22, 2009.

Kavanaugh, S. and Rafferty, K. (2014). Geothermal Heating and Cooling: Design of Ground-

Source Heat Pump Systems. ASHRAE.

Kjellsson, E., Hellström, G., and Perers, B. (2010). Optimization of systems with the com-

bination of ground-source heat pump and solar collectors in dwellings. Energy, 35 (6):

2667–2673.

Lhendup, T., Aye, L., and Fuller, R. J. (2014). Thermal charging of boreholes. Renewable
Energy, 67: 165–172.

Man, Y., Yang, H., Spitler, J. D., and Fang, Z. (2011). Feasibility study on novel hybrid
ground coupled heat pump system with nocturnal cooling radiator for cooling load

dominated buildings. Applied Energy, 88 (11): 4160–4171.

MathWorks (2012). MATLAB Release 2012a, The Mathworks, Inc., Natick, Massachusetts,
United States.

Mattsson, S. E., Elmqvist, H., and Otter, M. (1998). Physical system modeling with mod-
elica. Control Engineering Practice, 6 (4): 501–510.
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 120

Park, C., Clark, D. R., and Kelly, G. E. (1985). An overview of hvacsim+, a dynamic
building/hvac/control systems simulation program. In 1st Building Energy Simulation
Conference, International Building Performance Simulation Association, Seattle, Wash-
ington, August 21 - August 22, 1985.

Rad, F. M., Fung, A. S., and Leong, W. H. (2013). Feasibility of combined solar thermal
and ground source heat pump systems in cold climate, canada. Energy and Buildings,
61: 224–232.

Sankaranarayanan, K. P. (2005). Modeling, verification and optimization of hybrid ground

source heat pump systems in EnergyPlus. Master Thesis, Oklahoma State University, Still-

water, Oklahoma, United States.

SEL (2014). TRNSYS 17 (TRaNsient SYstems Simulation program). University of Wisconsin -

Madison, Solar Energy Laboratory (SEL). Available at http://sel.me.wisc.edu/trnsys/.

Spitler, J., Cullin, J., Bernier, M., Kummert, M., Cui, P., Liu, X., Lee, E., and Fisher, D.

(2009). Preliminary intermodel comparison of ground heat exchanger simulation mod-

els. In 11th International Conference on Thermal Energy Storage, Effstock, Sweden, June 14

- June 17, 2009.

Staffell, I., Brett, D., Brandon, N., and Hawkes, A. (2012). A review of domestic heat

pumps. Energy & Environmental Science, 5 (11): 9291–9306.

TESS (2014). TESS Component Library Package. Thermal Energy System Specialists, Inc.,
Madison, Wisconsin. Available at http://www.trnsys.com/tess-libraries/.

Tester, J., Drake, E., Driscoll, M., Golay, M., and Peters, W. (2012). Sustainable Energy:
Choosing Among Options. MIT Press, 2nd edition.

Wang, E., Fung, A. S., Qi, C., and Leong, W. H. (2012). Performance prediction of a hybrid

solar ground-source heat pump system. Energy and Buildings, 47: 600–611.
Chapter 7. Introduction to Hybrid Geothermal Heat Pump Systems 121

Wang, S., Liu, X., and Gates, S. (2015). Comparative study of control strategies for hybrid
gshp system in the cooling dominated climate. Energy and Buildings, 89: 222–230.

Xu, X. (2007). Simulation and optimal control of hybrid ground source heat pump systems. PhD
Dissertation, Oklahoma State University, Stillwater, Oklahoma, United States.

Yang, H., Cui, P., and Fang, Z. (2010). Vertical-borehole ground-coupled heat pumps: a
review of models and systems. Applied Energy, 87 (1): 16–27.

Yavuzturk, C. and Spitler, J. D. (2000). Comparative study of operating and control strate-

gies for hybrid ground-source heat pump systems using a short time step simulation
model. ASHRAE Transactions, 106:192.

Yi, M., Hongxing, Y., and Zhaohong, F. (2008). Study on hybrid ground-coupled heat

pump systems. Energy and Buildings, 40 (11): 2028–2036.

 CHAPTER 8
CORNELL-VERIZON HYBRID GEOTHERMAL HEAT PUMP PROJECT:
BACKGROUND INFORMATION AND VARNA SITE

8.1 Background Information

The equipment at a cellular site other than the antennas is typically housed in a shelter

for protection against environmental conditions and vandalism. The equipment com-
prises batteries, cabling, and telecommunication electronics and generates roughly 8 to
11 kWth of internal heat that remains fairly constant year-round. Given the construction

specifications and space use characteristics, the thermal load inside the shelter is usually

cooling-dominated, even in colder climates, with wall-mounted air-source HVAC units

being the standard cooling option. Tens of thousands of these shelters are installed across

the U.S.

Geothermal heat pumps are an attractive alternative to provide climate control for

cellular tower shelters. On warm days, the ground temperature is typically lower than

the ambient temperature which results in a higher coefficient of performance (COP) for
ground-source (geothermal) heat pumps (GSHPs) versus air-source heat pumps (ASHPs).

The reverse is true on colder days but then an air-economizer (AE) might be sufficient.

The higher performance of geothermal heat pumps leads to lower electricity consump-
tion and correspondingly lower CO2 emissions and lower operational costs. A second
advantage of geothermal heat pumps is superior reliability. Due to its isolation from the
ambient air, the cooling system is not subject to dust particles or extreme weather con-
ditions. This results in lower maintenance costs and a lower chance of system failure.
Failure of the cooling system can lead to down-time of the telecommunications equip-
ment and must unconditionally be avoided. Further, the thermal load is typically known

122
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 123

in advance which allows for efficient design of the cooling system. Nevertheless, because
this application is cooling-dominated, one should be aware of potential thermal imbal-
ance of the geothermal reservoir. In such case, oversizing the boreholes or hybridizing
the system, e.g. using a dry-cooler (DC), is recommended.

In an earlier “proof-of-concept” study conducted in our group in collaboration with

Verizon Wireless, LaBrozzi et al. (2010) argue that GSHPs in combination with AEs are
a more energy-efficient and in some cases also a more cost-effective alternative to con-
ventional HVAC units for cooling of cellular tower shelters. Using the commercial soft-

ware tools GLHEPRO v4.0 by the International Ground Source Heat Pump Association

(IGSHPA), and GLD2009 by Gaia Geothermal, LaBrozzi et al. assessed the overall life cy-
cle cost of ground-source heat pumps versus ASHPs, both with or without AEs, for seven

different climate zones across the country. They concluded that in warmer climate re-

gions, GSHP are not recommended with today’s energy costs, however, in colder regions,

hybrid GSHP with AEs are the most energy-efficient option. When utilizing horizontal

heat exchangers and assuming discounted prices for bulk purchases of GSHP units, they

are also the most cost-effective option.

After the promising study of LaBrozzi et al., a full-scale and fully-equipped demon-

stration system was designed and built during the period 2011-2013, to continue in more
depth the investigation of the performance and potential of (hybrid) geothermal heat

pumps for cellular tower shelter cooling. The demonstration site is located near a newly-
built cellular tower on the grounds of the Cornell University campus in the Cornell Plan-
tations and started full operation during the first half of 2014. In the remainder of the
dissertation, the field site will be referred to as the Varna Site. The set-up consists of six
borehole heat exchangers (BHEs), three ground-source heat pump units, an AE, a DC and
a data monitoring and logging system to closely observe the behavior and performance
of the geothermal reservoir and the hybrid cooling system. Alongside the experimental
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 124

analysis, the current more in-depth study includes systems numerical modeling and sim-
ulating beyond the capabilities of standard commercial software such as GHELPRO. The
specific goals in the current phase of the project are fourfold:

1. Develop a techno-economic model to simulate a hybrid ground-source heat pump

system providing climate control for cellular tower shelters.

2. Monitor the system performance at the Varna Site and utilize collected data to vali-
date the techno-economic model.

3. Using a validated techno-economic model, investigate various configurations of the

hybrid ground-source heat pump system. System variables are heat pump, AE, and

DC size and type, BHE geometry and type, borehole field layout, and management

strategy.

4. Assess the potential of energy, CO2 , and cost savings using hybrid ground-source

heat pumps for cellular tower shelters nationwide.

In this dissertation, the first, second, and partially the third goal are addressed in Chapters

8 and 9. In a follow-on study in our group (Aguirre, 2016), more of goal 3 and goal 4 is
addressed.

In Section 8.2, the hybrid GSHP system at the Varna Site is discussed in more detail,
along with diagrams and photographs of the set-up, and a full list of the components
and monitoring sensors installed. The geology and thermal properties of the geothermal
reservoir are characterized in Section 8.3, based on the results of a thermal response test,
and drill cuttings and rock sample analysis. This chapter is in large part based on a
conference paper presented at the 11th IEA (International Energy Agency) Heat Pump
Conference (Beckers et al., 2014). While the specific application studied in this and the
next chapter is cooling of cellular tower shelters, it is believed the results apply (to a first
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 125

approximation) to other cooling-dominated applications as well, e.g. cooling of small

data centers and cooled industrial storage of food and beverages.

8.2 Varna Site System Set-Up and Logging Data

A full-scale and fully-equipped hybrid ground-source heat pump system was designed
and built during the period 2011-2013 to provide the climate control (i.e. cooling) for a
newly-built Verizon Wireless cellular tower shelter in the Cornell Plantations. The system

start-up phase was completed by the end of 2013 and as of 2014, all telecommunication,

HVAC, and data acquisition equipment were in full operation. The site consists of a wide

array of sensors to monitor the technical and economic performance of the hybrid GSHP

system, and provide data to validate the models for simulating the geothermal reservoir,

BHEs, heat pumps, etc. Figure 8.1 shows the drill rig for drilling of the geothermal and

monitoring wells in May 2013. Figure 8.2 shows the equipment shelter and monopole

cellular tower towards the end of construction, in September 2013.

Due to relatively cold winters in Upstate New York, the hybrid system includes an

AE and DC. The AE brings cold outside air into the shelter when the ambient tempera-
ture drops below a certain setpoint temperature, which allows the geothermal reservoir

to partially thermally recover as no heat rejection from the shelter to the ground is taking

place when the AE is on. Nevertheless, the use of an AE requires regularly replacing the
air filters which increases shelter maintenance costs. In addition, there is increased risk
for dust particles entering the shelter and potentially damaging the electronics. The DC
allows for actively cooling down of the reservoir (“recharging”) to offset the imbalance
in heat exchange and enhance the long-term performance of the GSHP. In addition, the
DC could be utilized as pre-cooler in series with the geothermal reservoir. The projected
annual cooling demand of the shelter depends on the telecommunication technology in-
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 126

Figure 8.1 – Photograph taken at the beginning of construction of the Varna Site in May 2013, showing
the drill rig for drilling the geothermal and monitoring wells.

Figure 8.2 – Photograph of Varna Site taken during construction in September 2013, showing the equip-
ment shelter in the middle and the monopole cellular tower on the right. The geothermal borehole field
is behind the shelter on the left.
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 127

Figure 8.3 – Three-dimensional simplified schematic of the Varna Site.

stalled and is in the range of 80 to 100 MWhth (273 to 341 million BTU). The peak hourly

load is estimated at 12 kWth (41,000 BTU/hour).

Figure 8.3 shows the three-dimensional simplified sketch of the cellular site with the
location of the BHEs. The BHE field consists of four single-U (1-U) BHEs (81 m (265 ft)

depth each) and two double-U (2-U) BHEs (117 m (385 ft) depth each and pipes within

BHE are in parallel with each other), with each set of supply and return pipes individ-
ually controlled by a valve. The BHEs are located relatively close to one another (5.5 to
7.8 m) so that significant thermal interference is expected. The total BHE length is al-
most 560 m (1837 ft), about 30% more than the recommended length by the International
Ground Source Heat Pump Association (IGSHPA, 2009) for these conditions. The reasons
for this surplus of BHE length are (1) a potential future expansion of cellular equipment
which will increase the rate of heat generation, (2) to allow collecting experimental data
on both single-U and double-U BHEs, and (3) to allow investigating different BHE field
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 128

Table 8.1 – Component overview and specifications of shelter and HVAC system at Varna Site.

Shelter
Volume 65.5 m3
External wall + roof area 81.52 m2
Wall thickness 0.112 m
Wall R-value 3.0 K·m2 /W (R-17)
Wall R-value 3.0 K·m2 /W (R-17)
Thermal capacitance 2400 kJ/K
Heat Pump Units
Manufacturer ClimateMaster
Model number Tranquility 27 Two Stage Series 026
Cooling capacity 6.1 kWth (part load) ; 7.7 kWth (full load)
Number of units installed 3 (lead, lag, standby)
Dry-Cooler (DC)
Manufacturer Liebert
Model number DDNC092YS30938
Type Variable Frequency Drive (VFD)
Nominal fan power consumption 0.56 kW (0.75 hp)
Number of units installed 1
Air-Economizer (AE)
Manufacturer Greenheck
Model number BSQ-130-7
Type Variable Frequency Drive (VFD)
Nominal fan power consumption 0.56 kW (0.75 hp)
Number of units installed 1
Circulation Pumps
Manufacturer Bell & Gossett
Model number Series 90 1-/4AA
Type Variable Frequency Drive (VFD)
Nominal motor power consumption 1.12 kW (1.5 hp)
Number of units installed 2

management options, e.g. rotating between BHEs or making strategic use of the antici-
pated groundwater flow in the reservoir. The three heat pump units are identical two-
stage water-to-air heat pumps which operate rotationally in lead, lag, and standby mode.
Table 8.1 provides an overview and model specifications of the shelter and each HVAC
component. Two circulation pumps are installed for redundancy, and operate alternately.
Table 8.2 provides specifications on the geothermal BHEs and circulating fluid. Data on
the underground rock properties are provided in Section 8.3.
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 129

Table 8.2 – Geothermal BHE and circulating fluid specifications at Varna Site.

Borehole Heat Exchanger (BHE) Field

Depth of 1-U BHEs 80.8 m (265 ft)
Number of 1-U BHEs installed 4
Depth of 2-U BHEs 117.3 m (385 ft)
Number of 2-U BHEs installed 2
BHE radius 0.0889 m
Location of and spacing between BHEs See Figure 8.4
1-U BHE pipe inner radius 0.0134 m
1-U BHE pipe outer radius 0.0167 m
2-U BHE pipe inner radius 0.0170 m
2-U BHE pipe outer radius 0.0211 m
Spacing between center pipes 0.11 m
Pipe material HDPE PE4710
Pipe thermal conductivity 0.45 W/(m·K)
Pipe density 960 kg/m3
Pipe specific heat capacity 1930 J/(kg·K)
Grout thermal conductivity 1.73 W/(m·K)
Grout density 1670 kg/m3
Grout specific heat capacity (est.) 2000 J/(kg·K)
Far-field temperature 9◦ C
Circulating Fluid
Mixture (volume-based) 60% water - 40% ethylene glycol
Thermal conductivity 0.41 W/(m·K)
Specific heat capacity 3600 J/(kg·K)
Density 1060 kg/m3
Dynamic viscosity 0.0022 kg/(m·s)
Freezing point -25◦ C

The monitoring system logs a wide array of data in the shelter and BHE field includ-

ing ambient temperature, shelter temperature, fluid supply and return temperatures of

the BHE field, the fluid supply and return temperature of each heat pump, fluid flow rate
for each BHE and each heat pump, power consumption of different components, and the
reservoir temperature at 30 different locations. A full list of relevant data logged at the
Varna Site is provided in Table 8.3. The reservoir temperature sensors are distributed over
one single-U BHE, one double-U BHE, and three monitoring wells (see Figure 8.4). In each
well, the sensors are equally spaced at five different depths and are installed in duplicate
for a total of 60 field temperature sensors. Since groundwater flow was observed during
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 130

Figure 8.4 – Lay-out of BHE field, equipment shelter and cellular tower of Verizon hybrid geothermal
heat pump project at Varna Site.

drilling and the in-situ thermal response test (see Section 8.3.2), an open-hole water mon-

itoring well was also installed. This well is not permanently equipped with sensors but

allows lowering a tool at regular times to assess groundwater presence. Moreover, it is

anticipated that the large number of reservoir temperature sensors will reveal over time

information on groundwater flow as well. The data acquisition frequency is set at once

per hour for slow-changing data such as far-field temperature, while it is once per minute
or once per five minutes for fast-changing data such as supply and return temperatures.
In addition, a fully-equipped and Cornell-operated weather station is located about 300
m away from the cellular site. This weather station collects hourly data on ambient tem-
perature, humidity, solar radiation, rain and snow fall, wind speed, wind direction and
soil temperature at 4 and 8 inch (10 and 20 cm) depths.
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 131

Table 8.3 – List of relevant data logged at Varna Site. Status refers to on or off.

Shelter
Shelter temperature; Shelter relative humidity;
Outside temperature; Outside relative humidity;
Alternating Current (AC) supply voltages; AC supply kWh counter;
Alternating Current to Direct Current (ACDC) Plant power
Heat Pump Units
Compressor stage 1 status; Compressor stage 2 status; Electric current;
Circulating fluid valve status; Circulating fluid supply temperature;
Circulating fluid return temperature; Air supply temperature; Fan status
Dry-Cooler (DC)
DC status; DC supply voltage; DC supply current;
Circulating fluid valve status; Circulating fluid supply temperature;
Circulating fluid return temperature; Fan motor VFD frequency; Fan motor speed
Air-Economizer (AE)
AE status; AE supply voltage; AE supply current;
Circulating fluid valve status; Circulating fluid supply temperature;
Circulating fluid return temperature; Air supply temperature; Air return temperature;
Fan motor VFD frequency; Fan motor speed
Circulation Pumps
Pump 1 status; Pump 1 supply voltage; Pump 1 supply current; Pump 1 power;
Pump 1 motor VFD frequency; Pump 1 motor VFD speed;
Pump 1 run time; Pump 1 kWh counter;
Pump 2 status; Pump 2 supply voltage; Pump 2 supply current; Pump 2 power;
Pump 2 motor VFD frequency; Pump 2 motor VFD speed;
Pump 2 run time; Pump 2 kWh counter
Borehole Heat Exchanger (BHE) Field
Circulating fluid supply temperature; Circulating fluid return temperature;
265 ft monitoring wells temperature at 15, 75, 135, 195, and 255 ft depth;
265 ft 1-U BHE well temperature at 15, 75, 135, 195, and 255 ft depth;
385 ft 2-U BHE well temperature at 15, 105, 195, 285, and 375 ft depth;
BHE geofluid valve status

Figures 8.5 to 8.7 present some temperature data collected at the Varna Site since mea-
surements started during the Summer of 2014. Figure 8.5 shows the shelter and ambient
temperature. The shelter setpoint temperature has been modified at several occasions. In
general, the controller and cooling system, alternating between one heat pump in part
load and two heat pumps in full load, have been able to keep the shelter temperature
within the requested hysteresis around the specified setpoint. Some anomalies in shelter
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 132

Figure 8.5 – Shelter and ambient temperature at Varna Site from August 5th , 2014 till January 30th , 2016.
Due to technical issues, some data is missing around November 1st , 2014 and during the Fall of 2015.

temperature are a result of testing of equipment or maintenance at the site. Due to tech-

nical issues, some data had not been logged around November 1st , 2014 and during the

Fall of 2015.

Figure 8.6 shows the circulating fluid (geofluid) supply and return temperature. Be-

fore March 3rd , 2015, no hybrid component had been activated and all heat had been

injected in the geothermal reservoir. After March 3rd , 2015 until the end of the measure-
ments shown, the DC had been running continuously. The DC runs in parallel with the

heat pumps meaning that part of the circulating fluid leaving the BHE field enters the
heat pump while the other part enters the DC. This operation mode results in supply
and return temperatures in some cases overlapping each other, and closely following the
ambient temperature. The measured circulating fluid temperatures are only approximate
because the temperature sensors are attached to the outside the pipes (of copper material)
and will be influenced by the shelter temperature. Again, due to technical issues, some
data is missing around November 1st , 2014 and during the Fall of 2015.
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 133

Figure 8.6 – Circulating fluid (geofluid) return (leaving the shelter) and supply (entering the shelter)
temperature at Varna Site from August 29th , 2014 till January 30th , 2016. Around March 1st , 2015, the
DC was activated. Due to technical issues, some data is missing around November 1st , 2014 and during
the Fall of 2015.

The average circulating fluid supply temperature, which is equal to the heat pump

entering fluid temperature, is about 20◦ C. The total heat pump run time in part load and

full load so far is about 5,700 and 13,600 hours, respectively. Due to technical issues no

flow rate data has been collected with the installed flow meters at the site. Nevertheless,

based on the temperature data and flow rate measurements with an acoustic clamp-on

flow meter, the flow rates are expected to be in the range 4 to 7 GPM (0.25 to 0.44 l/s) per

heat pump running. Assuming an average flow rate of 5.0 GPM (0.32 l/s), and excluding
the circulating pump power, the average heat pump COP from end of August 2014 to
end of January 2016 is about 6.1. When including the power for the circulating pump, the
COP drops to 5.1 If also considering the power for the DC, the COP drops further to 4.8.
An ASHP during this time period would have had an average COP of about 3.3.

Figure 8.7 plots the temperature measured at four different depths in wellbore D,
whose location in the BHE field is shown in Figure 8.4. So far, only the two 2-U BHEs have
been in operation, which are located on the other side in the BHE field. Figure 8.7 shows
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 134

Figure 8.7 – Reservoir temperature at four different depths in borehole D (see Figure 8.4) at Varna Site
from August 5th , 2014 till January 30th , 2016. Due to technical issues, some data is missing during the
Fall of 2015.

a trend of slowly increasing reservoir temperature for all four depths shown, presumably

caused by the continuous injection of heat in the underground. Due to temperature sen-

sor malfunctioning, no data is collected in wellbore D at the depth of 135 ft (41.2 m). The

shallow reservoir temperature measured at 15 ft (4.6 m) depth follows a sinusoidal pro-

file caused by the seasonal sinusoidal ambient temperature profile (see Figure 8.5), but

shifted in time and decreased in amplitude. This behavior is matched by a simple tran-

sient heat conduction model simulating a semi-infinite domain with sinusoidal ambient
temperature at the boundary (Bandos et al., 2009). For a constant and uniform thermal
diffusivity of 0.62·10−6 m2 /s, the model predicts at 4.6 m depth a sinusoidal profile with
a time delay of 3.5 months and amplitude decrease of about 84% (from 14◦ C to 2.25◦ C),
corresponding well with the observed temperature profile. As discussed in Section 8.3,
a value of 0.62·10−6 m2 /s lies in the range of expected values for thermal diffusivity for
unconsolidated rocks making up the top soil part.
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 135

8.3 Varna Site Geothermal Reservoir Characterization

In order to conduct representative thermal simulations of the geothermal reservoir, as

well as to correctly interpret the measured in-situ reservoir and circulating fluid return

(i.e. leaving the BHE field) temperatures, proper understanding of the shallow geol-
ogy and corresponding thermophysical properties is necessary. This section presents the
methodology and results of the reservoir characterization at the Varna Site using drill cut-
tings analysis, a thermal response test, and thermal conductivity measurements of rock

samples.

8.3.1 Drill Cuttings Analysis

During drilling of one of the 265 ft (81 m) depth temperature monitoring wells, drill cut-

tings were collected at 10 ft (3.0 m) intervals from 0 to 60 ft (18.3 m) depth, and 20 ft (6.1 m)

intervals from 60 ft (18.3 m) depth to the bottom (see right diagram in Figure 8.8). Based

on analysis of these cuttings in combination with general knowledge of the local geology

(Tester et al., 2010; USDA, 2015; NYSDEC, 2015), the shallow underground geology at the
Varna Site is (to a first approximation) interpreted as Quaternary unconsolidated alluvial

and glacial deposits topping Devonian shale (left diagram in Figure 8.8). The drillers

reported shale starting at 123 ft (37.5 m) depth.

The dominant rock type in the reservoir is shale. Reported values in literature for shale
thermal conductivity (k), specific heat capacity (c p ), density (ρ), and thermal diffusivity
( )
α = ρck p are listed in Table 8.4. Based on this data, expected values for k and α, which
are the parameters used in the simulations, are in the range 1-2 W/(m·K), and around
10-6 m2 /s, respectively. Expected values for thermophysical properties for the unconsol-
idated sediments (soil) are in the range 1.2-2.4 W/(m·K) for thermal conductivity and
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 136

Figure 8.8 – Diagram of geothermal reservoir geology and drill cuttings collection depths at Varna Site.

0.5-1·10-6 m2 /s for thermal diffusivity (Hellström, 1991). The value 0.62·10−6 m2 /s derived

in Section 8.2 using temperature data collected at the Varna Site lies within the expected

range for thermal diffusivity. Lumped-parameter thermal properties for the entire reser-
voir could be obtained by averaging the parameter values for each layer weighted by the
layer thickness.

Groundwater was encountered during drilling. Darcy-like flow (matrix flow) is ex-
pected in the top part (unconsolidated sediments) while fracture-dominated flow is ex-
pected in the bottom part (shale). Groundwater flow enhances the heat dissipation and is
therefore advantageous for the system performance. Nevertheless, limited data are avail-

able on the quantity, direction and location of the fluid flow and hence, the corresponding
advective heat transfer contribution is difficult to model correctly in the simulations. In
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 137

Table 8.4 – Published thermophysical properties for shale.

Min Max Mean Reference

Thermal conductivity k (W/(m·K))
0.8 2.1 (Rybach and Muffler, 1981)
1.63 2.51 (Reiter and Tovar, 1982)
0.55 4.25 2.07 (Čermák and Rybach, 1982)
0.64 1.24 0.92 (Gilliam and Morgan, 1987)
0.96 1.8 (Beach et al., 1987)
1.05 1.45 (Blackwell and Steele, 1989)
0.8 4.0 (EPRI, 1989)
1.5 3.5 (Hellström, 1991)
0.6 4.0 Barker
Specific heat capacity c p (J/(kg·K))
880 1,440 1,180 (Čermák and Rybach, 1982)
880 1,090 980 (Gilliam and Morgan, 1987)
950 2,200 (Hellström, 1991)
Density ρ (kg/m3 )
2,350 2,850 (Čermák and Rybach, 1982)
2,300 2,750 (Schön, 2011)
Thermal diffusivity α (10-6 m2 /s)
0.53 1.6 0.94 (Čermák and Rybach, 1982)

response, an open-hole water monitoring well (currently unequipped) has been installed

in the BHE field (see Figure 8.4) which could be used in the future to collect more data on

the reservoir.

8.3.2 Thermal Response Test

A thermal response test (TRT), also called thermal conductivity test, is an on-site, multi-
day experiment to estimate average effective thermal properties of the geothermal BHE
field (Spitler and Gehlin, 2015). For a time period of usually 48 to 96 hours, one circulates
a fluid (typically water) with constant flow rate in a closed loop in a completed BHE while
continuously supplying constant heat of several kW to the fluid. By measuring the sup-
ply (injection) and return (production) temperature, one can derive a lumped-parameter
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 138

thermal conductivity of the geothermal reservoir, as well as the thermal resistance of the
BHE as installed. Generally, a TRT is conducted after the first BHE is installed to assess
whether or not the reservoir thermal conductivity matches the design value. In case of
underestimation (overestimation) of the thermal conductivity, one can adjust the design
to prevent oversizing (undersizing) of the geothermal BHE field. In May 2013, a TRT was
conducted at the Varna Site, shown in Figure 8.9.

Zhang et al. (2014) review different analytical and numerical methods to derive the av-
erage thermal properties using TRT. The impact of groundwater flow is studied in depth

by Gehlin (2002). An historical review on thermal response testing is provided by Spitler

and Gehlin (2015). Other interesting studies that provide an overview of TRT and model
equations to estimate the reservoir thermal conductivity and borehole thermal resistance

are the works by Yang et al. (2013) and Al-Khoury (2012).

In a simple analytical TRT model (Yang et al., 2013), one assumes the average cir-

culating fluid temperature T f,ave in the BHE as the average of the fluid inlet and outlet

temperature:
T f,inlet + T f,outlet
T f,ave = . (8.1)
2
Assuming quasi steady-state heat transfer within the BHE, the heat flux Q (in W/m) from

the fluid to the surrounding rocks is calculated as:

(T f,ave − T b )
Q= (8.2)
Rb

with T b and Rb the borehole wall temperature and borehole thermal resistance, respec-
tively. T b as a function of time t can be calculated as the ground temperature evaluated at
the borehole radius rb using a simple infinite line source model (ILS):
( 2 ) [ ( ) ]
Q rb Q 4αt
T b (t) = T 0 + E1  T0 + ln 2 − γ (8.3)
4πk 4αt 4πk rb
with T 0 , k, α and γ the far-field (or initial) reservoir temperature, reservoir thermal con-
ductivity, reservoir thermal diffusivity, and Euler’s constant (0.5772...), respectively. E1 (x)
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 139

Figure 8.9 – 96-hour thermal response test (TRT) at Varna Site conducted in May, 2013.

(∫ ∞ exp(−u)
) ( )
is the exponential integral function x u
du , which simplifies to ln 1
x
− γ for large

x, with ln the natural logarithm (Abramowitz and Stegun, 1964). This model assumes the

BHE is infinitely long. This is an acceptable assumption since the thermal diffusion length

during a TRT (∼0.6 m for a four-day TRT with ground thermal diffusivity of 10−6 m2 /s) is
much smaller than the actual borehole length (∼100 m). Combining Equations (8.2) and
5rb2
(8.3), one can find the following expression for T f,ave , valid for t > α
:
[ ( ) ]
Q 4αt
T f,ave = ln 2 − γ + T 0 + Q · Rb (8.4)
4πk rb
[ ( ) ]
Q Q 4α
= ln(t) + ln 2 − γ + T 0 + Q · Rb = κ · ln(t) + λ (8.5)
4πk
|{z} 4πk rb
κ
| {z }
λ

which means the average fluid temperature can be approximated as a line in a semi-log
plot. The geothermal reservoir thermal conductivity is then estimated using the slope κ
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 140

as:
Q q
k= = (8.6)
4πκ 4πκL
with q and L the total heat supplied by the heater to the fluid (in W), and the BHE length,
respectively. Using the intercept λ, one can find a value for the borehole resistance, as-
suming one has an estimate or assumed value for the reservoir thermal diffusivity and
far-field reservoir temperature:
[ ( ) ]
λ 1 4α T0
Rb = − ln 2 − γ − . (8.7)
Q 4πk rb Q

At the Varna Site, the TRT was conducted on one loop of a double-U BHE with 117.3

m (385 ft) depth. About 7.5 kW of heat was supplied to the circulating fluid which was

water. The total test duration was 96 hours, however, during the last six hours the heat

input slightly changed and hence, only 90 hours of measured test data are considered.

The measured average fluid temperature T f,ave from 1 to 90 hours is shown by the blue

curve in Figure 8.10 on a semi-log plot. A linear fit was applied to the data after discarding

the first 12 hours of measurements (a similar approach as by Spitler and Gehlin (2015)),

and is shown in Figure 8.10 as the red dashed line. Based on the slope and intercept of this

linear fit, Equations (8.6) and (8.7) result in a reservoir thermal conductivity and borehole

thermal resistance of 4.0 W/(m·K) and 0.16 m·K/W, respectively. All TRT model input

parameters and results are listed in Table 8.5.

An effective thermal conductivity of 4.0 W/(m·K) is rather high in comparison with

reported values for shale, which are in the range 0.6 to 4.25 W/(m·K) (see Section 8.3.1).
The reason for a high value at the Varna Site is the likely presence of groundwater flow,
as reported by the drillers. When replacing the ILS with a moving infinite line source
(MILS) model (Diao et al., 2004), one can, to a first approximation, account for uniform
horizontal Darcy-like groundwater flow. With this model, assuming a ground thermal
diffusivity of 10−6 m2 /s, a preliminary analysis shows that the measured average fluid
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 141

Figure 8.10 – Measured average fluid temperature and linear fit for thermal response test at Varna Site.

Table 8.5 – Model input parameter values and results for Varna Site thermal response test.

TRT Model Input Parameters

Thermal diffusivity (est.) α 10−6 m2 /s
Borehole radius rb 0.073 m
Borehole depth L 117.3 m
Heat input q 7,416 W
Far-field ground temperature T 0 10.6◦ C
Test total time duration 96 h
Linear fit time period 12-90 h
TRT Model Results
Linear fit slope κ 1.25
Linear fit intercept λ 11.2
Effective thermal conductivity k 4 W/(m·K)
Borehole thermal resistance Rb 0.163 m·K/W
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 142

temperature data of the TRT is best matched with a ground thermal conductivity of 2.0
W/(m·K), a uniform average groundwater velocity of 10−6 m/s, and a borehole thermal
resistance of 0.10 m·K/W.

8.3.3 Shale Sample Thermal Conductivity Measurement

The thermal conductivity is the most important rock property in conduction-dominated

reservoirs. It determines how effectively heat dissipates and what the required BHE
length is. Although groundwater flow and correspondingly advective heat transfer is

known to be present at the Varna Site, it is still desirable to accurately know the rock ther-

mal conductivity in order to simulate the worst-case scenario (conductive heat transfer

only) and assess the impact of the groundwater flow on the system performance.

A rock sample was taken from a mudstone outcrop in a gorge near the Varna Site.

The thermal conductivity of the rock sample was measured at the Southern Methodist

University (SMU) Geothermal Lab, using the apparatus shown in Figure 8.11. This ap-

paratus is based on the divided bar technique (Goss and Combs, 1976; Blackwell and

Spafford, 1982), illustrated schematically for this set-up in Figure 8.12. This method is a

steady-state test to estimate the thermal conductivity of a specimen (here shale sample)

by forcing a heat flux through the specimen and through a material of known thermal

conductivity, and measuring the temperature drops. Specifically for this apparatus, the
temperature drops are measured over an upper heat flux meter, over the specimen, and
over a lower heat flux meter, and compared with the temperature drops when running the
experiment for a specimen with known thermal conductivity (here silica glass or quartz
material). The heat flux meters consist of a layer of material with certain conductivity
in between two copper discs. The specimen is loaded under axial pressure to represent
in-situ conditions and limit the contact thermal resistances. The heat source and sink are
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 143

Figure 8.11 – Apparatus at Southern Methodist University (SMU) Geothermal Lab used for measuring
thermal conductivity of shale sample from outcrop near Varna Site.

Figure 8.12 – Schematic diagram of apparatus to measure thermal conductivity of specimen, based on
the divided bar technique. The temperatures T 1 to T 4 are measured inside the red copper discs of the
heat flux meters.
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 144

supplied by constant temperature circulating baths. The thermal conductivity of the shale
sample k shale is then calculated as (Blackwell and Spafford, 1982):
( )( ) 2
∆T 1,shale + ∆T 3,shale ∆T 1,re f + ∆T 3,re f T h shale Dre f
k shale = (8.8)
2∆T 2,shale 2∆T 2,re f T hre f D2shale

with ∆T 1 , ∆T 2 , and ∆T 3 calculated as T 1 -T 2 , T 2 -T 3 , and T 3 -T 4 , respectively, with T 1 to T 4

the temperatures measured inside the copper discs of the heat flux meters (red discs in
Figure 8.12). The subscripts shale and ref refer to the shale sample and to the reference
material. Further, T h shale , T hre f , Dre f , and D shale are the thickness and diameter of the shale

and reference specimen, respectively.

The thermal conductivity measured for the shale sample is about 1.0 W/(m·K), which

falls in the range of reported values in literature (see Table 8.4), and could be interpreted

as an approximate value for a lower bound for the effective thermal conductivity for a

reservoir that is subject only to heat conduction. At the Varna Site, we postulated that ad-

vection by groundwater flow is an added factor. As discussed earlier, the effective thermal

conductivity measured with the TRT is about four times higher, which substantiates the

significant impact of groundwater flow on reservoir heat dissipation.

References

Abramowitz, M. and Stegun, I. A. (1964). Handbook of mathematical functions: with formulas,

graphs, and mathematical tables. Number 55. Courier Corporation, Mineola, New York.

Aguirre, G. A. (2016). Nationwide Potential of Hybrid Ground-Source Heat Pump Systems

for Climate Control of Cellular Tower Shelters. Internal Report, Cornell Energy Institute,
Cornell University, Ithaca, New York, United States.

Al-Khoury, R. (2012). Computational Modeling of Shallow Geothermal Systems. CRC Press.

Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 145

Bandos, T. V., Montero, Á., Fernández, E., Santander, J. L. G., Isidro, J. M., Pérez, J., de
Córdoba, P. J. F., and Urchueguı́a, J. F. (2009). Finite line-source model for borehole heat
exchangers: effect of vertical temperature variations. Geothermics, 38 (2): 263–270.

Beach, R. D. W., Jones, F. W., and Majorowicz, J. A. (1987). Heat flow and heat gener-
ation estimates for the Churchill basement of the Western Canadian Basin in Alberta,
Canada. Geothermics, 16 (1): 1–16.

Beckers, K. F., Yavuzturk, C. C., and Tester, J. W. (2014). Techno-Economic Modeling and
Monitoring of Hybrid Ground-Source Heat Pump System with Borehole Heat Exchangers for
Cooling-Dominated Cellular Tower Application. 11th IEA Heat Pump Conference, Mon-

treal, Canada.

Blackwell, D. and Spafford, R. (1982). Standard Method of Test for Thermal Conductivity of

Rock Using Divided Bar. Submitted to AMTS 1982. SMU Geothermal Laboratory, Dallas,

Texas, United States.

Blackwell, D. D. and Steele, J. L. (1989). Thermal conductivity of sedimentary rocks: mea-

surement and significance. In Naeser, N. D., McCulloh, T. H. (Eds.), Thermal History of

Sedimentary Basins. Springer-Verlag.

Čermák, V. and Rybach, L. (1982). Thermal conductivity and specific heat of minerals

and rocks. In Beblo, M. et al. (Eds.), Geophysics - Physical Properties of Rocks, Chapter:
Landolt-Börnstein Numerical Data and Functional Relationships in Science and Technology,
New Series, Group V, Geophysics and Space Research, Vol. 1, pages 305–343. Springer.

Diao, N., Li, Q. and Fang, Z. (2004). Heat transfer in ground heat exchangers with ground-
water advection. International Journal of Thermal Sciences, 43 (12): 1203–1211.

EPRI (1989). Soil and Rock Classification for the Design of Ground-Coupled Heat Pump Systems.
Electric Power Research Institute, EPRI CU-6600, Research Projects 2892-3 and 2892-5.
Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 146

Gehlin, S. (2002). Thermal Response Test. Model Development and Evaluation. PhD Disserta-
tion, Luleå University of Technology, Luleå, Sweden.

Gilliam, T. M. and Morgan, I. L. (1987). Shale: Measurement of thermal properties.

ORNL/TM-10499, Oak Ridge National Laboratory, Oak Ridge, Tennessee, United

States.

Goss, R. and Combs, J. (1976). Thermal conductivity measurement and prediction from geophys-

ical well log parameters with borehole application. Institute for Geosciences, The University
of Texas at Dallas, Richardson, Texas, United States.

Hellström, G. (1991). Ground Heat Storage: Thermal Analyses of Duct Storage Systems. PhD

Dissertation, Lund University, Lund, Sweden.

IGSHPA (2009). Ground Source Heat Pump Residential and Light Commercial Design and In-

stallation Guide. International Ground Source Heat Pump Association, Oklahoma State

University, Stillwater, Oklahoma, United States.

LaBrozzi, B., Dodge, E., and Tester, J. (2010). Utilization of Closed Loop Geothermal Heat

Pumps at Verizon Wireless Cellular Towers. Cornell Energy Institute, Cornell University,

Ithaca, New York, United States.

NYSDEC (2015). Final Supplemental Generic Environmental Impact Statement on the Oil, Gas
and Solution Mining Regulatory Program. New York State, Department of Environmental

Conservation.

Reiter, M. and Tovar, J. C. (1982). Estimates of terrestrial heat flow in northern Chihuahua,
Mexico, based upon petroleum bottom-hole temperatures. Geological Society of America
Bulletin, 93 (7): 613–624.

Rybach, L. and Muffler, L. J. P. (1981). Geothermal systems: principles and case histories. John

Wiley & Sons.

Chapter 8. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
Background Information and Varna Site 147

Schön, J. (2011). Physical properties of rocks: A workbook, volume 8. Elsevier.

Spitler, J. D. and Gehlin, S. E. (2015). Thermal response testing for ground source heat
pump systemsan historical review. Renewable and Sustainable Energy Reviews, 50: 1125–
1137.

Tester, J. W., Joyce, W. S., Brown, L., Bland, B., Clark, A., Jordan, T., Andronicos, C.,
Allmedinger, R., Beyers, S., Blackwell, D., Richards, M., Frone, Z., and Anderson, B.

(2010). Co-Generation Opportunities for Lower Grade Geothermal Resources in the

Northeast - A Case Study of the Cornell Site in Ithaca, NY. In GRC Transactions, 34:
440–448.

USDA (2015). Web Soil Survey. Soil Survey Staff, Natural Resources Con-

servation Service, United States Department of Agriculture. Available at

http://websoilsurvey.nrcs.usda.gov/.

Yang, W., Chen, Z., Shi, M., and Zhang, C. (2013). An in situ thermal response test for

borehole heat exchangers of the ground-coupled heat pump system. International Jour-

nal of Sustainable Energy, 32 (5): 489–503.

Zhang, C., Guo, Z., Liu, Y., Cong, X., and Peng, D. (2014). A review on thermal response

test of ground-coupled heat pump systems. Renewable and Sustainable Energy Reviews,

40: 851–867.
 CHAPTER 9
CORNELL-VERIZON HYBRID GEOTHERMAL HEAT PUMP PROJECT:
TRNSYS SYSTEM ANALYSIS

9.1 TRNSYS Systems Model

An in-depth analysis of the performance of hybrid geothermal heat pump systems for

cooling of cellular tower shelters requires the development of a versatile and rigorous
computer simulation model. A preliminary model was developed in MATLAB and pre-
sented at the 11th IEA Heat Pump Conference in Montreal (Beckers et al., 2014). Building

upon this model, a more detailed and elaborate simulation tool was developed using

the software platform TRNSYS (SEL, 2014a), which is presented in this section. TRNSYS

was chosen for its proven success with simulating geothermal heat pump systems (see

Section 7.3).

An example of the TRNSYS model developed for this study is shown in Figure 9.1.

This model consists of various blocks representing the different system components, e.g.

shelter, heat pumps, air-economizer (AE), dry-cooler (DC), etc., all discussed in the fol-
lowing sections. Simulation results utilizing this model are presented in Section 9.3. Most

components selected for use in the TRNSYS model are similar to those at the Varna site.

There were a few exceptions because of economic or technical reasons. For example, a
smaller DC is selected to better match the circulation flow rates and shelter heat require-
ments. Further, not six but about three to four borehole heat exchangers (BHEs) will min-
imize the total cost of ownership (TCO). Also, the reservoir heat transfer model assumes
conduction-only heat transfer, even though groundwater flow is present at the Varna Site.
Nevertheless, an “effective” reservoir thermal conductivity can, to a first approximation,
capture the effects of both conduction and advection.

148
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 149

Figure 9.1 – Example of TRNSYS model of hybrid geothermal heat pump system. This model corre-
sponds to Case 3 (GSHP + DC) studied in Section 9.3.

9.1.1 Shelter Component

The shelter which houses the telecommunication equipment and requires climate con-
trol is modeled using the lumped capacitance building component (Type 88) (SEL, 2014b)
and labeled as “Shelter” in Figure 9.1. Type 88 is a simplified single-zone building simu-
lation component which assumes the internal thermal mass can be represented by a single
building thermal capacitance parameter, and the heat exchange through the walls by the
overall external wall area and wall R-value. The Varna Site shelter parameter values have
been provided in Table 8.1. The shelter thermal capacitance has been experimentally de-
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 150

termined by measuring the shelter temperature drop over a 30 minute period when all
three heat pumps are in first stage cooling mode. Further, in the TNRYS model, as well
as with the actual system at the Varna Site, only the shelter dry-bulb temperature is con-
trolled, not the humidity. Based on logging data at the Varna Site, the relative humidity
inside the shelter fluctuates between 10 and 50%.

In calculating the heat exchange with the environment, only heat transfer through
the external walls and roof are considered; heat losses through the ground are neglected.
Also, the emergency diesel generator room, which is directly attached to one of the small
walls of the shelter, is not climate-controlled and has been neglected in the TRNSYS

model.

9.1.2 Alternating Current to Direct Current Plant Heat Generation

Component

The Alternating Current to Direct Current (ACDC) Plant converts alternating current to

direct current to provide power for the telecommunication equipment in the shelter. This

power is dissipated as heat inside the shelter, requiring the shelter to be constantly cooled.

Figure 9.2 shows the ACDC Plant heat generation during the first 30 days of December
2014. The heat generation stays fairly constant with an average value of 11.5 kWth . Since
the daily fluctuations are small, a constant value was assumed in the TRNSYS model
using the equation calculator. Because of heat exchange between the shelter and the envi-
ronment, the instantaneous cooling load fluctuates between about 10 and 12 kWth . In the
analysis in Section 9.3, different values for heat generation are considered because differ-
ent shelters across the country can house different telecommunication equipment. Also,
even for one shelter, the telecommunication equipment can be replaced over time, with
the adoption of new technology.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 151

Figure 9.2 – ACDC Plant heat generation in kW at Varna Site during first 30 days of December 2014.

9.1.3 Weather Component

The ambient (dry-bulb) temperature has an impact on the operation and performance of

the system at several levels. It determines whether or not the AE and DC can be utilized
at a given moment and directly affects their performance. It determines the heat losses or
gains through the shelter walls and the annual average ambient temperature is a proxy

for the local initial ground temperature. In the TRNSYS model, the ambient temperature
is incorporated with a TMY3 (Typical Meteorological Year 3) weather file (Wilcox and
Marion, 2008), using Type 15, as shown in Figure 9.1. Locally measured temperature data
at the Varna site can easily be converted into the TMY3 data format using Excel. For
the nationwide analysis, existing TMY3 weather files for hundreds of U.S. cities based on
historical climatic data are publicly available. For long-term simulations at the Varna Site,
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 152

the TMY3 weather file for the weather station at the Elmira Corning Regional Airport is
utilized.

Other weather aspects that have a minor influence on the system performance, but
that are not considered in this study, are: solar irradiance which impacts the shelter heat

exchange with the environment, ambient relative humidity which affects the AE and air-
source heat pump (ASHP) performance, and rain fall which can influence the rate of
underground heat dissipation.

9.1.4 Heat Pump Component

A TRNSYS macro component has been developed to represent a ClimateMaster Tran-

quility Two-Stage (TT) 026 water-to-air heat pump. This component is labeled HP Cli-

mateMaster TT 026 in Figure 9.1, and simulates the cooling performance of the heat pump

based on the manufacturer’s datasheet, which are included in Appendix A. The macro-

subcomponents are the TRNSYS equation calculator and Type 42 to read external data

files (see Figure 9.3). Calculated are the total and sensible cooling, power consumption,

heat rejection, and coefficient of performance (COP), based on the entering water tem-
perature, water flow rate, entering dry- and wet-bulb air temperature, air flow rate, and

whether the heat pump is operating in part load or full load.

Due to technical issues with the flow rate measurement system, accurate validation of
the TRNSYS heat pump component could not be performed with the data obtained at the
Varna Site. Nevertheless, calibrated data has been collected at a residential heat pump
system in Lansing, NY. This heat pump is the ClimateMaster Tranquility Two-Stage (TT)
049 water-to-air heat pump, which is the same model and from the same manufacturer
but of larger size than the Varna Site heat pumps. Figure 9.4 shows excellent agreement
between the measured fluid outlet temperature and simulated fluid outlet temperature
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 153

Figure 9.3 – Subcomponents in TRNSYS macro component for ClimateMaster TT 026 heat pump. HP,
PL and FL refer to heat pump, part load and full load, respectively.

using the same manufacturer’s datasheet, assuming the measured fluid inlet temperature

as known input in the model. The data shown is collected during August 2013 when

the heat pump was in cooling mode with fluid inlet temperatures of the same order as
at the Varna Site. The ambient temperature measured at the site is included to show the

positive correlation between the ambient temperature and measured fluid inlet and outlet

temperatures.

9.1.5 Borehole Heat Exchangers Component

A novel reservoir heat transfer model to simulate the BHEs has been developed, and is
presented and validated in Appendix B. TRNSYS does have built-in BHE models but their

use is not meant for short-term transient behavior on the order of minutes to hours, which
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 154

Figure 9.4 – Measured fluid inlet and outlet temperature, simulated fluid outlet temperature and mea-
sured local ambient temperature for ClimateMaster TT 049 residential heat pump system in Lansing,
NY. Excellent agreement exists between measured and simulated fluid outlet temperature, justifying
the use of the Manufacturer’s datasheet for the TRNSYS heat pump component. No ambient tempera-
ture data was collected prior to August 5th , 2013.

are the time scales occurring in this study. As mentioned above, this model assumes the

reservoir heat transfer is conduction only, which underestimates the heat dissipation in
the presence of groundwater flow, such as at the Varna Site. Nevertheless, groundwater
flow is not only difficult to model, but also challenging to measure and monitor. There-

fore, a conduction-only situation was assumed for modeling to represent the most con-
servative scenario. In addition, conduction and advection can, to a first approximation,
be lumped together into an “effective” thermal conductivity, as already mentioned in Sec-
tion 8.3.2, and further discussed in Section 9.3.3. The BHE model has been implemented
in MATLAB and is coupled to the TRNSYS model using Type 155, labeled as “BHE” in
Figure 9.1.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 155

The BHE model can handle various flow rates during the same simulation, e.g. when
alternating between one and two active heat pumps. Also, it is possible to simulate turn-
ing off the system temporarily, e.g. when only providing cooling with the AE. The cir-
culation pumps installed at the Varna Site can provide flow rates in the range 0 to about
40 GPM (0 to 2.5 l/s). With the DC turned off, a flow rate of 9 GPM requires about 0.2 kW
of electricity consumption (for 4 x 234 m of pipe length). The underground initial (and
far-field) temperature is assumed uniform at 9◦ C (48.2◦ F), the average annual temperature
from the Elmira Corning Regional Airport TMY3 weather data.

9.1.6 Dry-Cooler Component

The dry-cooler (DC) cools down the circulating fluid by forced cross flow heat exchange

with cold ambient air. Based on the manufacturer’s datasheet, the following cooling cor-

relation was derived for the model installed at the Varna site (Liebert 092 model):
9
QDC = 293.1 · · (T f,DC,supply − T amb ) · (−0.310m2 + 1.620m + 1.700) · c f (9.1)
5
with T f,DC,supply , T amb , m, and c f , the fluid supply temperature to the DC (in ◦ C), the ambient

temperature (in ◦ C), the fluid mass flow rate (in kg/s), and a correction factor dependent

on the fluid type and fluid supply temperature, respectively. For the water-ethylene gly-
col mixture in our case, c f has a value of 0.975 and about 0.94 for a T f,DC,supply of 26.67◦ C
(80◦ F) and 10◦ C (50◦ F), respectively. The Liebert 092 model is designed for fluid flow rates

between 18 to 36 GPM (1.1 to 2.3 l/s), much higher than the flow rates occurring at the
Varna site as well as in our TRNSYS simulations, which are in the range 3.5 to 7 GPM
(0.22 to 0.44 l/s). Therefore, the DC model selected for the TRNSYS model is the smallest
Liebert model (033), designed for flow rates in the range 6-12 GPM (0.38-0.76 l/s), and
with cooling correlation based on the manufacturer’s datasheet as follows:
9
QDC = 293.1 · · (T f,DC,supply − T amb ) · (−1.398m2 + 2.654m + 0.465) · c f . (9.2)
5
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 156

The DC component is included in the model using the equation calculator, and is labeled
“Dry-Cooler” in Figure 9.1.

9.1.7 Air-Economizer Component

The AE draws in cold outside air to efficiently cool the shelter without having to operate
the heat pumps. Thus, electricity use operating costs will decrease when using the AE,
however, maintenance increases and costs for replacing air filters are significant factors,
as well as the risk of equipment malfunction due to entering dust particles. The fan has

a rated air flow capacity of 802 l/s (1,700 CFM) and a power consumption of 560 W (0.75

HP) at 1 atm and 21.1◦ C (70◦ F). The nominal cooling capacity QAE (in W) is calculated as:

QAE = 0.802 · ρair · c p,air · (T shelter − T amb ) (9.3)

with ρair the air density (≈ 353/(273.15 + T amb )), c p,air the air specific heat capacity (≈1

kJ/(kg·K)), and T shelter the shelter temperature (≈26◦ C), respectively. The controller can

set different VFD setpoints for the AE fan. Following the affinity laws for axial and cen-

trifugal fans and pumps (Peng, 2008), it is assumed the AE cooling QAE scales linearly

and power WAE scales cubicly with the flow rate, respectively. Although this assumption
significantly simplifies the AE behavior (e.g. fan performance maps are neglected), the

impact on the overall system performance is minor because the electric power consump-
tion of the AE is much smaller than for the heat pumps. This component is modeled in
TRNSYS using the equation calculator. It is assumed in this study that the same correla-
tion for the AE (Equation (9.3)) holds for both hybrid GSHP and hybrid ASHP systems.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 157

9.1.8 Controller Component

The controller regulates the cooling systems to keep the temperature in the shelter around
a setpoint of 25.6◦ C (78◦ F). Figure 9.5 shows the shelter and ambient temperature during

the month of April and part of May in 2015. The shelter temperature fluctuates within a
hysteresis of about 1.7◦ C (3◦ F) between 25◦ C (77◦ F) and 26.7◦ C (80◦ F).

A controller component has been written and compiled in FORTRAN and incorpo-
rated in the TRNSYS model as an external DLL (Dynamic-Link Library), labeled “Con-
troller” in Figure 9.1. The controller regulates the heat pump units, circulation pump, AE

and DC to keep the shelter temperature between 25◦ C (77◦ F) and 26.7◦ C (80◦ F). Specific

information such as setpoints in different cases is provided in Section 9.3.

9.1.9 Air-Source Heat Pump Component

The performance of the hybrid geothermal heat pump system will be compared with the

current industry standard which is a wall-mounted HVAC system, with or without an

air economizer. The cooling capacity and performance are modeled in TRNSYS using
empirical correlations based on the performance datasheet of a single-stage traditional

Marvair wall-mounted HVAC unit (Beckers et al., 2014). The correlation for the cooling

capacity QAS HP (in kW) is:

QAS HP = 20.3 − 0.189 · T amb (9.4)

with T amb the ambient temperature in ◦ C. The electrical power consumption WAS HP (in
kW) is estimated as:
Q
WAS HP = . (9.5)
3.9234 − 0.0367 · T amb
No data is provided by the manufacturer on the effect of different shelter dry- and wet-
bulb temperatures on the ASHP cooling capacity and power consumption. Therefore, the
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 158

impact is assumed the same as with the ClimateMaster TT 026 water-to-air heat pump
(see Figure A.4 in Appendix A).

9.1.10 Performance Metrics and Financial Parameters

The financial performance metric investigated for a certain system (either GSHP or ASHP)
is the total cost of ownership or TCO (also called life cycle cost), calculated as:
∑
lt
CFi
TCO = (9.6)
t=0
(1 + id − iin f )i
with CFi the annual cash flow, and (id − iin f ) the net discount rate. It is assumed the cap-
ital costs are incurred in year zero (t = 0). Table 9.1 provides an overview of financial

parameters considered in calculating the TCO (Beckers et al., 2014). An electricity rate of

14 ¢/kWhe is based on average commercial end-use rates for NY State. The drilling cap-

ital cost, 50 $/m, is based on the market survey by Battocletti and Glassley (2013). Other

metrics analyzed are the operational energy consumption and operational CO2 emissions

over the lifetime of the system. For the Varna Site, the average CO2 emissions rate for

grid-generated electricity is assumed 186 g/kWhe , based on EPA (Environmental Protec-

tion Agency) eGRID 2012 data for New York State (EPA, 2015). The software MATLAB

(MathWorks, 2012) is utilized to perform the performance metrics calculations based on

the TRNSYS simulation results.

In the GSHP cases in Section 9.3, only two GSHP units are considered ($5,000 each),
unlike at the Varna Site where the number of units installed is three. It is assumed that the
backup cooling system in case one of the GSHP units fails is a small, inefficient HVAC unit
whose cost (∼$500) is included in CAPother ($5,000). Further, the ASHP units are replaced
after 10 years. All other components have a lifetime of 20 years. In Section 9.3.3, the
impact of several of the parameters of Table 9.1, including drilling cost and net discount
rate, on the TCO is investigated using a sensitivity analysis.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 159

Table 9.1 – TRNSYS model financial parameter values.

Inflation rate iin f 2%

Discount rate id 5%
System lifetime lt 20 years
Electricity rate el 14 ¢/kWhe
GSHP unit capital cost CAPGS HP $5,000
GSHP capital cost for pumps, piping & installation CAPother $5,000
Drilling capital cost CAPDrilling 50 $/m
DC capital cost CAPDC $1,000
AE capital cost CAPAE $1,000
ASHP unit capital cost CAPAS HP $2,500
AE maintenance cost MT NAE $200
GSHP maintenance cost MT NGS HP $200/year
ASHP maintenance cost MT NAS HP $580/year

9.2 Overall TRNSYS Model Validation

In Section 9.1, various individual model components have been validated (e.g. heat

pump, BHE) or are directly based on experimental measurements at the Varna Site (e.g.

shelter). In this section, an overall validation of the TRNSYS model (excluding BHEs,

DC and AE) is performed using data collected at the Varna Site during six weeks of high-

frequency logging (1 min. interval) from April 1 to May 13, 2015. During this time period,

the shelter setpoint temperature stayed constant, there were both cold and warm days,

and no equipment maintenance or testing occurred. The shelter and ambient tempera-

ture measured at the Varna Site are shown in Figure 9.5.

The TRNSYS model simulation incorporates the actual ambient temperature, heat

pump inlet temperature, and ACDC Plant heat generation measured at the site. The
circulating fluid supply temperature is not simulated for but directly based on the Varna
Site measurements because the BHE model does not include groundwater advection and
due to the lack of data on the local groundwater flow. The TRNSYS model controller alter-

nates between one heat pump operating in part load and two heat pumps in full load (as
the Varna Site controller operates), to keep the shelter temperature between 25◦ C (77◦ F)
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 160

Table 9.2 – Overall TRNSYS model validation results. Validation is based on six weeks of high-
frequency data (1 min. interval) collected at the Varna Site from April 1st to May 13th , 2015. A flow rate
4, 5, and 6 GPM corresponds to 0.26, 0.32, and 0.38 l/s. HP refers to heat pump. A heat pump power
factor of 0.9 is assumed for calculating the electricity consumption.

TRNSYS TRNSYS TRNSYS

Varna Site
Metric Model Model Model
(Measured)
4.0 GPM 5.0 GPM 6.0 GPM
HP Electricity
2010 kWh 2020 kWh 1913 kWh 1850 kWh
Consumption
HP Part Load
325 h 380 h 399 h 409 h
Running Time
HP Full Load
1321 h 1272 h 1235 h 1214 h
Running Time

and 26.7◦ C (80◦ F). Due to flow meter technical issues, no accurate flow measurements

have been obtained. However, approximate flow rates have been determined based on

the temperature measurements as well as independent measurements with an acoustic

clamp-on flow meter. The average flow rate per heat pump running is believed to be be-

tween 4.0 and 7.0 GPM (0.26 and 0.44 l/s) and therefore TRNSYS simulations for a range

of flow rates have been performed.

The Varna Site measurements and TRNSYS model simulation results are provided in

Table 9.2. The data collected on heat pump voltage, current, and running time are reli-

able and therefore the comparison with the simulations is based on electricity consump-
tion and running time. In addition, the electricity consumption drives the TCO and is

therefore arguably the most important parameter to validate. The results shows that the

TRNSYS model assuming a flow rate between 4.0 and 6.0 GPM (0.26 and 0.38 l/s) per
heat pump running, matches approximately the electricity consumption and heat pump

running time data.

Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 161

Figure 9.5 – Varna Site shelter and ambient temperature measured during April and May 2015. Plotting
interval is 5 minutes.

9.3 System Analysis

9.3.1 Overview of Cases Studied

Various GSHP and ASHP configurations have been investigated. Five of them, believed
to be the most relevant, are presented in detail below. Their base case parameters are

summarized in Table 9.3. All cases consider the Varna shelter as described in Table 8.1,
the weather based on the Elmira TMY3 climatic data, and 11.5 kWth of heat generation
in the base case scenario. The time step in the TRNSYS simulations is 10 minutes for the
GSHP cases, and 1 minute for the ASHP cases. A shorter time step is required in the
ASHP cases because the shelter cools down and heats up faster. The reason for this is that
the ASHP cooling (Equation (9.4)) fluctuates between 0 and up to 20 to 30 kWth while the
GHSP units working together provide more moderate cooling around 11 kWth .
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 162

Table 9.3 – TRNSYS base case parameters. BHE field data refers to properties of circulating fluid,
pipe, grout and BHE inner geometry. Number and depth of BHEs is determined on a case by case to
minimize the TCO.

Case 1: Case 2: Case 3: Case 4: Case 5:

Parameter
GSHP GSHP + AE GSHP + DC ASHP ASHP + AE
Weather Data Elmira Corning Regional Airport TMY3
Shelter Data as in Table 8.1
Financial Data as in Table 9.1
Heat Generation 11.5 kWth
◦
AE Setpoint N/A 10 C N/A N/A 10◦ C
DC Setpoint N/A N/A 15◦ C N/A N/A
BHE Field Data as in Table 8.2 N/A N/A
BHE Spacing 10 m N/A N/A
Flow Rate per HP on 3.5 GPM N/A N/A
−6 2
Reserv. Therm. Diff. 10 m /s N/A N/A
Reserv. Therm. Cond. 2.0 W/(m·K) N/A N/A

Case 1: GSHP only

The first case is the “GSHP only” case (case (a) in Figure 7.3) in which all the cooling is

provided using two ClimateMaster TT 026 geothermal heat pump units (Table 8.1). Since

the heat pumps operate most efficiently in first stage, the controller is set to run one heat

pump in first stage and the second heat pump alternating between first and second stage,

keeping the shelter temperature between 77◦ F (25◦ C) and 80◦ F (26.7◦ C). If the cooling

requirements are not met on warm summer days, both heat pumps run in second stage.
If too much cooling is provided during cold winter days, the second heat pump alternates

between off and first stage, while the first heat pump runs permanently in first stage.

The dominant rock type in the geothermal reservoir at the Varna Site is shale with
expected thermal conductivities around 1 W/(m·K) (see Table 8.4 and Section 8.3.3).
Groundwater flow is present which likely explains the high effective ground thermal
conductivity of 4 W/(m·K) obtained using a TRT at the site. The effective thermal con-
ductivity occurring at any moment in the reservoir can be expected between these two
values and therefore the BHE model assumes a conduction-only reservoir with thermal
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 163

conductivity of 2 W/(m·K) in the base case scenario. Different effective reservoir thermal
conductivities are investigated using a sensitivity analysis. The BHE field consists of three
1-U BHE in parallel, each with depth of 150 m, which minimizes the TCO for this case.
Pipe dimensions and material properties, as well as grout and circulating fluid properties
are the same as at the Varna site (see Table 8.2). The flow rate is set at 3.5 GPM (0.22 l/s)
per heat pump running, both for first or second stage operation, slightly lower than at the
Varna site in order to lower the pump power. The eventual pump power requirements
are case-specific and difficult to calculate exactly and therefore conservative values are

assumed: 3.5 GPM (0.22 l/s) requires 0.05 kW and 7.0 GPM (0.44 l/s) requires 0.2 kW.

Case 2: GSHP + AE

The second case is similar to the first case but with an AE providing the cooling when the

ambient temperature drops below 10◦ C (50◦ F) (case (b) in Figure 7.3). The VFD of the AE

is controlled in order to provide cooling around 11 kWth . The TCO is minimized using

only two BHE’s in parallel, each with depth of 150 m. When the AE is activated, the heat

pumps and circulation pump are turned off.

Case 3: GSHP + DC

The third case is similar to the first case but a DC supplies supplemental heat rejection
(case (d) in Figure 7.3). When the ambient temperature drops below 15◦ C (59◦ F), the
circulating fluid leaving the heat pump units is cooled by the DC only and bypasses the
BHE field. The TCO is minimized using a BHE field consisting of two boreholes with 130
m depth each. Other configurations have been investigated including the DC in series
with the BHE field. However, those configurations resulted in higher TCOs than the
configuration selected here.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 164

Following the rules of thumb presented in Section 7.3 on design of supplemental heat
rejection for hybrid geothermal heat pump systems, the DC in the base case does not
provide annual net reservoir heat balance but covers roughly 65% of the cooling load.
Other setpoints have been analyzed but 15◦ C is believed to be a good compromise: with
a too low setpoint, the system does not taking full advantage of the efficient DC cooling
capability with cold ambient air; a too high setpoint means lower cooling returns for the
electricity invested.

Case 4: ASHP only

In the “ASHP only” case, all cooling is provided using an industry-standard wall-

mounted Marvair HVAC unit. This option has the lowest initial investment cost but a

high electricity consumption over the lifetime of the system.

Case 5: ASHP + AE

The fifth case is similar to the fourth case but with an AE providing cooling when the

ambient temperature drops below 10◦ C (50◦ F). Again, the VFD of the AE is set to provide

AE cooling fluctuating around 11 kWth , as in Case 2.

9.3.2 Base Case Simulation Results and Discussion

The simulation results for the five cases are presented in Table 9.4. Provided are the TCO
in $1,000, the electricity consumption in MWhe , the CO2 emissions in million metric tons,
the average heat pump COP, and the BHE configuration that minimizes the TCO for the
case considered. The electricity consumption and CO2 emissions are calculated over the
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 165

Table 9.4 – TRNSYS base case simulation results. Lifetime electricity consumption and CO2 emissions
do not account for energy consumption and emissions during construction and installation of the sys-
tem.

Electricity CO2 Emissions Average

TCO BHE
Case Consumption (million metric HP COP
($1,000) Configuration
(MWhe ) tons) [-]
GSHP 87.6 460 85.6 4.21 3 x 150 m
GSHP + AE 59.9 222 41.3 4.27 2 x 150 m
GSHP + DC 75.7 421 78.2 5.47 2 x 130 m
ASHP 78.7 589 109.6 3.28 N/A
ASHP + AE 54.0 314 58.4 3.02 N/A

lifetime of the system but do not account for the installation. The average heat pump COP
is calculated as the lifetime average heat pump cooling divided by the lifetime average

heat pump power and circulation pump power (if applicable).

Case 5 (ASHP + AE) has the lowest TCO for the parameters assumed in the base case

scenario, followed by Case 2 (GSHP + AE). The systems with AE perform well because

the cold Upstate New York weather allows the use of an AE, which provides cooling

very efficiently, for a significant amount of time throughout the year. If the use of an AE

is undesirable for reasons highlighted in the Section 8.2 such as increased maintenance,

the the GSHP + DC has the lowest TCO of the three remaining cases. Advantageous for

ASHP-based systems is the low upfront capital costs, moderate ambient temperatures

which result in reasonable COP’s, no temperature built-up which negatively impacts the
performance of GSHP systems over time, and no additional power requirements, e.g. for

circulation pumps as with GSHP-based systems.

If the goal is to minimize energy consumption and CO2 emissions, then GSHP-based
systems perform best. While the cases presented in Table 9.4 have been optimized for
minimum TCO, even lower amounts of electricity consumption and CO2 emissions can
be obtained by increasing the total BHE depth, which only slightly increases the TCO.
This is further discussed in Section 9.3.3 where the impact of several parameters on the
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 166

performance metrics is studied with a sensitivity analysis.

The results presented in this and the next section should not be interpreted as optimal

or final results but rather as representative numbers given the whole list of parameters in
Table 9.3. Parameters not studied in detail but believed to only have a minor role are the

inner BHE geometry and the heat pump manufacturer. Different AE and DC setpoints
have been explored but not necessarily optimized for. Different control strategies of the
DC, e.g. operating the DC at variable speeds depending on the ambient weather and
fluid temperature, are available but are not analyzed here. The heat exchange between

the shelter and ground, and the shelter and attached diesel generator room has been ne-

glected. Nevertheless, based on first order calculations, this simplification is expected to

only have a small impact. The weather is expected to have a major impact on the system

performance, as already highlighted in Section 9.3. However, the study of its impact is

part of the ongoing nationwide analysis in our group and is not further discussed in this

work. Further, the economic results do not account for any subsidies, tax incentives or a

potential carbon tax. Also leasing costs for land for the BHE field, which likely increase

with number of BHE, are not included.

Where possible, correlations and models that describe the various system components

have been validated or are directly based on measurements at the Varna Site, including for
the heat pumps, shelter, circulation pump, BHEs, controller, and ACDC plant. However,

inherent TRNSYS model uncertainty persists due to uncertainties with measurements and
manufacturer’s datasheet correlations, varying operating conditions, and model simpli-
fications. In addition, the performance of some components, e.g. ASHP and DC, is solely
based on the manufacturer’s datasheet and has not been independently verified.

The circulation pump power, up to 0.7 kWe at the Varna Site, has a significant impact
on the overall TCO and performance of the system. Lowering the pump power can be
achieved by lowering the flow rates (which has been done for the five TRNSYS cases) but
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 167

this also impacts the heat pump performance. For the same heat pump inlet tempera-
ture, a lower flow rate results in a slight decrease in heat pump COP. However, a lower
flow rate typically leads to lower inlet temperatures which increases the performance and
therefore the overall impact cannot directly be derived. However, much lower flow rates
are discouraged because they would fall outside the recommended range based on the
manufacturers datasheet (minimum 2.3 GPM (0.15 l/s) in part load and 3 GPM (0.19 l/s)
in full load for one heat pump). Further, in the TRNSYS simulation, the circulation pump
power correlation is kept the same for each BHE configuration, although a lower total

BHE depth will decrease the pump power. Another way to decrease the pump power is
to utilize a circulating fluid with lower viscosity. However, a low fluid freezing point is

required for Case 3 (GSHP + DC) and desired for Case 2 (GSHP + AE), limiting the fluid

options. Maybe in Case 1 pure water can be utilized since the heat pump mode is cooling

only but then the system may not be shut down, even temporarily, during the winter to

prevent any fluid freezing. All together, it is believed a small decrease in TCO, maybe

up to 5 or 10%, for the GSHP cases may be obtained in case of circulation flow rate and

pump optimization, but this is not further investigated here.

Different system configurations are possible besides the five cases presented in this

chapter. One case is the GSHP + AE + DC as investigated in one configuration by Beckers

et al. (2014). Due to the additional operational complexity and cost, this case is not further
analyzed here. Another system is the GSHP + DC but with the DC in series with the BHE

field. Based on a first analysis with the TRNSYS model, this configuration did not turn
out to perform better than the configuration in Case 3. Moreover, other technologies
are available, e.g. dual stage ASHP, GSHP with horizontal heat exchangers, or direct
exchange GSHP, but they are not investigated here.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 168

9.3.3 Sensitivity Analysis

A sensitivity analysis has been performed to address the uncertainty of several of the
model parameters, as well as to assess the performance of each system configuration for

different shelter operational and financial conditions. Specifically, the impact of total BHE
depth, shelter heat generation, reservoir thermal conductivity, electricity rate, drilling
costs, and net discount rate on TCO, electricity consumption and CO2 emissions has been
investigated and the results are presented in this section.

Total BHE Depth Sensitivity

For each case involving a GSHP system (Cases 1, 2 and 3), the TCO has been minimized

with respect to the total BHE depth. Increasing the total BHE depth, either by increasing

the number of BHEs or increasing the depth of each individual BHE, generally decreases

the average circulating fluid supply temperature, which increases the average COP of

the heat pumps and therefore lowers the lifetime electricity consumption. For example,

Figure 9.6 shows the lifetime average heat pump COP and electricity consumption as a

function of total BHE depth for Case 1 (GSHP only) in the base case scenario. The total
number of boreholes is three, four and six if the total BHE depth is in the range 360 to

450 m, 480 to 600 m, and 630 to 720 m, respectively. The non-monotonic behavior can

occur when the total BHE depth is adjusted by changing simultaneously the number of
boreholes and the individual BHE depth, for example, 600 m = 4 x 150 m and 630 m = 6
x 105 m. Figure 9.6 illustrates that the system performance depends considerably on the
total BHE depth, which was also highlighted in other works, for example (Casasso and
Sethi, 2014) and (Esen and Turgut, 2015).

Although the operational electricity costs decrease with increasing total BHE depth,
the capital investment costs increase due to higher drilling costs. The effect of the total
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 169

Figure 9.6 – Effect of total BHE depth on lifetime average heat pump COP (red crosses; left axis) and
lifetime electricity consumption (blue bars; right axis) for Case 1 (GSHP only) in the base case scenario.
The total number of boreholes is three, four and six for a total BHE depth of 360 to 450 m, 480 to 600
m, and 630 to 720 m.

Figure 9.7 – Effect of total BHE depth on TCO for Case 1 (GSHP only) in base case scenario. TCO
is broken down into capital cost (bottom red bars) and O&M cost (top blue bars). The O&M cost
includes the costs for electricity consumption and maintenance (see also Table 9.1). The total number
of boreholes is three, four and six for a total BHE depth of 360 to 450 m, 480 to 600 m, and 630 to 720 m.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 170

BHE depth on the TCO, consisting of both the capital and discounted lifetime O&M costs,
is shown in Figure 9.7 for Case 1 in the base case scenario. This figure shows that the TCO
is weakly dependent on the total BHE depth. The minimum TCO was $87,600 for a total
BHE depth of 450 m (3 x 150 m). For 540 m (4 x 135 m), the TCO is very close with $87,800,
and the lifetime electricity consumption is only 417 MWhe (9% decrease). Full simulation
results are provided in Table C.1 in Appendix C.

The minor effect of the total BHE on the TCO, and considerable effect on the energy
consumption around the minimum was found for each case studied involving a GSHP

system. Therefore, if the goal is to decrease the energy consumption, one could slightly

increase the total BHE depth from the numbers reported in Table 9.4 and in the tables in
Appendix C, which all report BHE depths that minimize the TCO. Nevertheless, it is not

recommended to increase the individual BHE depth beyond 150 m, because the effect of

the geothermal gradient will start to influence performance, and there is a likely increase

in risks in drilling and completing the wells. Equally important is that drilling permits

are required in the U.S. for depths greater than 500 ft (152.4 m).

Shelter Heat Generation Sensitivity

Technology and size of telecommunication equipment can differ among shelters and even
for the same shelter over time, and therefore different amounts of heat generated can be

expected. Figure 9.8 shows the dependence of the TCO on the shelter heat generation for
the different cases in Section 9.3.1, keeping all other parameters the same. The heat gen-
eration of 11.5 kWth in the base case scenario is believed to be on the high end side and
therefore smaller amounts of heat generation are considered. Lowering the heat genera-
tion lowers the TCO across the board for all five cases, and decreases the spread between
the highest and lowest TCO (Case 1 and Case 5). Full simulation results are provided in
Table C.2 in Appendix C.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 171

Figure 9.8 – Effect of shelter heat generation on TCO for Cases 1 to 5. All other parameters are the same
as in the base case scenario.

Reservoir Thermal Conductivity Sensitivity

In the base case scenario, the effective reservoir thermal conductivity was set at 2

W/(m·K) as a compromise between the high value of 4 W/(m·K) obtained using a TRT

and a low value of 1 W/(m·K), experimentally measured and often reported in literature

as thermal conductivity for dry shale, the dominant rock type in the reservoir. The effect
of a different thermal conductivity on the TCO for Cases 1 to 3 is shown in Figure 9.9,
keeping all other parameters the same. The TCO for Cases 4 and 5 are independent of the
reservoir thermal conductivity and are solely included in the plot for easy comparison. As
expected, a higher thermal conductivity allows for a lower total BHE depth and decreases
the TCO of the system. Full simulation results are provided in Table C.3 in Appendix C.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 172

Figure 9.9 – Effect of reservoir thermal conductivity on TCO for Cases 1 to 3. TCO for Cases 4 and 5 are
independent of reservoir thermal conductivity and are included for comparison. All other parameters
are the same as in the base case scenario.

Electricity Rate Sensitivity

The electricity rate of 14 ¢/kWhe in the base case scenario is relatively high in comparison

with electricity rates in other states, but might be low in comparison with future electricity

rates. Figure 9.10 shows the impact of the electricity rate on the TCO for all five cases

studied. As expected, higher electricity rates penalize systems with higher electricity
consumption, e.g. Case 4 (ASHP). Full simulation results are provided in Table C.4 in

Appendix C.

Net Discount Rate Sensitivity

A net discount rate of 3% in the base case scenario reflecting a long-term horizon might
be a too low discount rate for the fast-changing telecommunication industry. Figure 9.11
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 173

Figure 9.10 – Effect of electricity rate on TCO for Cases 1 to 5. All other parameters are the same as in
the base case scenario.

shows the impact of the discount rate on the TCO for all five case studied. As expected,

a higher discount rate lowers the TCO for all cases since future expenses are valued less

today. For the other assumptions used, systems with lower upfront capital costs (ASHP-

based systems) are favored with increasing discount rate. Full simulation results are pro-
vided in Table C.5 in Appendix C.

Drilling Cost Sensitivity

A drilling cost of 50 $/m used for the Varna Site might be high in comparison with drilling
costs observed in other regions in the U.S. (Battocletti and Glassley, 2013), but too low
in the case trouble arises during the drilling process. Figure 9.12 shows the effect of the
drilling costs on the TCO for Cases 1 to 3. Cases 4 and 5 are included for easy comparison.
As expected, higher drilling costs increase the TCO of the GSHP systems and also shifts
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 174

Figure 9.11 – Effect of net discount rate on TCO for Cases 1 to 5. All other parameters are the same as
in the base case scenario.

Figure 9.12 – Effect of drilling cost on TCO for Cases 1 to 3. TCO for Cases 4 and 5 are independent
of drilling cost and are included for comparison. All other parameters are the same as in the base case
scenario.
Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 175

the optimum towards shallower total BHE depths. Full simulation results are provided
in Table C.6 in Appendix C.

References

Battocletti, E. C. and Glassley, W. E. (2013). Measuring the Costs and Benefits of Nation-
wide Geothermal Heat Pump Deployment. Report prepared for U.S. Department of
Energy, Geothermal Technologies Program under Award Number DE-EE0002747.

Beckers, K. F., Yavuzturk, C. C., and Tester, J. W. (2014). Techno-Economic Modeling and

Monitoring of Hybrid Ground-Source Heat Pump System with Borehole Heat Exchangers for

Cooling-Dominated Cellular Tower Application. 11th IEA Heat Pump Conference, Mon-

treal, Canada.

Casasso, A. and Sethi, R. (2014). Sensitivity analysis on the performance of a ground

source heat pump equipped with a double u-pipe borehole heat exchanger. Energy

Procedia, 59 : 301–308.

EPA (2015). eGRID2012. U.S. Environmental Protection Agency, Washington, DC, United

States.

Esen, H. and Turgut, E. (2015). Optimization of operating parameters of a ground coupled

heat pump system by taguchi method. Energy and Buildings, 107 : 329–334.

MathWorks (2012). MATLAB Release 2012a, The Mathworks, Inc., Natick, Massachusetts,
United States.

Peng, W. W. (2008). Fundamentals of Turbomachinery. John Wiley & Sons.

SEL (2014a). TRNSYS 17 (TRaNsient SYstems Simulation program). Univer-

Chapter 9. Cornell-Verizon Hybrid Geothermal Heat Pump Project:
TRNSYS System Analysis 176

sity of Wisconsin - Madison, Solar Energy Laboratory (SEL). Available at

http://sel.me.wisc.edu/trnsys/.

SEL (2014b). TRNSYS 17 (TRaNsient SYstems Simulation program) Documentation - Volume

4 Mathematical Reference . University of Wisconsin - Madison, Solar Energy Laboratory

(SEL). Available at http://sel.me.wisc.edu/trnsys/.

Wilcox, S. and Marion, W. (2008). Users Manual for TMY3 Data Sets, Technical Report

NREL/TP-581-43156. National Renewable Energy Laboratory (NREL), Golden, Col-

orado, United States.
 CHAPTER 10
TECHNO-ECONOMIC ANALYSIS OF HYBRID GEOTHERMAL HEAT PUMP
SYSTEMS FOR COOLING-DOMINATED APPLICATIONS: KEY CONCLUSIONS

As part of an ongoing study between Cornell and Verizon Wireless to investigate the per-
formance of hybrid geothermal heat pump (GSHP) systems for cooling of cellular tower
shelters, a TRNSYS model simulating GSHP and air-source heat pump (ASHP) systems
has been developed. The model has been successfully tested with data from a full-scale

hybrid GSHP system in Varna, NY, and has been applied to test various systems config-
urations under different parameters. For conditions at the Varna Site, in particular Up-

state New York weather, shelter configuration and requirements, borehole heat exchanger

(BHE) field and fluid properties, and for financial conditions assumed in Chapter 9 not

considering any incentives, subsidies or taxes, the following conclusions can be drawn,

specifically for conditions at the Varna Site:

• An ASHP combined with an air-economizer (AE) has the lowest TCO, followed by

a GSHP with AE. If an AE is to be omitted, a GSHP + DC is the following option

with lowest total cost of ownership (TCO). A GSHP only system has the highest
TCO illustrating benefits of hybridizing a GSHP system.

• ASHP systems obtain low TCO because of: (1) low upfront capital costs for equip-
ment and design, (2) no deteriorating performance due to reservoir temperature

increase over time as with GSHP systems, and (3) good overall performance be-
cause of moderate ambient temperatures at the Varna Site and no auxiliary power
requirements such as circulation pumps with GSHP systems.

• If the goal is to minimize energy consumption and CO2 emissions, GSHP-based

cases perform best. Increasing the total BHE depth can significantly increase the

177
Chapter 10. Techno-Economic Analysis of Hybrid Geothermal Heat Pump Systems
for Cooling-Dominated Applications: Key Conclusions 178

energetic performance and lower the energy consumption, while only slightly in-
creasing the TCO.

• Favoring GSHP-based systems are high ground thermal conductivities, low drilling

costs, low discount rates, high electricity rates, and low set-point temperatures.

The TRNSYS model developed is being applied to other regions in the country as part
of the ongoing nationwide analysis of utilizing hybrid GSHP systems for cooling of cellu-
lar tower shelters. It is expected that the local weather and average ground temperatures

have a major impact on the performance of each configuration studied.

Not every possible configuration has been analyzed, not all cooling or heat pump

technologies have been considered, and of the systems studied, not a full rigorous opti-

mization has been performed since there are simply too many degrees of freedom. Nev-

ertheless, it is believed the results are representative for the specific conditions used for

the cases investigated, emphasizing that hybrid GSHP systems are a means of saving en-

ergy and CO2 emissions, but not necessarily for saving money under these conditions.

While the focus has been on cooling of cellular tower shelters, the results are expected to

be transferable to other cooling-dominated applications, e.g. cooling of walk-in fridges

and freezers in the food and beverages industry.

 Part III

Slender-Body Theory for Transient Heat

Conduction

179
 CHAPTER 11
INTRODUCTION TO SLENDER-BODY THEORY

An earlier publication (Beckers et al., 2015) covers the material presented in Part III.

Simulating and predicting the heat exchange of slender bodies undergoing transient
heat transfer in conductive media is of great interest and importance in the design and
analysis of several applications. Examples are (1) wellbore heat transmission in oil, gas
or geothermal wells, (2) reservoir heat transfer in ground-source heat pumps or under-

ground thermal energy storage systems using slinky-coil, horizontal pipe or borehole

heat exchangers, and (3) cooling or heating elements in concrete foundations and indus-
trial processes.

Various heat transfer solution methods are available and in some cases sufficient, de-

pending on the complexity and compactness of the slender body heat transfer problem.

Simple geometries, e.g. a straight finite cylinder, can be handled with analytical methods

such as the Laplace transform, separation of variables and Bessel functions, Green’s func-

tions and perturbation methods (Carslaw and Jaeger, 1959; Arpaci, 1966; Myers, 1971;

Aziz and Na, 1984; Beck et al., 1992; Ozisik, 1993). More complex but compact geome-

tries, meaning the slender body is e.g. irregularly curved but closely packed together,
can be simulated with finite element (FEM) or finite volume methods (Patankar, 1980;

Minkowycz et al., 2006). Nevertheless, using these numerical methods for non-compact
geometries typically requires long computational times in order to achieve accurate re-
sults. In this paper, we present a hybrid method that efficiently handles these type of
geometries. The method uses slender-body theory, an asymptotic analysis to allow for thin
non-compact geometries, in combination with a numerical implementation to allow these
geometries to have complex shapes. Hereby, one can reduce from discretizing a three-
dimensional domain to discretizing over a one-dimensional axis of the slender body. Es-

180
Chapter 11. Introduction to Slender-Body Theory 181

pecially for longer time-scales, when the domain needed to be studied becomes O(Lt3 ) with
Lt the total slender body length, much shorter computational times can be obtained and
much less computer memory would be needed. However, without smart implementation
strategies, the computational time is O(Nt ·NE3 + Nt2 ·NE2 ), while for FEM only O(NE3 + Nt ·NE2 ),
with NE and Nt the number of elements and time steps, respectively. Also, the method
requires a linear problem so no temperature dependence of the thermal conductivity is
allowed. Further, simple medium geometries are desired so that the Green’s function is
either unbounded or can be readily found, e.g. with method of images.

Slender-body theory (SBT) in general refers to approximating the equations or un-

derlying physics in a specific science field when dealing with bodies that are long in one

dimension and short in the other two. These approximations are typically achieved by ap-

plying matched asymptotic expansions of inner and outer solutions, although sometimes

only the inner or outer solution is considered. It originated and is most commonly used in

the field of fluid dynamics. Munk (1924) was the first to apply slender-body approxima-

tions in order to calculate the aerodynamic forces on airship hulls. He treated the airship

as an elongated body of revolution in a potential flow and argued that when calculating the

velocity potential and air forces, different longitudinal cross-sections can be considered

independent from each other. In 1927, von Kármán applied a different SBT approach in
solving the same problem using a distribution of monopoles and dipoles on the axis of

symmetry (von Kármán, 1927). Other examples of potential flow problems simplified
with SBT are the study of wings (Jones, 1946), not-so-slender wings (Adams and Sears,
1953), ship hydrodynamics (Newman, 1964, 1970), and fish locomotion (Lighthill, 1960,
1971). Slender bodies in Stokes flow have been widely studied using SBT methods as well.
Examples are the general works by Cox (1970; 1971), Keller and Rubinow (1976), Batche-
lor (1970), and Johnson (1980), and the studies on flagellar hydrodynamics (Johnson and
Brokaw, 1979), fiber particle suspensions (Mackaplow and Shaqfeh, 1996, 1998), and mo-
tion of a slender torus Johnson and Wu (1979). Other applications of SBT can be found
Chapter 11. Introduction to Slender-Body Theory 182

in the fields of acoustic scattering (Junger and Feit, 1972; Ye et al., 1997), electromagnetic
scattering (Geer, 1980), charged colloids (Chen and Koch, 1996), and steady-state heat
transfer. A steady-state heat transfer problem often approximated using SBT is the heat
transport in composites with slender inclusions (Rocha and Acrivos, 1973a,b; Chen and
Acrivos, 1976; Acrivos and Shaqfeh, 1988; Shaqfeh, 1988; Fredrickson and Shaqfeh, 1989;
Mackaplow et al., 1994), e.g. to derive the effective thermal conductivity. To the best of
our knowledge, SBT has never been applied to transient heat conduction.

In Chapter 12, the slender body geometry, constraints and heat transfer problem are
formally defined. A theoretical framework for solving the problem using Green’s func-

tions and matched asymptotic expansions is derived and the result is physically inter-

preted. Further, model extensions are discussed to treat various time-scales, medium

properties and initial and boundary conditions. In Chapter 13, the theoretical basis is

translated into a numerical method and strategies are provided for efficient implementa-

tion in order to obtain high speed, accuracy and versatility. Finally, these strengths are

demonstrated in Chapter 14 using two case-studies in which the presented SBT numeri-

cal model is compared with a finite element model. The first case-study is the simulation

of the heat exchange of two parallel finite cylinders at constant temperature; the second

case-study is the simulation of one loop of a slinky-coil heat exchanger.

References

Acrivos, A. and Shaqfeh, E. (1988). The effective thermal conductivity and elongational
viscosity of a nondilute suspension of aligned slender rods. Physics of Fluids (1958-1988),
31 (7): 1841–1844.

Adams, M. C. and Sears, W. R. (1953). Slender-Body Theory-Review and Extension. Jour-

nal of the Aeronautical Sciences (Institute of the Aeronautical Sciences), 20 (2): 85–98.
Chapter 11. Introduction to Slender-Body Theory 183

Arpaci, V. S. (1966). Conduction heat transfer. Addison-Wesley Pub. Co., Reading, Mass.

Aziz, A. and Na, T. Y. (1984). Perturbation methods in heat transfer. Hemisphere Publishing
Corporation, Washington DC.

Batchelor, G. K. (1970). Slender-body theory for particles of arbitrary cross-section in

stokes flow. Journal of Fluid Mechanics, 44 (03): 419–440.

Beck, J. V., Cole, K. D., Haji-Sheikh, A., and Litkouhi, B. (1992). Heat conduction using
Green’s functions. Hemisphere Publishing Corporation, Washington DC.

Beckers, K. F., Koch, D. L., and Tester, J. W. (2015). Slender-body theory for transient

heat conduction: theoretical basis, numerical implementation and case studies. Pro-
ceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences,

471 (20150494).

Carslaw, H. S. and Jaeger, J. C. (1959). Conduction of heat in solids. Clarendon Press, Oxford.

Chen, H. S. and Acrivos, A. (1976). On the effective thermal conductivity of dilute suspen-

sions containing highly conducting slender inclusions. Proceedings of the Royal Society of

London A: Mathematical, Physical and Engineering Sciences, 349 (1657): 261–276.

Chen, S. B. and Koch, D. L. (1996). Electrophoresis and sedimentation of charged fibers.

Journal of colloid and interface science, 180 (2): 466–477.

Cox, R. G. (1970). The motion of long slender bodies in a viscous fluid. Part 1. General
theory. Journal of Fluid mechanics, 44 (04): 791–810.

Cox, R. G. (1971). The motion of long slender bodies in a viscous fluid. Part 2. Shear flow.
Journal of Fluid Mechanics, 45 (04): 625–657.

Fredrickson, G. H. and Shaqfeh, E. S. (1989). Heat and mass transport in composites of

aligned slender fibers. Physics of Fluids A: Fluid Dynamics (1989-1993), 1 (1): 3–20.
Chapter 11. Introduction to Slender-Body Theory 184

Geer, J. (1980). Electromagnetic scattering by a slender body of revolution: axially incident

plane wave. SIAM Journal on Applied Mathematics, 38 (1): 93–102.

Johnson, R. E. and Brokaw, C. J. (1979). Flagellar hydrodynamics. A comparison between

resistive-force theory and slender-body theory. Biophysical journal, 25 (1): 113–127.

Johnson, R. E. and Wu, T. Y. (1979). Hydromechanics of low-Reynolds-number flow. Part

5. Motion of a slender torus. Journal of Fluid Mechanics, 95 (02): 263–277.

Johnson, R. E. (1980). An improved slender-body theory for stokes flow. Journal of Fluid

Mechanics, 99 (02): 411–431.

Jones, R. T. (1946). Properties of low-aspect-ratio pointed wings at speeds below and above the
speed of sound. National Advisory Committee for Aeronautics, Washington DC.

Junger, M. C. and Feit, D. (1972). Sound, structures, and their interaction. MIT Press.

Keller, J. B. and Rubinow, S. I. (1976). Slender-body theory for slow viscous flow. Journal

of Fluid Mechanics, 75 (04): 705–714.

Lighthill, M. J. (1960). Note on the swimming of slender fish. Journal of fluid Mechanics, 9

(02): 305–317.

Lighthill, M. J. (1971). Large-amplitude elongated-body theory of fish locomotion. Pro-

ceedings of the Royal Society of London B: Biological Sciences, 179 (1055): 125–138.

Mackaplow, M. B., Shaqfeh, E. S. G., and Schiek, R. L. (1994). A numerical study of heat
and mass transport in fibre suspensions. Proceedings of the Royal Society of London A:
Mathematical, Physical and Engineering Sciences, 447 (1929): 77–110.

Mackaplow, M. B. and Shaqfeh, E. S. G. (1996). A numerical study of the rheological

properties of suspensions of rigid, non-Brownian fibres. Journal of Fluid Mechanics, 329:
155–186.
Chapter 11. Introduction to Slender-Body Theory 185

Mackaplow, M. B. and Shaqfeh, E. S. G. (1998). A numerical study of the sedimentation

of fibre suspensions. Journal of Fluid Mechanics, 376: 149–182.

Minkowycz, W., Sparrow, E. M., and Murthy, J. (2006). Handbook of numerical heat transfer.
John Wiley & Sons, London, UK, 2nd edition.

Munk, M. M. (1924). The aerodynamic forces on airship hulls. National Advisory Committee
for Aeronautics, Washington DC.

Myers, G. E. (1971). Analytical methods in conduction heat transfer. McGraw-Hill Book

Company, New York.

Newman, J. N. (1964). A slender-body theory for ship oscillations in waves. Journal of

Fluid Mechanics, 18 (04): 602–618.

Newman, J. N. (1970). Applications of slender-body theory in ship hydrodynamics. An-

nual Review of Fluid Mechanics, 2 (1): 67–94.

Ozisik, M. N. (1993). Heat conduction. John Wiley & Sons, London, UK.

Patankar, S. (1980). Numerical heat transfer and fluid flow. CRC Press.

Rocha, A. and Acrivos, A. (1973a). On the effective thermal conductivity of dilute disper-

sions: General theory for inclusions of arbitrary shap. The Quarterly Journal of Mechanics
and Applied Mathematics, 26 (2): 217–233.

Rocha, A. and Acrivos, A. (1973b). On the effective thermal conductivity of dilute disper-
sions: highly conducting inclusions of arbitrary shape. The Quarterly Journal of Mechan-
ics and Applied Mathematics, 26 (4): 441–455.

Shaqfeh, E. S. (1988). A nonlocal theory for the heat transport in composites containing
highly conducting fibrous inclusions. Physics of Fluids (1958-1988), 31 (9): 2405–2425.
Chapter 11. Introduction to Slender-Body Theory 186

von Kármán, T. (1927). Berechnung der Druckverteilung an Luftschiffkörpern. Abhand-

lungen aus dem Aerodynamischen Institut an der Technischen Hochschule Aachen, (6): 3–17.

Ye, Z., Hoskinson, E., Dewey, R. K., Ding, L., and Farmer, D. M. (1997). A method for
acoustic scattering by slender bodies. I. Theory and verification. The Journal of the Acous-

tical Society of America, 102 (4): 1964–1976.

 CHAPTER 12
SLENDER-BODY THEORY FOR TRANSIENT HEAT CONDUCTION:
THEORETICAL DERIVATION

12.1 Slender Body Geometry, Constraints and Heat Transfer Problem

Figure 12.1 – Arbitrary slender body with center-line defined by F(s), local radius by R(s) and local
radius of curvature by RC (s) with s the arc-length along the center-line. The slender body has a sur-
face temperature T (s, t) with t the time, and exchanges heat Q(s, t) with the surrounding conductive
medium. A is a general point on the slender body center-line whose coordinates are provided by F(sA ).
B is an arbitrary point in the medium with coordinates b. The distance vector between the points A
and B is represented by d.

Figure 12.1 is a sketch of the type of slender-body transient heat conduction problem
studied in this work. The geometry of the slender body is set by the functions F(s) and

R(s), which define, as a function of the arc-length s along the center-line, the location of

the center-line and the radius in parametric form in the global (xyz)-coordinate system.

A local (x′ y′ z′ )-coordinate system is also defined with z′ parallel to s. The total length of
the slender body is Lt and its radius of curvature is represented by the function RC (s).
The slender body is situated in an infinite medium with constant and uniform thermal
diffusivity α and thermal conductivity k. For slender-body theory (SBT) to be applicable,
the pipe radius must be small compared with its length, radius of curvature, and the

187
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 188

thermal diffusion length:

R(s)
<< 1 , (12.1)
Lt
R(s)
<< 1 , (12.2)
RC (s)
R(s)
√ << 1 (12.3)
ατ

with τ a representative time-scale. Further, the pipe radius must be small compared to
the distance to a nearby slender body. In Section 12.4, a finite medium will be considered
with then the additional constraint that the pipe radius is small compared to the distance

from boundaries. Also, in Section 12.3, the model will be expanded for handling thermal
diffusion lengths on the order of the pipe radius or smaller and hereby lifting the con-

straint posed by Equation (12.3). The governing equation for transient heat conduction in

the medium is:

1 ∂T m
− ∇2 T m = 0 (12.4)
α ∂t
with T m (x, y, z, t) the temperature in the medium and t the time. It is assumed that the

initial medium temperature is zero and the body is in perfect thermal contact with the

medium.

We aim to find a relation between the temperature T at the surface of the slender body
and the heat exchange Q between the slender body and the conductive medium. For

the constraints given above, T and Q will show no angular dependence in the matching

region to a first approximation, since higher order multipoles will decay away, and there-

fore will only be a function of s and t. The linear relationship sought between T and Q
can now be written in its most general form as:
∫ t∫ Lt
T (s, t) = f (s∗ , t∗ , s, t)Q(s∗ , t∗ )ds∗ dt∗ (12.5)
0 0

implying that the temperature at a specific location and time depends on all past and
current heat exchange along the entire body. In simple problems, either T or Q is set
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 189

and the other one is calculated for using Equation (12.5). In more complex problems with
heat transfer processes occurring inside the slender body, e.g. internal advective heat flow,
Equation (12.5) must be combined with other equations, as discussed in Section 12.4.

12.2 SBT Derivation for Transient Heat Conduction

The idea of SBT is to develop an inner and outer solution for the problem, which are valid
nearby and faraway from the body, respectively, and then impose their compatibility in

the matching region. For the inner solution in our case, the temperature disturbance lo-

cally is assumed independent from nearby sections, i.e. the body is replaced with an

infinite cylinder with no axial or angular heat variation and the temperature calculation

simplifies to a two-dimensional axisymmetric heat transfer problem. For the outer solu-

tion, the slender body is replaced with a distribution of point sources along its center-line,

and the medium temperature calculation becomes a convolution over s and t of this heat

distribution with the Green’s function. This approach is equivalent to SBT for Stokes

Flow, where in the inner region, the local velocity has no axial dependence, whereas in

the outer region, the slender body can be replaced to a first approximation with a distri-

bution of Stokeslets (Batchelor, 1970).

12.2.1 Inner Solution

To estimate the temperature of the conductive medium in close proximity to the slender
body, i.e. r′ → R in the local coordinate system in Figure 12.1, the slender body is locally
simplified as an infinitely long cylinder undergoing steady-state heat transfer. The cylin-
der has no angular or axial variation in the heat exchange, which is justified provided the
constraints in Equations (12.1) to (12.3). Hence, the problem in the inner region simpli-
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 190

fies to a two-dimensional axisymmetric steady-state heat conduction problem in radial

coordinates with the temperature disturbed by a heat source located at the finite circular
boundary. Higher-order multipoles due to curvature or asymmetric heat processes in-
side the body, e.g. with borehole heat exchangers, decay away quickly and are therefore
neglected. At an arbitrary point A at the slender body (see Figure 12.1) at a certain time
t, the slender body is at temperature T (s A , t) and exchanges heat Q(s A , t). The medium
temperature in close proximity to A, defined as the inner solution T in , is estimated in the
local (r′ z′ )-coordinate system as (Carslaw and Jaeger, 1959):
( ′ )
′ Q(sA , t) r
T in (sA , r , t) = T (sA , t) − ln (12.6)
2πk R(sA )

with k the medium thermal conductivity, ln(x) the natural logarithm, and R(s A ) the local

radius of the slender body. Equation (12.6) cannot be viewed as a solution of the inner re-

gion problem independent from the outer solution because we cannot specify both T and

Q. Instead, we need to match this equation to an outer solution to obtain the relationship

between T and Q.

12.2.2 Outer Solution

To estimate the medium temperature at time t at an arbitrary point B far away from the
slender body, defined as the outer solution T out , the slender body is replaced by a distri-

bution of transient point sources with strength Q(s, t) along its center-line. Using Green’s
functions, T out is then calculated at B in the global (xyz)-coordinate system as (Carslaw
and Jaeger, 1959):
∫ t∫ Lt
T out (b, t) = G(b − F(s), t − t∗ )Q(s, t∗ )dsdt∗ (12.7)
0 0

with b − F(s) the distance vector between B and every point along the center-line. Equa-
tion (12.7) represents the convolution of the heat distribution Q(s, t) with the point source
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 191

solution G(d, τ) (Green’s function) calculated as:

( )
1 |d|2
G(d, τ) = exp − (12.8)
ρc p (4πατ)(3/2) 4ατ

with ρ, c p , and α the density, specific heat capacity and thermal diffusivity of the conduc-
tive medium, respectively, and |d| the length of the distance vector d.

When matching, the outer solution will be evaluated for the general point B approach-
ing asymptotically the center-line of the body at an arbitrary location, e.g. point A at

s = sA . However, with the double integral in its current form, a singularity arises due to
dividing by zero at point A for t∗ = t. The resolution of this singular lies in locally, around
A, simplifying the slender-body as a straight pipe (for which an analytical solution exists)

and adding and subtracting the resulting straight line source with constant heat exchange

Q(sA , t) to the right-hand side (RHS) of Equation (12.7), an approach comparable with SBT

for steady-state heat transfer (Shaqfeh, 1988). The starting time of the line source, t0 , can

be any time between 0 and t but the end time must be t. The straight line has a length 2l

much smaller than the local radius of curvature (2l << RC (s A )) due to geometry reasons,

but much larger than the local radius (2l >> R(sA )), which will be beneficial for the short

time-scale model extension (Section 12.3). Equation (12.7) becomes:

∫ t ∫ sA +l
T out (b, t) = G(b − F(s), t − t∗ )dt∗ ds Q(sA , t)
|t0 sA −l {z }
K
∫ t ∫ sA +l
+ G(b − F(s), t − t∗ ) [Q(s, t∗ ) − H(t∗ − t0 )Q(sA , t)] dt∗ ds (12.9)
|0 sA −l {z }
M
∫ t ∫ Lt
+ G(b − F(s), t − t∗ )Q(s, t∗ )dt∗ ds
0 0
s<(sA −l,sA +l)

with H the Heaviside step function. In Equation (12.9), the singularity has been removed
because, at point A for t∗ = t, Q(s, t∗ ) = Q(s A , t) and therefore the argument in the double
integral in term M is 0. Also, keeping in mind that the slender body is locally represented
as a straight line section, term K can be calculated analytically which will produce an
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 192

ln(r′ ) term that can be matched with the ln(r′ ) term of the inner solution. In preparation
for matching, the location of point B is expressed in the local (r′ z′ )-coordinate system
attached to point A. Term K is then calculated as:
∫ t ∫ l exp ( −r′2 −(z′ −s)2 )
4α(t−t∗ )
K(r′ , z′ ) = dt∗ ds . (12.10)
t0 −l ρc p (4πα(t − t ∗ ))(3/2)

When matching, point B asymptotically approaches point A. Term K will be evaluated

for z′ = 0 and can therefore be rewritten as:
∫ t ( ′2 ) ∫ l ( )
−r
exp 4α(t−t ∗) −s2
′
K(r , 0) = exp dsdt∗
t0 ρc p (4πα(t − t ∗ ))(3/2)
−l 4α(t − t ∗)

∫ t exp ( −r′2 ) ( )
4α(t−t∗ ) l
= erf √ dt∗
t0 4πk(t − t ∗)
4α(t − t ∗)
 ( ) ∫ ∞ ( ) ( ) 
1  r′2 1 r′2 1√ 
= E1 − exp −ϕ erfc ϕ dϕ  (12.11)
4πk 4α(t − t0 ) l2 ϕ
α(t−t0 )
4l 2 2
(∫ ∞ exp(−u) )
l2
with ϕ = α(t−t 0)
, E 1 (x) the exponential integral function x u
du , erf(x) the error
( ∫x )
function √2π 0 exp(−u2 )du , and erfc(x) the complimentary error function (1 − erf(x))
(Abramowitz and Stegun, 1964).

12.2.3 Matching

We now match the inner and outer solution, after adding and subtracting Q(sA , t∗ ) in term

M of Equation (12.9) to split M into two parts, and with further taking into account that
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 193

E1 (x2 ) ≈ −γ − 2ln(x) for small and positive x with γ Euler’s constant (0.5772...):

lim T in = lim T out (12.12)

r′ →∞ ′r →0

⇕
 ( ) ∫ ∞ ( ) 
 γ 1 R(s ) 1 1 1 √ 
T (sA , t) = − ϕ dϕ Q(sA , t)
A
− ln √ − erfc
4πk 2πk 4α(t − t0 ) 4πk α(t−t l 2 ϕ 2
∫ t ∫ sA +l ( ) 0)

exp −(s−s
2
A)
4α(t−t∗ )
+ [Q(sA , t∗ ) − H(t∗ − t0 )Q(sA , t)] dt∗ ds (12.13)
sA −l ρc p (4πα(t − t ))
∗ (3/2)
0
∫ t ∫ sA +l ( 2)
exp −(s−s A)
∗
[Q(s, t∗ ) − Q(sA , t∗ )] dt∗ ds
4α(t−t )
+
0 sA −l ρc p (4πα(t − t ∗ ))(3/2)
∫ t ∫ Lt
+ G(F(sA ) − F(s), t − t∗ )Q(s, t∗ )dt∗ ds .
0 0
s<(sA −l,sA +l)

Equation (12.13) provides the relationship, posed by Equation (12.5), between the slender

body temperature and heat exchange. It consists of four terms: the first term is the tem-

perature as if the slender body locally is a finite line source (FLS) with constant strength

Q(sA , t) during the time period t0 to t and over the total length 2l. Careful analysis reveals

this term consists of two parts. The first part, in the form −γ − 2ln(x), is a simplification

of E1 (x2 ) for small x and represents the temperature change due to an infinite line source

(Carslaw and Jaeger, 1959). The second part, the integral term, accounts for the finite
length (2l) by subtracting the temperature change due to two semi-infinite line sources,

one on either end. The second and third terms in Equation (12.13) can be viewed as cor-

rections to the first term by now accounting for the local heat source variation in space and
time. The second term is the temperature contribution from the same finite line source
but now with a heat pulse strength that is still constant along the line source but varies
over time, Q(sA , t∗ ), subtracted by Q(sA , t) for the time period t0 to t. In the third term, the
line source has the actual heat source Q(s, t∗ ) varying over space and time, subtracted by
Q(sA , t∗ ). Finally, the fourth term in Equation (12.13) is the temperature contribution from
all other previous and current point sources along the center-line of the slender body.
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 194

12.3 Extension of Model to Short Time-Scales

Equation (12.13) is only valid for heat transfer processes occurring over long time-scales,
√
i.e. (R(s) << ατ) (see Equation (12.3)). For short time-scale phenomena, with the heat

transfer diffusion length on the order of the pipe radius or smaller, Equation (12.13)
should be adjusted to represent a cylindrical source rather than a line source. The temper-
ature in the finite cylindrical source (FCS) model, T FCS , with radius R and constant heat
pulse Q from t0 to t is given by (Carslaw and Jaeger, 1959):
 ( ) ( ) 
 2 ∫ ∞ 1 − exp −ϕ2 α(t−t 0)
1
∫ ∞
1 1 √ 
T FCS (t) =  3
2
R
dϕ − erfc ϕ dϕ  Q (12.14)
π k 0 ϕ3 (J12 (ϕ) + Y12 (ϕ)) 4πk α(t−t
l2
)
ϕ 2
0

with J1 (x) and Y1 (x) the first order Bessel function of the first kind and second kind, re-

spectively. Equation (12.14) readily replaces the FLS model in the first term of Equa-

tion (12.13). The effect of Q(s A , t) changing over time, represented by the second term in

Equation (12.13), should be calculated with the FCS model as well. In continuous form,

this can only be written analytically in the Laplace domain. However, a solution can be

found in the time domain when discretizing Q as a series of n constant heat pulses (as will

be applied when numerically implementing),

∑[ ( )]
Q(sA , t) = Qn (sA ) H(t − tn−1 ) − H(t − tn ) (12.15)
n

with Qn (s A ) the constant heat pulse at time step n. No FCS model is available when
Q(s, t) varies spatially along the cylindrical source, represented by the third term in Equa-

tion (12.13). However, when considering 2l to be the length of a computational element

on which Q is constant (as will be applied in Chapter 13), allowed in the case of spatially-
smooth variations in Q along the center-line of the body, this term can be neglected. Fi-
nally, the fourth term remains unaltered, since it accounts for temperature contributions

caused by heat sources at a distance larger than R(s A ) away for the point of interest (A).
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 195

All together, the solution for T (sA , t) applicable at short time-scales becomes:
 ( ) ( ) 
 2 ∫ ∞ 1 − exp −ϕ2 α(t−t 0)
1
∫ ∞
1 1 √ 
T (sA , t) =  3
2
R(sA )
dϕ − erfc ϕ dϕ  Q(sA , t)
π k 0 ϕ3 (J1 (ϕ) + Y1 (ϕ))
2 2 4πk α(t−t l2 ϕ 2
[ ∫ ( α(t−t
) ( 0)
α(t−t
)
∑ 2 ∞ exp −ϕ 2 n
R(sA )2
)
− exp −ϕ 2 n−1 )
R(sA )2
+ dϕ (12.16)
n
πk 0
3
ϕ (J1 (ϕ) + Y1 (ϕ))
3 2 2

∫ α(t−t
l2 ( ) ]
1 n) 1 1√ ( )
− erfc ϕ dϕ Qn (sA ) − H(tn − t0 )Q(sA , t)
4πk α(t−tl ) ϕ
2 2
∫ t ∫ Lt
n−1

+ G(F(sA ) − F(s), t − t∗ )Q(s, t∗ )dt∗ ds

0 0
s<(sA −l,sA +l)

with the assumption that t0 now corresponds to one of the time steps. Equation (12.16)

is also valid for long time-scales, however, from a computational point of view, using
Equation (12.13) in calculations is preferred, as will be further discussed in Chapter 13.

12.4 Other Model Extensions

The equations developed so far assume an infinite, conductive medium with uniform

zero initial temperature, and constant, isotropic and uniform thermal properties. This

section discusses how to remove several of these constraints, as well as how to incorporate

heat transfer processes inside the slender body.

The effect of a far-field or initial temperature distribution can be accounted for by

modeling it as an imposed field T i (x, y, z, t) without the slender body and then superim-
posing this field to the disturbance field caused by the slender body, i.e. adding T i (x, y, z, t)
to the RHS of Equation (12.9) and therefore to the RHS of Equation (12.13) or (12.16). In
the most simple cases, the imposed field is found directly such as for a uniform non-
zero initial temperature, T i (x, y, z, t) = T 0 , or for a linear far-field temperature profile,

T i (x, y, z, t) = T 0 (z) = T sur f ace + ωz (with ω the geothermal gradient).

Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 196

For a non-infinite medium, the infinite medium model can still be an accurate repre-
sentation in cases when the distance from the slender body to the medium boundary is
much larger than the thermal diffusion length. An example is the simulation of the heat
transmission with deep geothermal, oil or gas wells. Otherwise, a different (and more
onerous) Green’s function must be used with a complete list of Green’s functions for all
types of boundary conditions provided by Beck et al. (1992). Some can be readily found
by using method of images, such as constant temperature or zero heat flux condition at
a planar boundary (Beck et al., 1992; Duffy, 2015). Several of the solution methods pro-

vided by Beck et al. can be derived using the imposed field technique, introduced above
for dealing with the initial condition. For example, a distribution of heat sources at a

boundary must be convoluted over space and time with the Green’s function for an infi-

nite medium, and then superimposed to the disturbance field caused by the slender body,

calculated with SBT using a zero heat flux Green’s function.

Simple cases of a medium involving fluid flow, or anisotropic or non-uniform thermal

properties can be simulated by using a different (and again more complicated) Green’s

function. One-dimensional constant velocity fluid flow causing heat advection, instead

of pure heat conduction, can be accounted for by using the moving point source Green’s

function (Carslaw and Jaeger, 1959; Molina-Giraldo et al., 2011). However, this is only di-
rectly applicable for the long time-scale solution, and should be combined with a different
model, e.g. the solid cylindrical source model (Zhang et al., 2013), for the short time-scale

solution. Also a Green’s function exists in the case of anisotropic thermal properties, for
which the thermal conductivity can now be represented by a tensor (Yan-Po, 1977). Fur-
ther, for a two-layered medium with isotropic properties, an exact Green’s function is
provided by Carslaw and Jaeger (1959). For any number of layers, Abdelaziz et al. (2014)
provide an approximate method using interlayer heat exchange adjustments, and laws
of composite materials for estimating representative thermal properties. Again, the ex-
tensions for anisotropic and non-uniform thermal properties are only directly applicable
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 197

at long time-scales when the cylindrical source models can be represented by line source
models. Incorporating temperature-dependent thermal properties with the SBT model is
not possible since this would lead to non-linear equations.

In simple cases, either T or Q at the surface of the slender body is specified with the

other being solved for. In more complex problems with heat processes occurring inside
the body, additional relations are required and calculated for simultaneously with Equa-
tion (12.13) or (12.16). For example, for advective heat transfer due to pipe fluid flow, a
first additional relation is (Bergman et al., 2011):

T f (s, t) − T (s, t)
Q(s, t) = (12.17)
Rt (s)

with T f (s, t) the temperature of the fluid and Rt (s) the effective thermal resistance of the

pipe calculated as the sum of the thermal resistances of the conductive heat transfer

through the pipe and the convective heat transfer between the fluid and the pipe wall

(Bergman et al., 2011):

ln(R(s)/Rinner (s)) 1
Rt (s) = + (12.18)
2πk p (s) 2πh(s)Rinner (s)

with R(s) and Rinner (s) the outer and inner radius of the pipe, respectively. Further, k p (s)

is the thermal conductivity of the pipe material and h(s) is the convective heat transfer

coefficient which can be calculated with appropriate Nusselt-correlations for forced inter-

nal convection (Bergman et al., 2011). A third equation is required, which represents the

energy balance of the fluid inside the pipe (with neglecting the heat conduction within
the fluid):
∂T f (s, t) ∂T f (s, t) Q(s, t)
ρ f c p, f + ρ f c p, f U(s) = (12.19)
∂t ∂s πR2inner (s)
with ρ f , c p, f and U(s) the density, specific heat capacity and velocity of the fluid, respec-
tively, and the fluid inlet temperature being a boundary condition. Various finite dif-
ference methods are available to handle the time and spatial derivative, with different
levels of accuracy and complexity (Minkowycz et al., 2006). For the second case-study for
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 198

example (Section 14.2), Equation (12.19) is numerically implemented using an upwind

differencing scheme. For other types of heat exchangers such as single-U (1-U), double-U
(2-U) and coaxial borehole heat exchangers, similar equations can be developed.

Finally, one can derive from Equation (12.13) the SBT solution for steady-state heat

conduction by replacing Q(sA , t) with Q(sA ), and letting time approach infinity. The time
convolution integral of the transient Green’s function (Equation (12.8)) for any starting
time t0 translates into the steady-state Green’s function (Beck et al., 1992):
( )
∫ t exp − |d| ∗
2

4α(t−t ) 1
lim dt∗ = . (12.20)
t→∞ t ρc p (4πα(t − t∗ ))(3/2) 4πk |d|
0

The SBT solution for steady-state heat transfer with curved slender bodies in a conductive

medium is now found as:

( ) ∫ sA +l ∫ Lt
1 2l Q(s) − Q(sA ) Q(s)
T (sA ) = ln Q(sA ) + ds + ds (12.21)
2πk R(sA ) sA −l 4πk |s − sA | 0 4πk |F(sA ) − F(s)|
s<(sA −l,sA +l)

with T (sA ) the surface temperature of the slender body at any point sA along its center-line.

The first two terms in Equation (12.21) represent the temperature due to a local straight

line source with length 2l. The third term in Equation (12.21) accounts for the line dis-

tribution along each of the other cylindrical elements making up the curved body. The
second term in Equation (12.13) is not present in Equation (12.21) because the local heat

source no longer varies with time. Equation (12.21) provides a description of the interac-

tion through steady-state heat conduction among a collection of cylindrical elements that
approximate a curved slender body that is equivalent to the treatment in Equation (2.18)
of Mackaplow et al. (1994) of steady-state heat conduction in a composite with many
discrete straight highly-conductive fiber inclusions.
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 199

References

Abdelaziz, S. L., Ozudogru, T. Y., Olgun, C. G., and Martin, J. R. (2014). Multilayer finite

line source model for vertical heat exchangers. Geothermics, 51: 406–416.

Abramowitz, M. and Stegun, I. A. (1964). Handbook of mathematical functions: with formulas,

graphs, and mathematical tables. Number 55. Courier Corporation, Mineola, New York.

Batchelor, G. K. (1970). Slender-body theory for particles of arbitrary cross-section in

stokes flow. Journal of Fluid Mechanics, 44 (03): 419–440.

Beck, J. V., Cole, K. D., Haji-Sheikh, A., and Litkouhi, B. (1992). Heat conduction using

Green’s functions. Hemisphere Publishing Corporation, Washington DC.

Bergman, T. L., Lavine, A. S., Incropera, F. P., and Dewitt, D. P. (2011). Introduction to heat

transfer. John Wiley & Sons, 6th edition.

Carslaw, H. S. and Jaeger, J. C. (1959). Conduction of heat in solids. Clarendon Press, Oxford.

Duffy, D. G. (2015). Greens functions with applications. CRC Press, Boca Raton, Florida.

Mackaplow, M. B., Shaqfeh, E. S. G., and Schiek, R. L. (1994). A numerical study of heat

and mass transport in fibre suspensions. Proceedings of the Royal Society of London A:

Mathematical, Physical and Engineering Sciences, 447 (1929): 77–110.

Minkowycz, W., Sparrow, E. M., and Murthy, J. (2006). Handbook of numerical heat transfer.

John Wiley & Sons, London, UK, 2nd edition.

Molina-Giraldo, N., Blum, P., Zhu, K., Bayer, P., and Fang, Z. (2011). A moving finite
line source model to simulate borehole heat exchangers with groundwater advection.
International Journal of Thermal Sciences, 50 (12): 2506–2513.

Shaqfeh, E. S. (1988). A nonlocal theory for the heat transport in composites containing
highly conducting fibrous inclusions. Physics of Fluids (1958-1988), 31 (9): 2405–2425.
Chapter 12. Slender-Body Theory for Transient Heat Conduction:
Theoretical Derivation 200

Yan-Po, C. (1977). Analytical solution for heat conduction in anisotropic media in infinite,
semi-infinite, and two-plane-bounded regions. International Journal of Heat and Mass
Transfer, 20 (10): 1019–1028.

Zhang, W., Yang, H., Lu, L., and Fang, Z. (2013). The analysis on solid cylindrical heat
source model of foundation pile ground heat exchangers with groundwater flow. En-
ergy, 55: 417–425.
 CHAPTER 13
SLENDER-BODY THEORY FOR TRANSIENT HEAT CONDUCTION:
NUMERICAL FRAMEWORK

In Chapter 13, we develop one possible numerical framework for the slender-body theory
(SBT) model for transient heat conduction derived in the previous chapter. Figure 13.1
shows with a simplified diagram the 6 steps taken in developing this framework. The
slender body heat transfer problem is discretized in the space (step 1) and time domain

(step 2), which converts the integrals in the SBT solution (Equation (12.13) or (12.16))
into summations. The most cumbersome calculation is the double summation over all

previous heat pulses emitted by all neighboring slender body elements. Therefore, sim-

plifications are applied in the space (step 3) and time domain (step 4) when calculating

these heat pulse contributions. The computational time can further be decreased by com-

bining point sources in the space (step 5) as well as in the time domain (step 6), a method

also known as the Fast Multipole Method or FMM. The following sections explain each

step in more detail and develop the SBT numerical equations, which are implemented in

a computer algorithm.

13.1 Space and Time Discretization of SBT Model

A computer implementation of the SBT model requires to numerically solve Equa-

tion (12.13) or (12.16) for each location s and time t. The integrals in these equations
can be converted into summations by discretizing the slender body into a set of straight

pipe elements i and discretizing the time into time steps m (steps 1 and 2 in Figure 13.1).
This converts the continuous functions Q(s, t) and T (s, t) into the discrete functions Qi,m
and T i,m , respectively. Qi,m represents the constant heat exchange (in units of W/m) for the

201
Chapter 13. Slender-Body Theory for Transient Heat Conduction:
Numerical Framework 202

Figure 13.1 – Diagram illustrating the 6 steps in developing the SBT numerical framework. A slender
body is discretized into straight pipe elements (step 1). The heat exchange at each element is discretized
in time as a series of heat pulses (step 2). In many cases, without much loss in accuracy but with
less computational complexity, pipe elements can be simplified as infinite or finite line sources, point
sources or even neglected both in space (step 3) and time domain (step 4). To further decrease the
computational time, point sources can be combined in space (step 5) as well as in time domain (step 6).
Chapter 13. Slender-Body Theory for Transient Heat Conduction:
Numerical Framework 203

entire pipe element i during time step m; T i,m represents the temperature at the mid-point
of pipe element i at the end of time step m.

The local geometry of the slender body sets a maximum length for the pipe elements
by requiring that the set of elements still accurately represents the shape of the body.

In numerical tests with the SBT algorithm, we used sample calculations with a heated
toroidal ring to show that a maximum element length to radius of curvature ratio of about
0.7 results in at least 98% accuracy. On the other hand, there is no minimum length and
the user can discretize the body as fine as deemed necessary to achieve the desired accu-

racy with element lengths on the order of the pipe radius or smaller possible. However,

to simplify the numerical implementation, we choose here to set the element lengths, la-
beled as Li for element i, equal to the length of the local constant line source (2l) of the

outer solution (see Section 12.2). Each element length Li is now larger than the local pipe

radius but not necessarily identical to the other element lengths, since l can vary along the

slender body axis. For spatially-smoothly varying Q along the center-line of the slender

body, this simplification is not believed to jeopardize the accuracy. Further, it automati-

cally sets the third term in Equation (12.13) to zero, a simplification already applied when

deriving Equation (12.16).

There are no model constraints on the time step length. The user can choose non-
uniform time steps as fine or coarse as required in obtaining the desired trade-off between

accuracy and computational time. To simplify the numerical implementation, we choose

to have the starting time t0 of the constant line source Q(sA , t) so that t − t0 corresponds to
each time step length.

For the space and time discretization schemes selected above, Equation (12.16) (or
(12.13)) is now converted into a linear equation in Q with T i,m the temperature at the
selected element i after current time step m and Q j,n the heat source emitted by element j
Chapter 13. Slender-Body Theory for Transient Heat Conduction:
Numerical Framework 204

during time step n:

∑∑
T i,m = ( fi, j,m,n ·Q j,n )
j n

= ( fi,i,m,m ·Qi,m ) (current heat pulse by selected element)

∑
+ ( fi,i,m,n ·Qi,n ) (previous heat pulses by selected element) (13.1)
n,m
∑
+ ( fi, j,m,m ·Q j,m ) (current heat pulse by neighboring elements)
j,i
∑∑
+ ( fi, j,m,n ·Q j,n ) (previous heat pulses by neighboring elements) .
j,i n,m

Terms 1 and 2 in Equation (12.13) and (12.16) correspond to the first and second term in

Equation (13.1) (current and previous heat pulses emitted by selected element ( j = i)).

Further, term 4 in Equation (12.13) or term 3 in Equation (12.16) is split into two parts in

Equation (13.1) (third and fourth term) by distinguishing between current (n = m) and

previous (n < m) heat pulses emitted by neighboring elements ( j , i). The terms fi, j,m,n

now represent standard, easy to calculate models, which are the topic of Section 13.2.

Other approaches exist to discretize or numerically solve Equation (12.13) or (12.16).

One could decouple 2l from the pipe element lengths to allow for very fine grids in order

to capture abrupt spatial changes in Q, occurring over length-scales on the order of the

pipe radius or smaller. Another approach is to represent Q(s, t) as a summation of polyno-

mials and numerically solve for the polynomial coefficients, as highlighted by Batchelor

(1970). However, these methods are not further investigated in this discussion.

13.2 Equations for fi, j,m,n

The term fi, j,m,n , introduced in Equation (13.1), can be interpreted as the temperature re-
sponse at the mid-point of pipe element i at time step m caused by a unit heat pulse of 1
Chapter 13. Slender-Body Theory for Transient Heat Conduction:
Numerical Framework 205

W/m emitted by pipe element j during time step n. From Equations (12.13) and (12.16),
it is derived that fi, j,m,n represents simple models such as cylindrical, line or point source
models, which can often be calculated analytically. To save on computational time, a
method is presented in Appendix D that uses decision trees and non-dimensional num-
bers, based on length- and time-scales of the heat transfer problem, to select the most
simple model that still accurately calculates each term fi, j,m,n . For example, heat sources
occurring far away in space and/or time (see steps 3 and 4 in Figure 13.1), could be repre-
sented by point sources or even neglected rather than calculated with the full cylindrical

or line source models. Appendix D provides all equations to calculate f for each term
in Equation (13.1) and explains in more detail the accuracy obtained by applying these

simplifications.

13.3 Fast Multipole Method, other Implementation Strategies, and

Computational Time

The computational time of the SBT algorithm can further be decreased by combining

point sources in the space and time domain (steps 5 and 6 in Figure 13.1). This method is

called the Fast Multipole Method (FMM) and was pioneered by Greengard and Rokhlin
(1987). Although advanced decision criteria could be developed to determine which point

sources to combine at which times, here a more rudimentary method is adopted. Previ-
ous pulses once represented by the most simple point source model (i.e. the PS3 model,
√
introduced in Appendix D) can be combined if TDR· FoL,PP,E , with TDR and FoL,PP,E non-
dimensional numbers defined in Table D.1 and appearing in the PS3 equation, is larger
than a certain value, e.g. 100, and correspondingly the f become relatively small (but
not negligible). A certain number of point sources, e.g. five, could then be grouped to-
gether and represented by a new point source with a Q·(tn − tn−1 ) equal to the sum of the
Chapter 13. Slender-Body Theory for Transient Heat Conduction:
Numerical Framework 206

Q·(tn − tn−1 ) of the original point sources (conservation of energy) and with the new tn and
tn−1 (end and start time of a heat pulse) set in this way that the new f exactly matches
the sum of the old f , e.g. at the time of combining the pulses. This simplification can be
done at multiple levels, i.e. five new point sources each replacing five old point sources
can later be combined themselves. Also, in the space domain, different pipe elements
located at different positions can have a similar spacing, e.g. within 5%, with respect to a
selected element. Once represented by a point source, the heat pulses from these different
elements can simply be added together and only one f needs to be calculated.

Other implementation strategies to decrease the computational time are to calculate

certain terms that often reoccur throughout the simulation once in advance. Examples
are (Li /4πk), several non-dimensional numbers, special functions such as erf(x) and E1 (x)

and the integrals in the cylindrical source model. Also, because each Equation (13.1) for

every pipe element is coupled with each other, the solution algorithm requires solving a

linear set of equations at each time step, for which either direct or iterative methods can be

used. Since the matrix system in some cases is sparse and/or symmetrical, fast solution

methods could be utilized, which are built-in into software packages such as MATLAB

(MathWorks, 2012).

An SBT computer program based on the numerical framework presented requires im-
plementing Equation (13.1), the non-dimensional numbers, heat source model equations

and decision trees from Appendix D to calculate fi, j,m,n , and depending on the problem
possibly equations from Section 12.4 (model extensions), e.g. to include heat transfer
phenomena inside the slender body or for specific boundary conditions. A simulation
requires solving Equation (13.1) at each time step simultaneously for each pipe element.
For SBT, the total number of elements NE is O(NL ) with NL the discretization of the slen-
der body. Without any simplifications (steps 3 to 6 of Figure 13.1 neglected), and with
utilizing a direct solver based on LU decomposition for solving the matrix system, the
Chapter 13. Slender-Body Theory for Transient Heat Conduction:
Numerical Framework 207

computational time is O(Nt ·NE3 + Nt2 ·NE2 ), with Nt the number of time steps. In contrast, for
FEM (finite element methods), with L a characteristic length for the medium domain, the
√
number of elements NE is O(NL ·(L/R)2 ) or O(NL ·( αt/R)2 ), whichever is smaller, but the
computational time is only O(NE3 + Nt ·NE2 ). The number of elements NE for SBT is typically
several orders of magnitude smaller than NE for FEM which results in lower computa-

tional times in many cases. However, the SBT model is no longer advantageous for more
spatially-complex but compact problems, especially in combination with a large number
of time steps. With the simplifications and strategies provided, the computational time

will decrease but it is difficult to determine by how much because it depends on several
factors such as decision tree limits, matrix solver algorithm, and spatial and time lengths

and discretizations. In terms of computer memory, the SBT model can also be superior

since an FEM simulation typically requires large amounts of data storage for meshing,

matrix assembly and solving.

References

Batchelor, G. K. (1970). Slender-body theory for particles of arbitrary cross-section in

stokes flow. Journal of Fluid Mechanics, 44 (03): 419–440.

Greengard, L. and Rokhlin, V. (1987). A fast algorithm for particle simulations. Journal of

computational physics, 73 (2): 325–348.

MathWorks (2012). MATLAB Release 2012a, The Mathworks, Inc., Natick, Massachusetts,
United States.
 CHAPTER 14
SLENDER-BODY THEORY FOR TRANSIENT HEAT CONDUCTION:
CASE-STUDIES

In this chapter, two case-studies are presented to validate and illustrate the applicabil-
ity, speed and versatility of the slender-body theory (SBT) numerical model developed in
Chapter 13. The first case-study is a theoretical example of two parallel cylinders in close
proximity and at constant temperature, representing probably the most fundamental heat

transfer problem with interacting slender bodies, for which no analytical solution exists.

The second case-study is an applied example of one loop of a slinky-coil heat exchanger,
typically used in geothermal heat pump systems that employ horizontal trenches for heat

exchange with the ground. In each case, the SBT numerical model, implemented in MAT-

LAB R2012a using all of the simplifications and strategies discussed, is compared with the

FEM (finite element methods) software package COMSOL Multiphysics 4.3 (COMSOL,

2012) in terms of computational time. In order to make a fair comparison, the discretiza-

tion schemes are chosen such that comparable accuracy is obtained. All simulations are

carried out on a standard Intel Core i3-2120 (3.3 GHz) desktop computer with 6 GB RAM

and Windows 7 as operating system. Both in MATLAB and COMSOL, a direct solver is

used to solve the linear system of equations.

14.1 Two Parallel Cylinders at Constant Temperature

Two parallel cylinders with length Lt = 10 m, radius R = 0.05 m and spacing S = 1 m are
placed in an infinite conductive medium with thermal conductivity k = 1 W/(m·K) and
thermal diffusivity α = 10−6 m2 /s (Figure 14.1). The medium has an initial temperature
of 0◦ C and both cylinders are kept at a constant temperature of 10◦ C. We aim to solve for
the transient heat exchange Q (W/m) at both short and long time-scales, occurring at the

208
Chapter 14. Slender-Body Theory for Transient Heat Conduction: Case-Studies 209

Figure 14.1 – Two case-studies to demonstrate the SBT numerical model developed in Chapter 13. The
first case-study (left) is the simulation of the heat flux Q at the mid-point and end-point of two parallel
identical cylinders at constant temperature and in close proximity. The second case-study (right) is the
simulation of the outlet temperature of one loop of a slinky-coil heat exchanger. Drawings are not to
scale.

mid-point (δm = 5 m) and close to the end-point (δe = 0.25 m) of the cylinders, averaged
over the circumference. Due to the symmetry of the problem, both cylinders will have the

same heat flux.

In the SBT model, each cylinder is discretized into 20 identical pipe elements for a
total number of 40 grid points, which is identical to modeling one cylinder and applying
the method of images to force a zero flux boundary condition at the mid-plane between
both cylinders. The time is discretized into 150 time steps from 0 to 109 s. The COMSOL
model is a 3-dimensional model of a quarter of one cylinder (axial cross section of the half
length), with zero heat flux condition at the appropriate boundaries, in order to make full
Chapter 14. Slender-Body Theory for Transient Heat Conduction: Case-Studies 210

use of the symmetry embedded in this problem, and constant temperature of 0◦ C at the
other (far-field) boundaries. This quarter of a cylinder is placed in a cuboid with height
of 35 m and width and depth of 30 m, at its correct location close to one of the corners.
The number of grid points along the cylindrical body is 500, the total number of mesh
elements is about 130,000 and the relative tolerance is set to 0.1. With these spatial and
time discretization schemes, the SBT and COMSOL model show a comparable accuracy
of about 99% on average over the entire time span, with reference to a highly accurate
COMSOL model with very fine mesh (1,000,000 cells) and very low relative tolerance

(10−6 ). This reference COMSOL model has been validated at short time-scales with the
analytical solution of a single, infinitely long cylinder at constant temperature (Carslaw

and Jaeger, 1959).

The result for Q at the mid-point and end-point as a function of time is shown in Fig-
√
ure 14.2 with the heat flux and time non-dimensionalized as Q/(k∆T ) and αt/S , respec-
tively, with ∆T the temperature difference between the cylindrical surface and the far-field

(∆T = 10◦ C). The non-dimensional time can be interpreted as the ratio of the diffusion
√
distance αt to the spacing between the cylinders S . The result for one cylinder only is
included to illustrate the tube-tube interactions which start to occur at a non-dimensional
time of about 0.5 corresponding to a thermal diffusion length of 0.5 m. The difference

in Q between the mid-point and end-point becomes evident at a non-dimensional time

of about 0.2, corresponding to a thermal diffusion length of 0.2 m. The computational

time for the SBT model is 7 s and for the COMSOL model about 10 min. This case-study
illustrates the high computational speed of the SBT model, and its versatility in easily
handling short and long time-scales and multiple, disconnected, interacting slender bod-
ies of finite length. Even higher speeds in this as well as in the next case-study can likely
be obtained when implementing the SBT model using a compiled language such as C
instead of an interpreted language as with MATLAB. However, this is not further inves-
tigated.
Chapter 14. Slender-Body Theory for Transient Heat Conduction: Case-Studies 211

Figure 14.2 – Results for the two case-studies presented in Figure 14.1. The result for the first case-
study
√ (left) is the non-dimensional heat exchange (Q/(k∆T )) as a function of the non-dimensional time
( αt/S ) at the mid-point and end-point of one and two parallel cylinders at constant temperature. The
result for the second case-study (right) is the fluid outlet temperature of one loop of a slinky-coil heat
exchanger.

14.2 Slinky-Coil Heat Exchanger

In the second case-study, the outlet temperature of water flowing inside one loop of a
slinky-coil heat exchanger as a function of time is simulated for (see Figure 14.1). The

heat exchanger has a total length Lt of 3.325 m, a loop radius RC of 0.37 m, a pipe radius R

of 1 cm, a pipe wall thickness T h p of 0.1 cm, and a pipe material thermal conductivity k p of
1 W/(m·K). At the cross-over point, the pipes are within 0.5 cm of each other. The density
ρ f , specific heat capacity c p, f , inlet temperature T inlet , and mean velocity U of the water are
1000 kg/m3 , 4200 J/(kg·K), 20◦ C, and 0.1 m/s, respectively. The convective heat transfer
coefficient between the water and the pipe is set at h = 1000 W/(m2 ·K). Further, the infinite
conductive medium has a thermal diffusivity α of 10−6 m2 /s, thermal conductivity k of 3
W/(m·K), and initial temperature of 0◦ C.
Chapter 14. Slender-Body Theory for Transient Heat Conduction: Case-Studies 212

For the SBT equations, the heat exchanger is discretized into 30 identical pipe ele-
ments and the time is discretized into 50 linear time steps from 1 to 50 s, followed by 100
logarithmically-spaced time steps from 50 s up to 107 s. In the COMSOL model, the heat
exchanger is placed at the center of a cube, representing the infinite medium, with edge
length of 10 m. The total number of elements is about 80,000 and the relative tolerance is
set to 0.02. Further, in both the SBT and COMSOL simulations, the conductive medium
model is coupled to a one-dimensional advective heat flow model representing the fluid
flow inside the slinky-coil heat exchanger (see Section 13.3) with a spatial discretization of

180 and 200 elements in the SBT and COMSOL model, respectively. The spatial and time
discretization schemes chosen for both models provide a comparable accuracy of about

99% on average over the entire time span, in comparison with a highly accurate COMSOL

model with very low relative tolerance of 10−6 and a fine mesh of about 350,000 cells.

The water outlet temperature simulated with the SBT and COMSOL model is shown

in Figure 14.2. The calculation time for the SBT model is 6 s while for the COMSOL model

about 19 min. This case-study illustrates the SBT model running at high computational

speeds and easily handling short and long time-scales, curved bodies, and heat transfer

inside a slender body. Moreover, the slender body approaching itself at the cross-over

point, which is a violation of one of the constraints provided in Section 12.1, does not
appear to have a noticeable impact on the outlet temperature. The reason is that the two
parts only lie within the inner region of one another for a short distance O(R), and since

Lt /R >> 1, the heat transfer from this region is small O(R/Lt ).

Slinky-coil heat exchangers are typically modeled using finite element modeling
(FEM) numerical software (Wu et al., 2010; Congedo et al., 2012; Fujii et al., 2012; Chong
et al., 2013) or g-functions (Xiong et al., 2015). This case-study shows SBT provides an al-
ternative, computationally-fast approach. Other types of heat exchangers for geothermal
or ground-source heat pump applications, e.g. parallel horizontal pipes, and single-U
Chapter 14. Slender-Body Theory for Transient Heat Conduction: Case-Studies 213

(1-U) and double-U (2-U) borehole heat exchangers, can be simulated using SBT as well.
Potential oversimplifications often applied in other models for simulating these heat ex-
changers, such as constant longitudinal heat exchange Q along a single-U borehole heat
exchanger, can then be avoided.

References

Carslaw, H. S. and Jaeger, J. C. (1959). Conduction of heat in solids. Clarendon Press, Oxford.

Chong, C. S. A., Gan, G., Verhoef, A., Garcia, R. G., and Vidale, P. L. (2013). Simulation of

thermal performance of horizontal slinky-loop heat exchangers for ground source heat

pumps. Applied Energy, 104: 603–610.

COMSOL (2012). COMSOL Multiphysics 4.3, COMSOL Inc., Burlington, Massachusetts,

United States.

Congedo, P., Colangelo, G., and Starace, G. (2012). CFD simulations of horizontal ground

heat exchangers: a comparison among different configurations. Applied Thermal Engi-

neering, 33: 24–32.

Fujii, H., Nishi, K., Komaniwa, Y., and Chou, N. (2012). Numerical modeling of slinky-coil
horizontal ground heat exchangers. Geothermics, 41: 55–62.

Wu, Y., Gan, G., Verhoef, A., Vidale, P. L., and Gonzalez, R. G. (2010). Experimental
measurement and numerical simulation of horizontal-coupled slinky ground source
heat exchangers. Applied Thermal Engineering, 30 (16): 2574–2583.

Xiong, Z., Fisher, D. E., and Spitler, J. D. (2015). Development and validation of a SlinkyTM
ground heat exchanger model. Applied Energy, 141: 57–69.
 CHAPTER 15
SLENDER-BODY THEORY FOR TRANSIENT HEAT CONDUCTION:
CONCLUSIONS

The slender-body theory (SBT) was initially developed and is most often used in the field
of fluid dynamics to simplify the flow and pressure field solution of a slender body prob-
lem using matched asymptotic expansions of an inner and outer solution. In this work,
the SBT was developed for slender, curved bodies with circular cross section undergoing
transient heat transfer in a conductive medium. The theoretical derivation is based on

matching the temperature field of an infinitely long cylindrical source as inner solution

with the temperature field of a finite line source as outer solution. By replacing the line

source model with the cylindrical source model in the final result, a solution was obtained

applicable at all time-scales. One possible numerical model was developed by discretiz-

ing the SBT theoretical result in the space and time domain. High computational speeds

were obtained by applying simplifications such as modeling cylindrical and line sources

as point sources, combining point sources with the Fast Multipole Method, or even ne-

glecting point sources when allowed. Two case-studies were included to demonstrate the

SBT model operating at high computational speeds, capturing tube-tube interactions and

finite lengths, and handling curved bodies, short and long time-scales, and heat transfer

processes inside a slender body.

The slender-body theory for transient heat conduction can be an attractive solution
technique, especially in the case of complex and non-compact geometries, when no an-
alytical solutions are available and finite element methods require long computational
times and large amount of computer memory. Examples of applications are simulat-
ing the heat transmission in deviated oil, gas or geothermal wells, designing geother-
mal reservoirs for underground energy storage or heat pump applications, and modeling
heating or cooling elements in industrial processes.

214
 CHAPTER 16
OVERALL CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK

16.1 Overall Conclusions

Techno-economic simulations with the newly-developed computer model GEOPHIRES

show that low-grade (∼30◦ C/km) deep geothermal resources are attractive for direct-use
heat and cogeneration but not necessarily for solely electricity production. Hybridizing
the system, e.g. with biomass, can increase the performance. Utilizing medium-grade

(∼50◦ C/km) and high-grade (∼70◦ C/km) resources for solely electricity production is cur-

rently already fairly attractive but expected to be even more so in the near future with

anticipated improvements in reservoir productivity and thermal drawdown rate.

A techno-economic analysis of hybrid geothermal heat pump systems for cooling-

dominated applications, with particular focus on cooling of cellular tower shelters, was

conducted. A full-scale and fully equipped hybrid geothermal heat pump system pro-

viding cooling to a Verizon Wireless cellular tower shelter in Varna, NY was successfully

designed, constructed, and tested, and has been reliably and efficiently operated for two

years. A TRNSYS model was developed, validated, and applied to various system con-
figurations to evaluate performance under transient conditions. The conditions analyzed

are those of Upstate New York weather, and shelter, cooling, and financial parameters
similar to those at the Varna Site. Without considering any subsidies or tax incentives, an
air-source heat pump combined with air-economizer is the most cost-effective option fol-
lowed by a ground-source heat pump hybridized with air-economizer. In terms of energy
consumption and CO2 emissions, hybrid geothermal heat pump systems perform best, il-
lustrating the advantages of hybridizing to counteract the reservoir thermal imbalance
due to the cooling-dominated nature of this application. Air-source heat pump systems

215
Chapter 16. Overall Conclusions and Recommendations for Future Work 216

are cost-effective for the conditions studied because of low upfront equipment and design
costs, no issues with reservoir temperature increase over time, and efficient system per-
formance due to moderate ambient temperatures and lack of auxiliary equipment such
as fluid circulation pumps.

In the third part of the dissertation, a novel heat transfer model using the slender-
body theory to simulate transient heat transfer of slender, curved bodies in a conduc-
tive medium has been presented. The model is most attractive to simulate heat transfer
with complex and non-compact geometries for which finite element numerical simula-

tions might be too cumbersome, and no analytical solutions exist. Two case-studies, the

first one of two parallel cylinders in close proximity at constant temperature, and the
second of one loop of a slinky-coil heat exchanger, show excellent agreement but much

faster computational times for the slender-body theory model in comparison with a finite

element model. Possible applications are transient heat transfer simulations for deep, de-

viated oil, gas and geothermal wells, shallow geothermal borehole and slinky-coil heat

exchangers, and heating and cooling elements in industrial processes.

16.2 Recommendations for Future Work

It is recommended to upgrade future versions of GEOPHIRES with updated capital and

operation & maintenance cost correlations, especially drilling costs since they typically
are the largest upfront cost and can change significantly over time. Also the heat-to-
electricity conversion efficiency correlations could be upgraded to account for the latest
advancements in power plant technology. Since the four reservoir models currently im-
plemented in GEOPHIRES are rather simple, it might be beneficial to incorporate more
advanced models, e.g. in TOUGH2, or perform simulations with actual geothermal pro-
duction data.
Chapter 16. Overall Conclusions and Recommendations for Future Work 217

GEOPHIRES is continuously being used in our research group, e.g. in a geothermal

play fairway analysis project for the Appalachian Basin, as well as in a few U.S. national
laboratories and universities. Modifying the code for different platforms beyond Win-
dows, perhaps even with an online version, might expand its user database and increase
its reach to other research groups and organizations. Beneficial for future upgrades would
be to convert the programming language from FORTRAN to e.g. Python and upgrade the
legacy code currently part of the GEOPHIRES source code.

While the GEOPHIRES simulations predict attractive performance of deep geother-

mal energy for direct-use heat applications, actual systems are still required to confirm

cost-competitiveness, scalability, and power output under real-life conditions. Across the

world, several successful examples already exist, e.g. geothermal district heating systems

in Iceland, Paris, and Boise. Nonetheless, for widespread deployment, non-hydrothermal

resources should be exploited, i.e. hot dry rock with EGS technology or hot sedimentary

aquifers. Proper research, development & demonstration support is recommended for

developing pilot plants serving various direct-use applications, including district heating

as well as industrial processes, e.g. biomass drying, food pasteurization, and fish farming,

potentially combined in a cascaded way.

The study on hybrid geothermal heat pump systems for cooling-dominated applica-
tions continues in our group with the analysis of the system performance in different

regions in the country, with different weather, geological and economic conditions. In ad-
dition, a detailed study of the impact of groundwater flow, the circulating fluid, advanced
dry-cooler control schemes, and different cooling technologies such as direct-exchange
heat pumps, on the overall system financial and technical performance would be inter-
esting.

While geothermal heat pumps are a well-known technology in the residential sector
allowing to save the homeowner energy and usually also money, great potential lies as
Chapter 16. Overall Conclusions and Recommendations for Future Work 218

well in the commercial and industrial sector. Specifically for cooling-dominated oper-
ation, analysis and development of hybrid geothermal heat pump systems for various
applications beyond cellular tower shelters is recommended, e.g. cooling of small data
centers, or cooled storage of food and beverages. In addition, a stable multi-year tax in-
centive program is recommended to shorten payback periods and level the playing field
with other energy saving technologies.
 APPENDIX A
PERFORMANCE DATASHEETS FOR CLIMATEMASTER HEAT PUMP
TRANQUILITY 27 TWO-STAGE (TT) 026

This appendix provides performance datasheets for the ClimateMaster Tranquility 27

Two-Stage (TT) 026 water-to-air heat pump (ClimateMaster, 2012), which are incorpo-
rated in the TRNSYS hybrid geothermal heat pump system model discussed in Chapter 9.
Performance tables for the heat pump in part load and full load are given in Figures A.1

and A.2, respectively. TC, SC and HR refer to total cooling, sensible cooling, and total heat
rejection in thousands of BTU/hour; kW refers to the power consumption in kW and EER

is the energy efficiency ratio (see also Chapter 7). These cooling performance parameters

are a function of the entering water temperature (EWT), the fluid flow rate (GPM), and

the dry-bulb (DB) and wet-bulb (WB) entering air temperature (EAT).

The performance data provided in Figures A.1 and A.2 is only valid for standard enter-

ing air temperatures and air flow rates. For deviations in dry-bulb and wet-bulb air tem-

perature, and air flow rate, correction tables are provided in Figure A.3 and Figure A.4,

respectively.

References

ClimateMaster (2012). Tranquility 27 Two-Stage (TT) Series Submittal Data, Models

TTD/H/V 026-072, 60Hz - HFC-410A. Technical Report, ClimateMaster, Oklahoma
City, Oklahoma, United States.

219
Appendix A. Performance Datasheets for ClimateMaster Heat Pump TT 026 220

Figure A.1 – Performance datasheet for ClimateMaster heat pump Tranquility 27 TT 026 in part load.
Appendix A. Performance Datasheets for ClimateMaster Heat Pump TT 026 221

Figure A.2 – Performance datasheet for ClimateMaster heat pump Tranquility 27 TT 026 in full load.
Appendix A. Performance Datasheets for ClimateMaster Heat Pump TT 026 222

Figure A.3 – Airflow correction tables for ClimateMaster heat pump Tranquility 27 TT 026 for part load
(a) and full load (b) operation.
Appendix A. Performance Datasheets for ClimateMaster Heat Pump TT 026 223

Figure A.4 – Air temperature correction tables for ClimateMaster heat pump Tranquility 27 TT 026 for
part load (a) and full load (b) operation.
 APPENDIX B
BOREHOLE HEAT EXCHANGER TRANSIENT HEAT TRANSFER MODEL

This appendix presents a computationally-efficient, hybrid heat transfer model for simu-
lating short- and long-term transient heat exchange with borehole heat exchangers (BHEs)
in a conduction-dominated medium. This model is incorporated in the TRNSYS model
discussed in Chapter 9.

B.1 Borehole Heat Exchanger Heat Transfer Problem

A simplified single-U (1-U) BHE is shown in Figure B.1. A heat exchanger fluid, typi-

cally water plus antifreeze, flows downwards and then upwards through the pipe while

exchanging heat with the surrounding grout and soil or rock (ground). The ground un-

dergoes heat transfer through conduction only with corresponding governing equation:

1 ∂T s
− ∇2 T s = 0 (B.1)
α s ∂t

with T s and α s the ground temperature and thermal diffusivity, respectively. It is assumed
the ground has uniform and constant thermophysical properties, and has a far-field, ini-

tial, and surface temperature of T 0 . The grout and the pipes undergo transient heat con-

duction as well with the same governing equation (Equation (B.1)) applicable with the
thermal diffusivity for the grout and pipe material, respectively. At each boundary, the

temperature T and heat flux Q are preserved.

The fluid at temperature T f exchanges heat with the pipe wall at temperature T p
though convection. The governing equation when neglecting thermal diffusion within
the water because of high Peclet numbers is:

∂T f ∂T f 2h
ρc p + ρc p U = (T p − T f ) (B.2)
∂t ∂z r p,i

224
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 225

Figure B.1 – Simplified diagram of single-U BHE. Drawing not to scale.

with ρ, c p , and U the density, specific heat capacity, and velocity of the fluid, respectively.

The heat transfer coefficient h can be calculated with appropriate Nusselt correlations

for forced fluid flow inside a pipe. The internal radius of the pipe is represented by r p,i .

The independent variables t and z are the time and vertical coordinate along the fluid

direction, respectively. The temperature at the inlet T inlet is a specified boundary condition

and the outlet temperature T outlet is being solved for.

B.2 Existing BHE Heat Transfer Models

The heat transfer problem presented in the previous section is rather difficult to solve
because of the presence of various medium properties, and both short and long time-
scales, and short and long length-scales. Several solution methods are available from
simple but less accurate analytical tools to full three-dimensional but computationally-
intense finite element models. Some excellent overviews of simulation tools have been
provided by Yavuzturk (1999), Xu (2007), Yang et al. (2010), Al-Khoury (2012), and He
(2012).
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 226

In simple models, the heat transfer in the ground is often simplified using the infinite
line source model (Ingersoll et al., 1954), infinite cylindrical source model (Carslaw and
Jaeger, 1959), or finite line source model (Zeng et al., 2002). The (average) temperature
at the borehole wall T b is calculated by typically assuming constant heat exchange along
the BHE with the surrounding ground. This heat exchange can be discretized in the time
domain, leading to response factor functions, also called g-functions (Eskilson, 1987). A
g-function can be defined as the (average) temperature rise at the borehole wall caused
by a unit heat pulse active for a certain time period. Superposition is used to account for

various heat pulses and for multiple BHE in a field. One is not restricted to the afore-
mentioned analytical tools to derive g-functions; any method including numerical tools

can be applied. In the BHE hybrid heat transfer model presented in the next section,

g-functions are calculated using the slender body-theory approach, discussed in Part III

of the dissertation. Originally developed for long time-scales, g-functions have been ex-

tended to short time-scales by Yavuzturk (1999) by adjusting them to account for transient

effects inside the BHE. Another popular model for the heat transfer in the ground is the

Duct Storage Model (DST), developed by (Hellström, 1991). In this model, the temper-

ature in the field is calculated as the superposition of the solution to three heat transfer

problems: (1) a global heat conduction problem of the heat exchange between the entire

field and the surrounding undisturbed ground (numerically solved), (2) a local heat con-
duction problem in the ground directly surrounding one BHE (numerically solved), and
(3) a steady-state heat conduction problem for heat transfer between the local and global

domain (analytically solved).

A ground heat transfer model is coupled at the borehole wall to a heat transfer model
inside the BHE. Often the thermal mass of the grout, pipes and fluid are neglected, which
is only accurate for long-term simulations. In this case, the BHE is solely represented by a
set of thermal resistances, e.g. the borehole thermal resistance Rb , and internal resistance
Ra , discussed in more detail in other works such as (He, 2012). Several models have been
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 227

developed to calculate these resistances with an overview provided by Lamarche et al.

(2010). The thermal mass inside the BHE can be accounted for by incorporating a thermal
capacity for the grout, fluid and maybe even pipes. Examples are the models by Bauer
et al. (2011a) and Maestre et al. (2015), which are valid at short time-scales.

Other interesting methods for solving the BHE heat transfer problem, applicable at
short and long time-scales, are utilizing the Fourier transform (Al-Khoury, 2010), or
Laplace transform (Bandyopadhyay et al., 2008; Beier, 2014). Finally, the most accurate
approach but requiring the longest computational times, are simulating the heat transfer

problem with full three-dimensional numerical models based on finite volume methods,

e.g. (Rees and He, 2013), or finite element methods, e.g. (Diersch et al., 2011a,b).

B.3 Novel Hybrid BHE Heat Transfer Model

In the study on hybrid geothermal heat pump systems (Part II of this dissertation), a

large number of cases are investigated, with system lifetimes up to 20 years and heat

pump operational time scales of less than one hour. Therefore, it is essential to have a

computationally-fast BHE heat transfer model that can handle both long and short time-

scales. Inspired by the work of Bauer et al. (2011a) and Maestre et al. (2015), a novel

model is presented in this section which includes elements of both these works while
incorporating some modifications as well.

For the heat transfer in the ground outside the BHE, we will rely on g-functions since
this methodology has been shown many times to perform well. In this approach, an
average borehole wall temperature T b , after discretizing the heat exchange Q in the time
domain, is calculated for time step n as:
[ ( ) ( )]
1 ∑ tn − ti−1 tn − ti
T b,n = T0 + Qi g −g (B.3)
2πk i ts ts
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 228

with t s a time scale, typically set at L2 /(9α) with L the length of the BHE, and with k the
thermal conductivity of the medium. As defined earlier, T 0 is the initial, far-field, and sur-
face temperature. The summation in Equation (B.3) ads up the temperature contributions
from each heat pulse Qi with i from 1 to n with t0 being the start time (0 s). The g-function
itself depends, besides the time, on the length and radius of the BHE as well as the BHE
field configuration. Figure B.2 illustrates the g-functions for several configurations for a
BHE length of 100 m and radius of 0.1 m. The spacing in between the BHEs is 5 m. They
are calculated using the slender-body theory model (Part III) and represent the average

borehole wall temperature assuming a unit heat pulse uniform in space along the BHE
and constant in time from 0 to t, with the BHE(s) placed in a semi-infinite medium with

constant temperature T 0 at the surface.

To decrease the computational time in simulations with many time steps, a load-

aggregation algorithm is applied, based on the work by Yavuzturk (1999). With this algo-

rithm, past heat pulses are combined within time blocks of e.g. 500 hours and represented

by their average value for each time period.

In contrast to some previous work on g-functions such as by Yavuzturk (1999), the

g-functions at short time-scales here are not adjusted to account for transient heat effects

occurring inside the BHE. Rather, they represent the temperature at the borehole wall
caused by a unit heat pulse occurring at the borehole wall, similar to their definition for

long time-scales. In this work, the short time-scale effects are captured by the internal
BHE heat transfer model discussed below. As a result, the heat exchange Q in Equa-
tion (B.3) represents the heat exchange between the BHE and the ground, which is not
necessarily equal to the heat exchange between the fluid and the BHE, unless the heat
exchange inside the BHE has reached steady-state.

The BHE itself is discretized internally along the vertical direction as well as in the
horizontal cross-sectional plane, shown in Figure B.3. Unlike the models of Bauer et al.
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 229

Figure B.2 – Plot of g-functions as a function of dimensionless time for several BHE field configurations,
calculated with slender-body theory model (Part III). The time scale t s is defined as L2 /(9α). The BHE
length L and radius rb are 100 m and 0.1 m, respectively. Spacing between boreholes is 5 m.

(2011a) and Maestre et al. (2015), the grout around each pipe is represented by several

nodes instead of only one, in order to allow for better capturing the short time-scale phe-
nomena (see Section B.4). The number of cylindrical shells around each pipe in Figure B.3

is two but it could be any number as long as the largest shells still fit together in the BHE.

Each pipe and grout zone has a heat capacity C and is represented by a temperature T .
The heat transfer in between contacting zones has a corresponding thermal resistance R.
The subscripts d and u refer to the side of the BHE at which the fluid flows downwards
and upwards, respectively. It is assumed the BHE is symmetrical, i.e. the upward and
downward pipe are identical and located on the horizontal axis at the same distance from
the center. The network of thermal resistances and capacities within one layer is shown
in Figure B.4.
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 230

Figure B.3 – Cross section of BHE illustrating spatial discretization in horizontal plane. Left section
(subscript d) refers to side with fluid flowing downwards. Right section (subscript u) refers to side
with fluid flowing upwards. Grout directly surrounding the pipes is discretized into cylindrical shells
to allow for accurate capturing of short-term transient heat exchange.

The units of C and R are J/K and m·K/W. The heat capacity of the pipe and grout

zones are simply calculated as the specific heat capacity and density of the material times
the area of the zone times the thickness of the layer times. For example, the heat capacity

of the pipe C p is calculated as:

C p = ρ p c p,p π(r2p,o − r2p,i )∆z (B.4)

with ρ p and c p,p the density and specific heat capacity of the pipe material, respectively,
and ∆z the layer thickness. The thermal resistances for the cylindrical shells can eas-
ily be calculated using simple expressions. For example, the thermal resistance for heat
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 231

Figure B.4 – Network of thermal resistances (R) and capacities (C) within 1 horizontal layer of a dis-
cretized BHE. Each black solid circle corresponds to a temperature T in the network. The subscripts f,
p, g, b, u, d, and n refer to fluid, pipe, grout, borehole wall, upwards, downwards, and nth grout zone.

exchange between the fluid and the pipe R f p involves a convective and conductive com-

ponent:
( )
1 1 r p,o + r p,i
Rf p = + ln (B.5)
πk f Nu 2πk p 2r p,i
with k f and k p the thermal conductivity of the fluid and pipe, respectively, and Nu the

Nusselt number for forced flow inside the pipe (Bergman et al., 2011). The thermal resis-

tance between the first grout zone and the pipe is calculated as:
( ) ( )
1 2r p,o 1 rg,1 + r p,o
Rg,1 = ln + ln (B.6)
2πk p r p,o + r p,i 2πkg 2r p,o
with kg the thermal conductivity of the grout. The more exotic thermal resistances Rg,n ,

Rgg , and Rg,b are calculated with equations provided by Bauer et al. (2011b).

The BHE is discretized in the vertical direction along the BHE into several layers, as

shown in Figure B.5. The layers are coupled through fluid advection within the upward
and downward pipe, and conduction in the grout and pipes as represented by thermal
resistances Rgg,n to R pp . These resistances are easily calculated using expressions for one-
dimensional conduction in rectangular coordinates. R pp for example, the thermal resis-
tance between two pipe segments, is calculated as:
∆z
R pp = (B.7)
k p π(r2p,o − r2p,i )
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 232

Figure B.5 – Diagram showing layered thermal network within BHE. Different layers are coupled
through conductive heat exchange in the grout and pipes with corresponding thermal resistances Rgg,n
to R pp , and fluid advection within the pipes labeled by blue arrows. Each layer has an identical borehole
wall temperature T b .

which assumes uniform vertical discretization for which ∆z, the layer thickness, is equal

to the distance between two pipe temperature nodes. The borehole wall temperature T b
is identical for each layer, a characteristic of the g-functions approach.

A simple energy balance can be constructed for each temperature node in the thermal
network. For example, for the first grout cylindrical shell on the downward side:
∂T g,d,1,i T g,d,2,i − T g,d,1,i T p,d,i − T g,d,1,i T g,d,1,i+1 − T g,d,1,i T g,d,1,i−1 − T g,d,1,i
Cg,1 = + + + (B.8)
∂t Rg,1 Rf p Rgg,1 Rgg,1
with the additional subscripts i, i + 1, and i − 1 referring to the layer number. A simple
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 233

implicit scheme is utilized for the time integration to ensure high stability. More advanced
schemes such as Runge-Kutta methods are available, but are not investigated here.

For the fluid energy balance, we directly incorporate the analytical solution for fluid
flow in a pipe with constant wall temperature. For a certain pipe segment and during

a certain time step, the wall temperature, here the pipe temperature T p , can be assumed
constant. When neglecting the thermal diffusion within the fluid, the outlet temperature
is calculated as (Van Genuchten and Alves, 1982):
( )
∆z
T outlet = T p + (T inlet − T p ) · exp − (B.9)
R f pv

with T inlet the segment inlet temperature, and assuming that the fluid reached the end of

the segment. In Equation (B.9), the parameter v is calculated as:

v = ρc p Uπr2p,i = mc p (B.10)

with m the fluid mass flow rate. For the heat transfer within the BHE, the time step size

chosen internally in the algorithm corresponds to the time period for the fluid flowing

through one pipe segment. As a result, the inlet temperature in Equation (B.9) is the

outlet temperature of the previous segment at the previous internal time step. By relying

on the analytical solution and with judicious choice of the internal time step, numerical
diffusion is limited, and high stability and accuracy are obtained. During a simulation,
the flow rate m is allowed to vary.

The above equations have been implemented as a simulation tool in MATLAB (Math-
Works, 2012) and coupled to TRNSYS (see Chapter 9). It can be considered a hybrid BHE
heat transfer model, since it relies on both analytical as well as numerical equations, and
involves both g-functions as well as a thermal resistance - capacity (“RC”) network. Other
types of heat exchangers (e.g. double-U (2-U) and coaxial) are not discussed here but can

be implemented with the presented model as well.

Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 234

Table B.1 – Case-study parameter values for validating BHE heat transfer model.

BHE length L 100 m

BHE radius rb 0.065 m
Pipe inner radius r p,i 0.0131 m
Pipe outer radius r p,o 0.0160 m
Spacing between center of pipes w 0.06 m
Ground thermal conductivity ks 2.2 W/(m·K)
Ground thermal diffusivity αs 9.9548·10−7 m2 /s
Grout thermal conductivity kg 2.3 W/(m·K)
Grout thermal diffusivity αg 1.0502·10−6 m2 /s
Pipe thermal conductivity kp 0.38 W/(m·K)
Pipe thermal diffusivity αp 1.6579·10−7 m2 /s
Fluid density ρ 1000 kg/m3
Fluid specific heat capacity cp 4130 J/(kg·K)
Fluid thermal conductivity k 0.6405 W/(m·K)
Fluid dynamic viscosity µ 0.54741·10−3 kg/(m·s)
Fluid mass flow rate m 0.25 kg/s
Fluid inlet temperature T inlet 80◦ C
Initial temperature T0 10◦ C
Vertical discretization N 20

B.4 Validation Case-Study

The presented hybrid BHE heat transfer model is validated in this section with a short

time-scale case-study, identical to one of the case-studies discussed by Bauer et al. (2011a).

The outlet temperature for a single-U BHE with fully described geometry and material
and fluid properties is simulated for known inlet temperature for a time period of 180

minutes. All parameters are listed in Table B.1.

The simulation results for the outlet temperature as a function of time are shown in
Figure B.6. The result of the model with one node for the pipe and four nodes for the grout
(at each side) compares well with the result of an ANSYS reference simulation (taken from
(Bauer et al., 2011a)). A model with only one node for the grout at each side overpredicts
the temperature initially but eventually produces the same result.
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 235

Figure B.6 – Case-study simulation results for validating hybrid BHE heat transfer model show good
agreement between model with one node for pipe and four nodes for grout (at each side) and ANSYS
reference case. Model with one node for the grout initially overpredicts the outlet temperature but
eventually predicts the same result.

References

Al-Khoury, R. (2010). Spectral framework for geothermal borehole heat exchangers. In-
ternational Journal of Numerical Methods for Heat & Fluid Flow, 20 (7): 773–793.

Al-Khoury, R. (2012). Computational Modeling of Shallow Geothermal Systems. CRC Press.

Bandyopadhyay, G., Gosnold, W., and Mann, M. (2008). Analytical and semi-analytical
solutions for short-time transient response of ground heat exchangers. Energy and Build-
ings, 40 (10): 1816–1824.

Bauer, D., Heidemann, W., and Diersch, H.-J. (2011a). Transient 3d analysis of borehole
heat exchanger modeling. Geothermics, 40 (4): 250–260.
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 236

Bauer, D., Heidemann, W., Müller-Steinhagen, H., and Diersch, H.-J. (2011b). Thermal
resistance and capacity models for borehole heat exchangers. International Journal of
Energy Research, 35 (4): 312–320.

Beier, R. A. (2014). Transient heat transfer in a u-tube borehole heat exchanger. Applied
Thermal Engineering, 62 (1): 256–266.

Bergman, T. L., Lavine, A. S., Incropera, F. P., and Dewitt, D. P. (2011). Introduction to heat

transfer. John Wiley & Sons, 6th edition.

Carslaw, H. S. and Jaeger, J. C. (1959). Conduction of heat in solids. Clarendon Press, Oxford.

Diersch, H.-J., Bauer, D., Heidemann, W., Rühaak, W., and Schätzl, P. (2011a). Finite
element modeling of borehole heat exchanger systems: Part 1. fundamentals. Computers

& Geosciences, 37 (8): 1122–1135.

Diersch, H.-J., Bauer, D., Heidemann, W., Rühaak, W., and Schätzl, P. (2011b). Finite

element modeling of borehole heat exchanger systems: Part 2. numerical simulation.

Computers & Geosciences, 37 (8): 1136–1147.

Eskilson, P. (1987). Thermal Analysis of Heat Extraction Boreholes. PhD Dissertation, Lund
University, Lund, Sweden.

He, M. (2012). Numerical modelling of geothermal borehole heat exchanger systems. PhD Dis-
sertation, De Montfort University, Leicester, United Kingdom.

Hellström, G. (1991). Ground Heat Storage: Thermal Analyses of Duct Storage Systems. PhD
Dissertation, Lund University, Lund, Sweden.

Ingersoll, L., Zobel, O. J., and Ingersoll, A. C. (1954). Heat Conduction: With Engineering
Geological And Other Applications. Oxford And Ibh Publishing Co.; Calcutta; Bombay;
New Delhi.
Appendix B. Borehole Heat Exchanger Transient Heat Transfer Model 237

Lamarche, L., Kajl, S., and Beauchamp, B. (2010). A review of methods to evaluate bore-
hole thermal resistances in geothermal heat-pump systems. Geothermics, 39 (2): 187–200.

Maestre, I. R., Gallero, F. J. G., Gómez, P. Á., and Pérez-Lombard, L. (2015). A new rc
and g-function hybrid model to simulate vertical ground heat exchangers. Renewable

Energy, 78: 631–642.

MathWorks (2012). MATLAB Release 2012a, The Mathworks, Inc., Natick, Massachusetts,

United States.

Rees, S. J. and He, M. (2013). A three-dimensional numerical model of borehole heat

exchanger heat transfer and fluid flow. Geothermics, 46: 1–13.

Van Genuchten, M. T. and Alves, W. J. (1982). Analytical solutions of the one-dimensional

convective-dispersive solute transport equation. Technical Report, United States De-

partment of Agriculture, Technical Bulletin No. 1661, 151p.

Xu, X. (2007). Simulation and optimal control of hybrid ground source heat pump systems. PhD

Dissertation, Oklahoma State University, Stillwater, Oklahoma, United States.

Yang, H., Cui, P., and Fang, Z. (2010). Vertical-borehole ground-coupled heat pumps: a
review of models and systems. Applied Energy, 87 (1): 16–27.

Yavuzturk, C. (1999). Modeling of vertical ground loop heat exchangers for ground source
heat pump systems. PhD Dissertation, Oklahoma State University, Stillwater, Oklahoma,

United States.

Zeng, H., Diao, N., and Fang, Z. (2002). A finite line-source model for boreholes in
geothermal heat exchangers. Heat TransferAsian Research, 31 (7): 558–567.
 APPENDIX C
TRNSYS SIMULATION RESULTS

This chapter provides detailed simulation results for the various TRNSYS cases studied
in Chapter 9. Reported are the total cost of ownership (TCO) in $1,000, the lifetime elec-
tricity consumption in MWhe , lifetime CO2 emissions in million metric tons, the average
heat pump (HP) coefficient of performance (COP) (dimensionless) and the borehole heat
exchanger (BHE) configuration. The lifetime electricity consumption and CO2 emissions

do not account for the installation of the system. For cases involving a ground-source heat
pump (GSHP), the average heat pump COP does account for the electricity consumption

of the fluid circulation pump. Table C.1 provides the results for the total BHE depth sen-

sitivity analysis for Case 1 (GSHP only). Tables C.2, C.3, C.4, C.5, and C.6 provide the

results of the sensitivity analysis of the shelter heat generation, reservoir thermal conduc-

tivity, electricity rate, net discount rate, and drilling costs, respectively. ASHP, AE, and

DC refer to air-source heat pump, air-economizer, and dry-cooler, respectively.

Table C.1 – Effect of total BHE depth on system performance in Case 1 (GSHP only) in base case
scenario.

Total Electricity CO2 Emissions Average

BHE TCO
BHE Consumption (million metric HP COP
Configuration ($1,000)
Depth (MWhe ) tons) [-]
360 m 3 x 120 m 93.8 566 105.3 3.42
390 m 3 x 130 m 90.9 522 97.2 3.70
420 m 3 x 140 m 88.8 487 90.5 3.97
450 m 3 x 150 m 87.6 460 85.6 4.20
480 m 4 x 120 m 88.6 455 84.7 4.25
510 m 4 x 127.5 m 88.0 435 80.8 4.45
540 m 4 x 135 m 87.8 417 77.6 4.64
570 m 4 x 142.5 m 87.8 403 75 4.8
600 m 4 x 150 m 88.1 391 72.7 4.95
630 m 6 x 105 m 90.1 396 73.7 4.88
660 m 6 x 110 m 90.4 384 71.5 5.03
690 m 6 x 115 m 91.0 376 69.9 5.15
720 m 6 x 120 m 91.8 369 68.6 5.25

238
Appendix C. TRNSYS Simulation Results 239

Table C.2 – Sensitivity of TRNSYS simulation results to shelter heat generation. All other parameters
are kept the same as in the base case scenario. Results for 11.5 kWth are provided in Table 9.4.

Electricity CO2 Emissions Average

TCO BHE
Case Consumption (million metric HP COP
($1,000) Configuration
(MWhe ) tons) [-]
8.5 kWth Shelter Heat Generation
GSHP 66.0 277 51.5 5.09 3 x 130 m
GSHP + AE 48.4 145 26.9 4.76 2 x 115 m
GSHP + DC 59.8 302 56.1 6.00 2 x 95 m
ASHP 62.1 430 79.9 3.28 N/A
ASHP + AE 45.1 228 42.4 3.02 N/A
10 kWth Shelter Heat Generation
GSHP 76.7 352 65.4 4.76 3 x 150 m
GSHP + AE 53.9 178 33.2 4.59 2 x 135 m
GSHP + DC 68.0 366 68.1 5.58 2 x 110 m
ASHP 70.4 510 94.8 3.28 N/A
ASHP + AE 49.5 271 50.4 3.02 N/A

Table C.3 – Sensitivity of TRNSYS simulation results to reservoir thermal conductivity. All other pa-
rameters are kept the same as in the base case scenario. Results for 2.0 W/(m·K) are provided in
Table 9.4.

Electricity CO2 Emissions Average

TCO BHE
Case Consumption (million metric HP COP
($1,000) Configuration
(MWhe ) tons) [-]
1.0 W/(m·K) Reservoir Thermal Conductivity
GSHP 114.4 507 94.4 3.81 6 x 150 m
GSHP + AE 73.9 258 47.9 3.67 4 x 127.5 m
GSHP + DC 85.3 447 83.1 5.09 3 x 133.33 m
3.0 W/(m·K) Reservoir Thermal Conductivity
GSHP 77.8 371 69.1 5.21 3 x 143.33 m
GSHP + AE 55.4 197 36.7 4.83 2 x 130 m
GSHP + DC 72.4 408 75.9 5.67 2 x 110 m
4.0 W/(m·K) Reservoir Thermal Conductivity
GSHP 73.0 359 66.7 5.39 3 x 120 m
GSHP + AE 53.0 189 35.1 5.06 2 x 115 m
GSHP + DC 70.6 405 75.3 5.72 2 x 95 m
Appendix C. TRNSYS Simulation Results 240

Table C.4 – Sensitivity of TRNSYS simulation results to electricity rate. All other parameters are kept
the same as in the base case scenario. Results for 14 ¢/kWhe are provided in Table 9.4.

Electricity CO2 Emissions Average

TCO BHE
Case Consumption (million metric HP COP
($1,000) Configuration
(MWhe ) tons) [-]
10 ¢/ kWhe Electricity Rate
GSHP 74.2 460 85.6 4.20 3 x 150 m
GSHP + AE 53.3 228 42.5 4.15 2 x 145 m
GSHP + DC 63.0 438 81.4 5.22 2 x 115 m
ASHP 61.2 589 109.6 3.28 N/A
ASHP + AE 44.7 314 58.4 3.02 N/A
12 ¢/ kWhe Electricity Rate
GSHP 80.9 460 85.6 4.20 3 x 150 m
GSHP + AE 56.6 222 41.3 4.27 2 x 150 m
GSHP + DC 69.4 425 79.1 5.40 2 x 125 m
ASHP 70.0 589 109.6 3.28 N/A
ASHP + AE 49.4 314 58.4 3.02 N/A
16 ¢/ kWhe Electricity Rate
GSHP 93.7 403 75.0 4.80 4 x 142.5 m
GSHP + AE 63.2 222 41.3 4.27 2 x 150 m
GSHP + DC 81.9 416 77.4 5.54 2 x 135 m
ASHP 87.5 589 109.6 3.28 N/A
ASHP + AE 58.7 314 58.4 3.02 N/A
Appendix C. TRNSYS Simulation Results 241

Table C.5 – Sensitivity of TRNSYS simulation results to net discount rate. All other parameters are
kept the same as in the base case scenario. Results for 3% net discount rate are provided in Table 9.4.

Electricity CO2 Emissions Average

TCO BHE
Case Consumption (million metric HP COP
($1,000) Configuration
(MWhe ) tons) [-]
5% Net Discount Rate
GSHP 79.1 460 85.6 4.20 3 x 150 m
GSHP + AE 55.1 222 41.3 4.27 2 x 150 m
GSHP + DC 68.0 425 79.1 5.40 2 x 125 m
ASHP 66.7 589 109.6 3.28 N/A
ASHP + AE 46.2 314 58.4 3.02 N/A
7% Net Discount Rate
GSHP 72.5 467.5 86.9 4.14 3 x 146.67 m
GSHP + AE 51.4 228 42.5 4.15 2 x 145 m
GSHP + DC 61.9 438 81.4 5.22 2 x 115 m
ASHP 57.4 589 109.6 3.28 N/A
ASHP + AE 40.1 314 58.4 3.02 N/A
9% Net Discount Rate
GSHP 67.3 467.5 86.9 4.14 3 x 146.67 m
GSHP + AE 48.3 243 45.3 3.89 2 x 135 m
GSHP + DC 57.0 445 82.8 5.12 2 x 110 m
ASHP 50.1 589 109.6 3.28 N/A
ASHP + AE 35.3 314 58.4 3.02 N/A

Table C.6 – Sensitivity of TRNSYS simulation results to drilling cost. All other parameters are kept the
same as in the base case scenario. Results for 50 $/m are provided in Table 9.4.

Electricity CO2 Emissions Average

TCO BHE
Case Consumption (million metric HP COP
($1,000) Configuration
(MWhe ) tons) [-]
30 $/m Drilling Cost
GSHP 76.1 391 72.7 4.95 4 x 150 m
GSHP + AE 53.6 189 35.3 5.02 3 x 133.33 m
GSHP + DC 70.1 405 75.4 5.72 2 x 150 m
70 $/m Drilling Cost
GSHP 96.6 460 85.6 4.20 3 x 150 m
GSHP + AE 65.8 228 42.5 4.15 2 x 145 m
GSHP + DC 80.5 438 81.4 5.22 2 x 115 m
90 $/m Drilling Cost
GSHP 105.5 467 86.9 4.14 3 x 146.67 m
GSHP + AE 71.4 243 45.3 3.89 2 x 135 m
GSHP + DC 85.0 445 82.8 5.12 2 x 110 m
 APPENDIX D
NON-DIMENSIONAL NUMBERS, HEAT SOURCE MODEL EQUATIONS, AND
DECISION TREES FOR CALCULATING fi, j,m,n

This appendix provides the procedure and equations for calculating fi, j,m,n , the terms that
emerged in Equation (13.1) after discretizing the (slender-body theory) SBT theoretical
result for developing the numerical framework (Chapter 13), which was used in the case-
studies (Chapter 14). As mentioned in Section 13.2, fi, j,m,n can be interpreted as the temper-

ature rise at the mid-point of element i at the end of time step m due to a unit heat pulse by
element j during time step n. Without simplifications, the numerical framework would

require for each pipe element at each time step to evaluate the full Equation (12.13) or

(12.16), i.e. calculating the full finite cylindrical or line source model and the contributions

from all previous and neighboring heat pulses. However, simplifications are possible to

save on computational time with no significant loss in accuracy. These simplifications are

neglecting the finite radius or length of a pipe element, or neglecting the element as a

whole, which translates into different models for calculating fi, j,m,n .

Six well-known models are identified to calculate fi, j,m,n , ordered from more to less rep-
resenting the actual geometry and from more to less computationally-intense to evaluate:

the Finite Cylindrical Source (FCS) model, Infinite Cylindrical Source (ICS) model, Finite

Line Source (FLS) model, Infinite Line Source (ILS) model, Point Source (PS) model, and

Table D.1 – Non-dimensional numbers utilized in equations and decision trees for calculating fi, j,m,n .
In these numbers, Fo, TDR, SLR, R, L, S, CP, PP, S and E refer to Fourier number, time durations ratio,
spacing to length ratio, pipe radius, pipe length, spacing between two pipes, current pulse, previous
pulse, start of pulse, and end of pulse, respectively.

α(tm −tm−1 ) α(tm −tm−1 ) α(tm −tm−1 )

FoR,CP = R2j
FoL,CP = L2j
FoS,CP = S i,2 j
α(tm −tn−1 ) α(tm −tn−1 ) α(tm −tn−1 )
FoR,PP,S = R2 FoL,PP,S = L2j
FoS,PP,S = S 2
j i, j

FoR,PP,E = α(tmR−t 2
n) α(tm −tn )
FoL,PP,E = L2 FoS,PP,E = α(tSm2−tn )
j j i, j

TDR = tntm−t−tn−1n
S
SLR = Li,jj

242
Appendix D. Non-Dimensional Numbers, Heat Source Model Equations, and
Decision Trees for Calculating fi, j,m,n 243

Table D.2 – Heat source model equations to calculate fi, j,m,n for selected pipe (Carslaw and Jaeger, 1959).
The different models are: Finite Cylindrical Source (FCS), Infinite Cylindrical Source (ICS), Finite Line
Source (FLS), Infinite Line Source (ILS), Point Source (PS), and Point Neglected (PN) Model.

fi,i,m,m (Selected Pipe & Current Pulse)

∫∞ 1−exp(−ϕ2 FoR,CP ) ∫∞ ( √ )
FCS: fi,i,m,m = 2
π3 k 0 ϕ3 (J12 (ϕ)+Y12 (ϕ))
dϕ − 1
4πk 1
1
ϕ
erfc 12 ϕ dϕ
∫∞
4FoL,CP
1−exp(−ϕ2 FoR,CP )
ICS: fi,i,m,m = 2
π k 0
dϕ
( ϕ3 (J1 (ϕ)+Y
) 1 (ϕ))∫ ∞ ( √ )
3 2 2

FLS: fi,i,m,m = 1
E 1
4πk 1 4FoR,CP
− 4πk 1
1
1
ϕ
erfc 12 ϕ dϕ
( )4Fo L,CP

ILS: fi,i,m,m = 1
E 1
4πk 1 4FoR,CP

fi,i,m,n (Selected Pipe & Previous Pulses)

∫∞ exp(−ϕ2 FoR,PP,E )−exp(−ϕ2 FoR,PP,S ) ∫ 1 ( √ )

fi,i,m,n = − ϕ dϕ
2 1 4FoL,PP,E 1
FCS: π3 k 0 ϕ3 (J12 (ϕ)+Y12 (ϕ))
dϕ 4πk 1 ϕ
erfc 12
∫∞
4FoL,PP,S
exp(−ϕ2 FoR,PP,E )−exp(−ϕ2 FoR,PP,S )
ICS: fi,i,m,n = 2
π3 k
dϕ
[0 (
ϕ3 (J12 (ϕ)+Y12 (ϕ))
) ( ) ( √ ) ]
∫ 1

fi,i,m,n = E1 4FoR,PP,S − E1 4FoR,PP,E − ϕ dϕ

1 1 1 4FoL,PP,E 1
FLS: 4πk 1 ϕ
erfc 12
[ ( ) ( )] 4FoL,PP,S
ILS: fi,i,m,n = 4πk
1 1
E1 4FoR,PP,S − E1 4FoR,PP,E
1
[ ( ) ( )]
PS1 : fi,i,m,n = 4πkRi erfc √
Li 1
− erfc √ 1

( 2 FoR,PP,S
) 2 FoR,PP,E

PS2 : fi,i,m,n = 4π √απk √tm −tn − √tm −tn−1

Li 1 1

PS3 : fi,i,m,n = 8πkTDR √Liαπ(t −t )

m n

Point Neglected (PN) model. The PN model simply means setting the temperature con-
tribution to zero by neglecting the element. The equations for these heat source models,
as well as the methodology to select the most appropriate model rely on dimensionless

numbers, listed in Table D.1. In these numbers, R j and L j are the radius and length of pipe
element j emitting the heat pulse, S i, j is the spacing between the centers of pipe element
i, where the temperature is calculated at, and pipe element j, the source of the heat pulse,
TDR refers to the ratio of time durations, SLR refers to spacing-to-length ratio, CP and PP
refer to current pulse and previous pulse, and S and E refer to start and end of a previous
heat pulse, respectively. Furthermore, the time steps are in general labeled as tk with t0
the time starting point set at 0 s. tm is the current time and tn is the end time of heat pulse
Appendix D. Non-Dimensional Numbers, Heat Source Model Equations, and
Decision Trees for Calculating fi, j,m,n 244

Table D.3 – Heat source model equations to calculate fi, j,m,n for neighboring pipes (Carslaw and Jaeger,
1959). The different models are: Finite Cylindrical Source (FCS), Infinite Cylindrical Source (ICS), Finite
Line Source (FLS), Infinite Line Source (ILS), Point Source (PS), and Point Neglected (PN) Model.

fi, j,m,m (Neighboring Pipes & Current Pulse)

∫ Lj ( S i, j (z′ )
)
FLS: fi, j,m,m = 1 1
S i, j (z(′ )
erfc √ dz′
4πk 0 ) m −tm−1 )
4α(t

PS1 : fi, j,m,m = 1

4πkSLR
erfc √ 1
(
2 FoS,CP
)
PS2 : fi, j,m,m = 1
4πkSLR
1− √ 1
πFoS,CP
PN: fi, j,m,m = 0

fi, j,m,n (Neighboring Pipes & Previous Pulses)

[ ( ) ( )]
∫ Lj S i, j (z′ ) S i, j (z′ )
FLS: fi, j,m,n = 1
4πk 0
1
S i, j (z′ )
− erfc
erfc √
4α(tm −tn−1 )
√
4α(tm −tn )
dz′
[ ( ) ( )]
PS1 : fi, j,m,n = 4πkSLR erfc √
1 1
− erfc √ 1

Lj
( 2 FoS,PP,S
) 2 FoS,PP,E

PS2 : fi, j,m,n = 4π √απk √tm −tn − √tm −tn−1

1 1

L
PS3 : fi, j,m,n = 8πkTDR √απ(t
j
m −tn )
PN: fi, j,m,n = 0

n. Also, (tm − tm−1 ) is the time duration of the current heat pulse, and (tm − tn−1 ) and (tm − tn )

are the time that has past since the beginning and end of heat pulse n, respectively.

The equations to calculate fi, j,m,n are provided in Tables D.2 and D.3. They are grouped

into four categories, corresponding to the four terms in Equation (13.1), by distinguishing

between selected ( j = i) and neighboring ( j , i) pipe elements, and current (n = m)

and previous (n < m) heat pulses. In the equations, J1 (x), Y1 (x), erfc(x), and E1 (x) refer
to first order Bessel function of the first kind, first order Bessel function of the second

kind, complimentary error function, and exponential integral, respectively (Abramowitz

and Stegun, 1964). The temperature contribution from a previous heat pulse Qi,n , emitted
during the time period tn−1 to tn , is calculated as the sum of the contributions by a heat
pulse with the same strength active from tn−1 to tm (current time) and a heat pulse with
Appendix D. Non-Dimensional Numbers, Heat Source Model Equations, and
Decision Trees for Calculating fi, j,m,n 245

Figure D.1 – Four decision trees are provided to select the least computationally-intense model to still
accurately calculate f . FCS, ICS, FLS, ILS, PS and PN refer to Finite Cylindrical Source, Infinite Cylin-
drical Source, Finite Line Source, Infinite Line Source, Point Source and Point Neglected, respectively.
The non-dimensional numbers are defined in Table D.1. The accuracy obtained in calculating f is
at least 99% for dominant terms, and 95% for non-dominant terms. These decision trees are valid if
L
10 ≤ R jj ≤ 1000 for each element j.

opposite strength active from tn to tm .

Figure D.1 provides decision trees that have been developed to recommend the least

computationally-intense model that still accurately calculates f . The limits in the decision
trees are calculated to obtain an accuracy of at least 99% in calculating dominant f and at
least 95% for non-dominant f in Equation (13.1). The dominant terms are the f for the cur-
rent and recent pulses for the selected pipe and the nearby pipes and are characterized by
having a value of up to several orders of magnitude larger than the non-dominant f . Dif-
ferent decision tree limits could be developed for low-, medium- and high-accuracy sim-
ulations with corresponding high, medium, and low computational speed. The decision
Appendix D. Non-Dimensional Numbers, Heat Source Model Equations, and
Decision Trees for Calculating fi, j,m,n 246

tree limits have been derived in MATLAB by finding the values of the non-dimensional
numbers for which the solution of a certain simplified model deviates no more than e.g.
99% with respect to the solution of the most accurate model available, e.g. the FCS model
in the case of the first two decision trees.

References

Abramowitz, M. and Stegun, I. A. (1964). Handbook of mathematical functions: with formulas,

graphs, and mathematical tables. Number 55. Courier Corporation, Mineola, New York.

Carslaw, H. S. and Jaeger, J. C. (1959). Conduction of heat in solids. Clarendon Press, Oxford.