Meraj X
Meraj X
Meraj X
by
A thesis
presented to the University of Waterloo
in fulfillment of the
thesis requirement for the degree of
Master of Applied Science
in
Civil Engineering
I hereby declare that I am the sole author of this thesis. This is a true copy of the thesis, including any
required final revisions, as accepted by my examiners.
I understand that my thesis may be made electronically available to the public.
ii
Abstract
Solar road panels are a technology that have the ability to revolutionize the way that roads are built
and how electricity is generated. Strong incentives towards sustainable solutions in both of these
fields have led to the design of innovative, multifaceted solutions, of which solar road panels are one
of the most recent entrants. This research presents some initial analysis into the design of solar road
panels from the perspective of Canadian pavement engineering. The hypothesis of this research was
as follows:
Through a process that covers the design, construction, and analysis of the structural elements of
a solar road panel prototype, this research evaluated the impact that solar road panels can have for
Canada’s pavement infrastructure. Specific elements researched include the material selection for
such a panel, the flexural response of the composite structure, how the panel will interact with
traditional pavement and geotechnical materials while in use, and the change in performance of
transparent layer materials as they are subjected to freeze-thaw cycling and scaling.
The research found that the initial prototype design included a two 10-mm tempered glass pane
transparent layers with a 12.7-mm GPO-3 optical layer and 19.1-mm GPO-3 base layer. The concept
being that the glass would provide the rigidity required to protect the fragile solar cells while the
fiberglass laminate has demonstrated performance as a traffic-supporting material in adverse
conditions.
Testing of this structure found that the performance was easily duplicated through finite element
analysis, given that the material properties were assumed to be more rigid than the averages for
tempered glass and GPO-3. Further finite element analysis demonstrated that the prototype solar road
panel would not fail through traditional fatiguing methods, and in all cases on concrete, asphalt,
granular, and subgrade bases the panels improved the performance characteristics of the structural
base.
The environmental conditioning of acrylic, glass, and polycarbonate specimens demonstrated that
glass is the ideal material choice for the transparent layer for Canadian solar road panels. It proved to
have the greatest freeze-thaw and scaling resistance of the three materials, and while the friction
iii
characteristic of the flat glass samples would not be suitable for driving on, avenues of research were
identified that could improve this characteristic.
In summary, the research conducted clearly proved the hypothesis; it is possible to build a
structure that can house a photovoltaic system while supporting the structural and environmental
loads that Canadian pavement are exposed to. The ideal panel would be constructed with a tempered
glass transparent layer, GPO-3 optical and base layers, and the structure would be installed on a
concrete structural base. The refinement of this design will be the scope for future research.
iv
Acknowledgements
I first and foremost would like to acknowledge my supervisor, Dr. Susan Tighe, for all of her support
in completing this research program. Without her this project would not have existed and her vision
and guidance were critical in helping me get through the challenges of this expansive project. I will
always be grateful for what I have learned from you and hope to pay the favor forward to many junior
engineers in the future.
I would next like to thank the Natural Science and Engineering Research Council of Canada,
Ontario Graduate Scholarship Program, University of Waterloo, and the Transportation Association
of Canada for various levels of funding support throughout this project, without which none of my
research would have been completed.
I would also like to thank all of the technical staff who helped me realize my research program.
This includes Richard Morrison for his support in designing my structural testing program, Rob
Sluban for helping me with all of my instrumentation needs, Douglas Hirst for designing my
hydraulic load system and for other miscellaneous design needs, Phill Laycock and Andrew Urschel
for providing design advice for fixture development, Rob Kraemer for consultation and construction
on the testing frame design, Mark Griffett for his training and equipment provided for the three-point
bending testing, and Dr. Michael Collins for providing training for and access to optical testing
equipment.
Special thanks are extended to all of my colleagues who proved reliable time and time again at
providing advice and for bouncing ideas off of. This includes, but is not limited to, Zaid Alyami, Dr.
Mohab El-Hakim, Marcelo Gonzalez, Aleks Kivi, Boyd Panton, and Magdy Shaheen. Developing my
research was a challenging process and your support helped make it possible.
Last, but certainly not least, I would like to thank my family, friends, and amazing girlfriend for
helping keep me relatively sane through the last three years of my exhaustive research program. It’s
been an amazingly stressful adventure and without your support I could not have accomplished it.
There are far too many people to name here, but you all know who you are and I am forever in your
debt for the help you have given me.
v
Table of Contents
Author's Declaration .............................................................................................................................. ii
Abstract ................................................................................................................................................. iii
Acknowledgements ................................................................................................................................ v
Table of Contents .................................................................................................................................. vi
List of Figures ........................................................................................................................................ x
List of Tables ...................................................................................................................................... xiv
Chapter 1 Introduction ........................................................................................................................... 1
1.1 Solar Road Panels ........................................................................................................................ 2
1.2 Research Scope and Hypothesis .................................................................................................. 4
1.3 Research Objectives ..................................................................................................................... 6
1.4 Research Contribution ................................................................................................................. 6
1.5 Organization of Thesis ................................................................................................................. 7
Chapter 2 Design Elements .................................................................................................................... 8
2.1 Pavement Design Elements.......................................................................................................... 8
2.1.1 Traditional Pavement Design ................................................................................................ 8
2.1.2 Non-Conventional Pavement and Flooring Materials......................................................... 10
2.1.3 Structural Testing of Pavement ........................................................................................... 16
2.1.4 Numerical Analysis in Pavement Design ............................................................................ 18
2.2 Solar Module Design Elements.................................................................................................. 18
2.2.1 Photovoltaic Energy Conversion ........................................................................................ 18
2.2.2 Solar Module Design Requirements ................................................................................... 20
2.2.3 Solar Module Design .......................................................................................................... 21
2.3 Solar Road Panel Systems ......................................................................................................... 25
2.3.1 State-of-the-Practice............................................................................................................ 25
2.3.2 Knowledge Gaps ................................................................................................................. 27
Chapter 3 Solar Road Panel Design and Construction......................................................................... 29
3.1 Design ........................................................................................................................................ 29
3.1.1 Design Requirements .......................................................................................................... 29
3.1.2 Material Analysis ................................................................................................................ 30
3.1.3 Electrical System Design .................................................................................................... 34
3.1.4 Transparent Layer Design ................................................................................................... 36
vi
3.1.5 Structural Layer Design ....................................................................................................... 37
3.1.6 Frame Design....................................................................................................................... 40
3.1.7 Prototype Model .................................................................................................................. 41
3.2 Construction ............................................................................................................................... 41
3.2.1 Bill of Materials ................................................................................................................... 41
3.2.2 Fiberglass Fabrication ......................................................................................................... 42
3.2.3 Housing Fabrication ............................................................................................................ 46
3.2.4 Panel Assembly ................................................................................................................... 47
3.2.5 Lessons Learned .................................................................................................................. 47
Chapter 4 Testing and Analysis Methodology ..................................................................................... 49
4.1 Structural Testing ....................................................................................................................... 49
4.1.1 Testing Objectives ............................................................................................................... 49
4.1.2 Testing Frame ...................................................................................................................... 49
4.1.3 Load Apparatus Design ....................................................................................................... 54
4.1.4 Instrumentation .................................................................................................................... 55
4.2 Finite Element Analysis ............................................................................................................. 58
4.2.1 Prototype Model Development............................................................................................ 58
4.2.2 Ontario Pavement Load Cases............................................................................................. 61
4.2.3 Static and Fatigue Property Analysis................................................................................... 66
4.3 Environmental Testing ............................................................................................................... 69
4.3.1 Material Selection ................................................................................................................ 69
4.3.2 Sample Conditioning and Testing Standards....................................................................... 69
4.3.3 Scaling Resistance ............................................................................................................... 69
4.3.4 Three-Point Flexural Testing ............................................................................................... 72
4.3.5 Friction Testing ................................................................................................................... 75
4.3.6 Optical Testing .................................................................................................................... 76
Chapter 5 Structural Prototype Testing and Analysis .......................................................................... 78
5.1 Structural Testing Results........................................................................................................... 78
5.1.1 Strain Response ................................................................................................................... 79
5.1.2 Deflection Response ............................................................................................................ 81
5.2 Prototype FE Model Development ............................................................................................. 82
5.2.1 Mesh Development .............................................................................................................. 82
vii
5.2.2 Structural Test Simulation .................................................................................................. 82
5.2.3 Model Validation ................................................................................................................ 84
5.2.4 Strain Modelling Deficiencies ............................................................................................ 85
5.2.5 Deflection Modelling .......................................................................................................... 86
5.3 Pavement Load Case Simulation ............................................................................................... 87
5.3.1 Model Development ............................................................................................................ 87
5.3.2 Static and Fatigue Load Simulation Results ....................................................................... 88
5.3.3 Fatigue Life Analysis .......................................................................................................... 97
5.4 Summary of Key Findings ....................................................................................................... 101
Chapter 6 Environmental Material Testing Results and Analysis ..................................................... 103
6.1 Scaling Resistance Testing....................................................................................................... 103
6.2 Three-Point Bending Testing ................................................................................................... 103
6.2.1 Testing Results .................................................................................................................. 103
6.2.2 Material Comparison......................................................................................................... 108
6.2.3 Statistical Analysis ............................................................................................................ 109
6.3 Friction Testing ........................................................................................................................ 111
6.4 Optical Testing ......................................................................................................................... 115
6.4.1 Acrylic Results and Analysis ............................................................................................ 115
6.4.2 Glass Results and Analysis ............................................................................................... 118
6.4.3 Polycarbonate Results and Analysis ................................................................................. 120
6.4.4 Material Comparison......................................................................................................... 122
6.5 Summary of Key Findings ....................................................................................................... 125
Chapter 7 Conclusions and Recommendations .................................................................................. 126
7.1 Conclusions .............................................................................................................................. 126
7.1.1 Design of a Solar Road Panel ........................................................................................... 126
7.1.2 Applicability of Solar Road Panels in Ontario .................................................................. 126
7.2 Recommendations .................................................................................................................... 127
7.2.1 Future Research................................................................................................................. 127
7.2.2 Additional Areas of Study for Solar Road Panels ............................................................. 128
References .......................................................................................................................................... 129
Appendix A Solar Road Panel Prototype Design Drawings .............................................................. 136
Appendix B Solar Road Panel Prototype Bill of Materials................................................................ 146
viii
Appendix C Structural Testing Frame Design Drawings ................................................................... 148
Appendix D Three-Point Bending Apparatus Fixture Drawings ....................................................... 156
Appendix E Prototype Model Validation Results .............................................................................. 164
Appendix F Pavement Load Case Model Validation Results ............................................................ 166
ix
List of Figures
Figure 1-1: Mean daily solar insolation across Canada (NRC, 2012) ................................................... 3
Figure 1-2: Exploded view of a conceptual solar road panel (Northmore & Tighe, 2012a).................. 4
Figure 1-3: Research methodology ........................................................................................................ 5
Figure 2-1: LEF relation to axle group load of AASHTO standard trucks (MTO, 1990) ..................... 9
Figure 2-2: Design depths for frost penetration in Southern Ontario (MTO, 1990) ............................ 10
Figure 2-3: Faun Trackway aircraft landing mat system (FAUN, 2012) ............................................. 11
Figure 2-4: Faun Trackway heavy ground mobility system installation (FAUN, 2012) ..................... 12
Figure 2-5: Golden Gate Bridge - Steel deck bridge............................................................................ 13
Figure 2-6: Aluminum deck bridge (Walbridge & de la Chevrotiere, 2012) ....................................... 14
Figure 2-7: Glass Floors at the CN Tower (left) and Grand Canyon (right) ........................................ 15
Figure 2-8: Typical glass floor panel structure (ISG, 2013) ................................................................ 15
Figure 2-9: Highway 401 test section instrumentation schematic (El-Hakim, 2009) .......................... 16
Figure 2-10: Schematic diagram of confined pavement test apparatus (Perkins, 1999) ...................... 17
Figure 2-11: Crystalline silicon module structure (El Amrani et al., 2007)......................................... 22
Figure 2-12: Solar Roadways prototype solar road panel (Solar Roadways, 2013) ............................ 26
Figure 2-13: TNO SolaRoad prototype panel (Ooms, 2013) ............................................................... 27
Figure 3-1: 125 mm square monocrystalline silicon solar cell (RMSolar, 2012) ................................ 35
Figure 3-2: Schematic solar road panel interconnection scheme (Northmore & Tighe, 2012b) ......... 36
Figure 3-3: Solar road panel prototype transparent layer..................................................................... 38
Figure 3-4: Cell compartment accommodation in the optical layer ..................................................... 39
Figure 3-5: Base plate configuration .................................................................................................... 40
Figure 3-6: Assembled frame member................................................................................................. 40
Figure 3-7: Solar road panel prototype model ..................................................................................... 41
Figure 3-8: Glass transparent layer ...................................................................................................... 41
Figure 3-9: Base layer fiberglass cutting ............................................................................................. 42
Figure 3-10: Base layer fiberglass epoxying........................................................................................ 43
Figure 3-11: Optical layer fiberglass sheet trimming .......................................................................... 43
Figure 3-12: Optical rib cutting with a chop saw................................................................................. 44
Figure 3-13: Fiberglass ribs marked for trimming ............................................................................... 45
Figure 3-14: Optical layer pre-epoxying .............................................................................................. 45
Figure 3-15: Optical layer epoxied joints ............................................................................................ 46
Figure 3-16: Optical layer post-sanding............................................................................................... 46
x
Figure 3-17: Housing frame with edge trim and foam inserts .............................................................. 47
Figure 3-18: Completed structural solar road panel prototype ............................................................. 48
Figure 4-1: Structural testing rig schematic ......................................................................................... 50
Figure 4-2: Deflection response of simulated panel to a centred 4,448-N [1,000-lbf] load ................. 51
Figure 4-3: Structural testing rig base plate.......................................................................................... 52
Figure 4-4: Deflection response of 101.6-mm x 51.2-mm HSS under 6,672-N [1,500-lbf] distributed
load ....................................................................................................................................................... 52
Figure 4-5: Structural testing rig loading ring ...................................................................................... 53
Figure 4-6: Structural testing rig legs ................................................................................................... 53
Figure 4-7: Structural testing rig .......................................................................................................... 54
Figure 4-8: Structural testing air-over-oil system................................................................................. 54
Figure 4-9: Structural testing load application points .......................................................................... 55
Figure 4-10: Structural testing load cell ............................................................................................... 56
Figure 4-11: Instrumentation locations with respect to the optical layer ............................................. 56
Figure 4-12: Rosette and LVDT arrangement ...................................................................................... 57
Figure 4-13: Structural testing setup .................................................................................................... 58
Figure 4-14: FE model load application for fatigue load case.............................................................. 62
Figure 4-15: FE tire load application locations, direction of travel up the page .................................. 62
Figure 4-16: FE subgrade model with complete single bias meshing .................................................. 66
Figure 4-17: S-N curve for cyclic loading performance of short fiber reinforced matrices (Demers,
1998)..................................................................................................................................................... 68
Figure 4-18: Concrete pavement fatigue life equations (Huang, 2004) ............................................... 68
Figure 4-19: Asphalt pavement fatigue life equation (Huang, 2004) ................................................... 69
Figure 4-20: Asphalt and granular rutting life equation (Huang, 2004) ............................................... 69
Figure 4-21: CPATT freeze-thaw cart arrangement ............................................................................. 71
Figure 4-22: Materials 2 Laboratory 500-kg Instron ............................................................................ 72
Figure 4-23: 3-point structural loading base......................................................................................... 73
Figure 4-24: 3-point bending test support noses, base collar, and load nose (left to right) .................. 73
Figure 4-25: 3-point bending test apparatus ......................................................................................... 74
Figure 4-26: Water-saturated friction test specimen ............................................................................ 75
Figure 4-27: Friction testing platform .................................................................................................. 75
Figure 4-28: Varian Cary 5000 UV/VIS/NIR Spectrophotometer ....................................................... 76
xi
Figure 4-29: Sample positioning for transmissivity and reflectivity testing (left to right) .................. 77
Figure 5-1: Structural testing load positions on transparent layer ....................................................... 78
Figure 5-2: Structural testing result orientation ................................................................................... 79
Figure 5-3: 4.44-kN centre load glass strain response ......................................................................... 79
Figure 5-4: 2.22-kN diagonal load glass strain response ..................................................................... 80
Figure 5-5: 2.22-kN centre load base strain response .......................................................................... 80
Figure 5-6: Deflection difference from testing between 2.22-kN and 4.44-kN centre loads ............... 81
Figure 5-7: 2.22-kN diagonal load simulation average material property glass strain contours.......... 83
Figure 5-8: 2.22-kN diagonal load simulation average material property base strain contours .......... 83
Figure 5-9: 2.22-kN diagonal load absolute strain difference for the average material property model
............................................................................................................................................................. 84
Figure 5-10: Average absolute strain error between testing and simulations ...................................... 85
Figure 5-11: Average absolute strain error between testing and simulations for non-boundary
locations ............................................................................................................................................... 85
Figure 5-12: 4.44-kN centre load absolute strain difference in the transparent layer for the rigid
material property model ....................................................................................................................... 86
Figure 5-13: Deflection difference from modelling between 2.22-kN and 4.44-kN centre loads ....... 87
Figure 5-14: Transparent layer stress when subjected to static load .................................................... 89
Figure 5-15: Transparent layer stress contours under static, centre load with HMA base .................. 89
Figure 5-16: Transparent layer stress contours under static, corner load with HMA base .................. 90
Figure 5-17: Transparent layer stress when subjected to fatigue load ................................................. 91
Figure 5-18: Transparent layer stress contours under fatigue, centre load with HMA base ................ 91
Figure 5-19: Base layer stress when subjected to static load ............................................................... 92
Figure 5-20: Base layer stress contours under static, centre load with HMA base .............................. 93
Figure 5-21: Base layer stress when subjected to fatigue load ............................................................ 93
Figure 5-22: Base layer strain when subjected to static load ............................................................... 94
Figure 5-23: Base layer strain contours under static, centre load with HMA base .............................. 95
Figure 5-24: Base layer strain when subjected to fatigue load ............................................................ 95
Figure 5-25: Structural base deflection when subjected to static load ................................................. 96
Figure 5-26: HMA layer deflection contours under static, centre load with HMA base ..................... 96
Figure 5-27: Structural base deflection when subjected to fatigue load .............................................. 97
Figure 5-28: Transparent layer fatigue life endurance limit check ...................................................... 98
xii
Figure 5-29: Base layer fatigue life endurance limit check .................................................................. 98
Figure 5-30: PCC layer fatigue life endurance limit check .................................................................. 99
Figure 5-31: HMA layer maximum horizontal strain ......................................................................... 100
Figure 5-32: HMA base maximum vertical compressive strain in the Granular B layer ................... 100
Figure 5-33: Granular base maximum vertical compressive strain in the Granular B layer .............. 101
Figure 6-1: Salt accumulation during scaling resistance testing ........................................................ 103
Figure 6-2: 3-point bending raw data for G-0-1 ................................................................................. 104
Figure 6-3: 3-point bending raw data for A-0-1 ................................................................................. 105
Figure 6-4: Typical glass 3-point bending test failure (Specimen G-15-US-3) ................................. 106
Figure 6-5: Irregular failures of samples a) G-15-N-1 and b) G-50-N-1 ........................................... 107
Figure 6-6: 3-point bending raw data for P-0-1 .................................................................................. 108
Figure 6-7: Comparative material flexural stress response to displacement ...................................... 109
Figure 6-8: Polycarbonate flexural modulus results ........................................................................... 110
Figure 6-9: Polycarbonate flexural modulus results, excluding '0' cycle data.................................... 111
Figure 6-10: Friction testing results ................................................................................................... 112
Figure 6-11: Acrylic friction results ................................................................................................... 113
Figure 6-12: Glass friction results ...................................................................................................... 114
Figure 6-13: Polycarbonate friction results ........................................................................................ 114
Figure 6-14: Acrylic transmission results........................................................................................... 116
Figure 6-15: Acrylic reflectivity results ............................................................................................. 117
Figure 6-16: Acrylic absorptivity results ............................................................................................ 117
Figure 6-17: Glass transmissivity results ........................................................................................... 119
Figure 6-18: Glass reflectivity results ................................................................................................ 119
Figure 6-19: Glass absorptivity data................................................................................................... 120
Figure 6-20: Polycarbonate transmissivity results.............................................................................. 121
Figure 6-21: Polycarbonate reflectivity results .................................................................................. 122
Figure 6-22: Polycarbonate absorptivity results ................................................................................. 123
Figure 6-23: Average transmissivity comparison ............................................................................... 123
Figure 6-24: Average reflectivity comparison.................................................................................... 124
Figure 6-25: Average absorptivity comparison .................................................................................. 125
xiii
List of Tables
Table 2-1: Simplified FE models of pavement structures .................................................................... 19
Table 3-1: Structural layer material mechanical properties ................................................................. 31
Table 3-2: Transparent layer material mechanical properties .............................................................. 33
Table 3-3: Maximum transparent layer bending stress as a function of glass pane thickness ............. 37
Table 4-1: Prototype mesh validation parameters................................................................................ 60
Table 4-2: Prototype material property ranges and analysis categories ............................................... 61
Table 4-3: Tire loading conditions ....................................................................................................... 61
Table 4-4: FE structural base layer thicknesses (ARA, 2011) ............................................................. 63
Table 4-5: FE structural base material properties ................................................................................ 64
Table 4-6: FE boundary conditions per load application model .......................................................... 65
Table 4-7: FE load case model validation parameters ......................................................................... 65
Table 4-8: Standard procedures for environmental testing .................................................................. 70
Table 4-9: 3-point bending specimen dimensions ............................................................................... 74
Table 5-1: Prototype panel mesh development results ........................................................................ 82
Table 5-2: Deflection result comparison .............................................................................................. 87
Table 5-3: Pavement load case model development results................................................................. 88
Table 6-1: Acrylic 3-point bending results ........................................................................................ 105
Table 6-2: Glass 3-point bending initial results ................................................................................. 106
Table 6-3: Glass 3-point bending final results ................................................................................... 107
Table 6-4: Polycarbonate 3-point bending results ............................................................................. 108
Table 6-5: 3-point bending ANOVA results ...................................................................................... 109
Table 6-6: Polycarbonate 3-point bending regression analysis.......................................................... 111
Table 6-7: Friction testing results ...................................................................................................... 112
Table 6-8: Friction ANOVA results................................................................................................... 113
Table 6-9: Friction linear regression results ...................................................................................... 115
Table 6-10: Acrylic optical testing results ......................................................................................... 115
Table 6-11: Acrylic optical ANOVA results ..................................................................................... 116
Table 6-12: Glass optical testing results ............................................................................................ 118
Table 6-13: Glass optical ANOVA results ........................................................................................ 118
Table 6-14: Polycarbonate optical testing results .............................................................................. 120
Table 6-15: Polycarbonate optical ANOVA results........................................................................... 121
xiv
Chapter 1
Introduction
Sustainability has become a leading goal of global infrastructure developments over the last several
years due to increased understanding of the impacts that engineering decisions have on society and
the environment. The leading definition for sustainability was created by the Brundtland Commission
and states that sustainable development is “development that meets the needs of the present without
compromising the ability of future generations to meet their own needs” (World Commission on
Environment and Development, 1987). This is traditionally accomplished through optimizing the
balance between design economic, environmental, and societal impacts while still delivering
infrastructure that meets the original design requirements (Adams, 2006). While it is often possible to
make traditional designs more sustainable by these metrics through evolutionary modifications,
revolutionary redesign of infrastructure from first principles allows for rediscovery of its full
potential.
This holds true within the realm of pavement infrastructure, where the majority of the materials
currently used in roads and highways have not changed over the last few decades. Recent pushes have
been made to make pavements more sustainable in a number of ways. One such way has been looking
at using more recycled materials in pavements, be it glass, asphalt shingles, recycled asphalt
pavements, recycled rubber tires, or other additives in asphalt mixes or recycled concrete and other
additives in concrete mixes. Other focuses for increasing pavement sustainability have included
reducing the processing temperatures required for asphalt and creating porous pavements which offset
the need for stormwater management. While these technologies combined help make pavements
incrementally more sustainable, the fact of the matter is they all look at pavement design through the
same lens as designers have for the last few decades.
This becomes a problem within the transportation industry due to how large transportation’s
impact on the environment is as a whole. It is well established that for our society to be more
sustainable there must be substantial changes in our energy consumption and greenhouse gas
emission patterns, and transportation is the cause of 19% of global energy consumption (US EIA,
2011) and 23% of global CO2 emissions (IEA, 2011). While most of this is a direct result of
transportation modes and not the infrastructure that satisfies these modes, the infrastructure is always
designed with the mode energy consumption and emissions in mind. For roads there are direct
correlations on the impact that roughness (Costello, Bargh, Henning, & Hendry, 2013) and grade
1
have on vehicle fuel consumption, traffic intersections are naturally designed to minimize delays and
as a result vehicle idling time, and the same applies to minimizing delays when performing road
maintenance. All of these factors are the responsibility of regional transportation authorities, so
solutions that minimize their energy or carbon footprint in a major way are highly sought after
solutions.
One way that many jurisdictions have started making their transportation corridors more
sustainable is by incorporating renewable energy projects into their territory. While using
transportation corridors for utility purposes other than transportation is not a novel idea, the interest in
incorporating renewable energy projects has only been building slowly over the last few decades.
Most of the longer term projects exist in Europe, where they have constructed solar railway tunnels in
Belgium (Enfiinty, 2011), solar highway tunnels in Italy, and solar highway tunnels and noise barriers
in Germany (US DoT, 2012). Recently in the United States, many state Departments of
Transportation have investigated the benefits of adding solar panels and wind turbines and using
highway lands to grow crops for conversion to biofuels (US DoT, 2012). These projects demonstrate
the interoperability of the right-of-way with renewable energy projects, though they for the most part
ignore that roads have the capability to be a multifunctional element of our infrastructure.
2
Figure 1-1: Mean daily solar insolation across Canada (NRC, 2012)
A few different methods have been researched to attempt to generate electricity from this
resource. The first was done by turning traditional asphalt pavements into solar thermal collectors
(Bijsterveld, 2001). While this was a logical choice due to the known heat retention of asphalt
pavements, maintenance of such structures is very challenging as the piping that runs through the
asphalt disrupts traditional maintenance approaches. More current research is being done on using
thermoelectric generators to directly extract electricity from the thermal gradient in asphalt pavement
structures (G. Wu & Yu, 2012), though this research is still in its infancy and has very low conversion
efficiency. The most promising technology in this field also happens to be the most innovative and is
the only one to completely disregard traditional paving practices: the development of solar road
panels.
Solar road panels are a new technology being designed to serve as the top layer of pavement
infrastructure. The key benefit from this approach is that it is a modular structure which incorporates
photovoltaic cells which directly convert the incident solar energy into electricity. This electricity
3
could have a number of uses, such as powering street lights, charging electric vehicles, or being sold
to local distribution companies. While the greatest influence of these panels would come from
replacing highway pavements, many lower speed and volume applications are currently being
considered as ideal locations for installation of these panels (Northmore & Tighe, 2012a).
The panels are a three layer composite structure that consists of transparent, optical, and base
layers as shown in Figure 1-2 (Northmore & Tighe, 2012a). The transparent layer handles direct
interaction with vehicles and allows solar radiation to pass through to the optical layer. The optical
layer transfers the load on the transparent layer to the base layer by directing it around the embedded
solar cells within the structural cutouts. Lastly the base layer transfers load to the structured base
beneath the panel (Northmore & Tighe, 2012b).
Figure 1-2: Exploded view of a conceptual solar road panel (Northmore & Tighe, 2012a)
Overall the research on solar road panels is also in its infancy. There are only two organizations
working towards the development of these panels: Solar Roadways in the United States and TNO in
the Netherlands. The prototypes being developed by these groups are being designed with less focus
on optimizing the design of a composite pavement structure to act as a solar road module in a variety
of structural scenarios and neither considers how to mitigate the issues of salting and freeze thaw
effects experienced during a typical Canadian winter.
This research involved the design, construction, and testing of solar road panel prototypes in a
laboratory environment at the Centre for Pavement and Transportation Technology (CPATT) at the
University of Waterloo. Additional testing work was performed through the Civil Engineering
5
Structures Laboratory, the Solar Thermal Research Laboratory, and the Mechanical Materials
Laboratory at the University of Waterloo.
Laboratory testing of the solar road panel prototypes was done to determine the physical response
of the panel to loading and to replicate the stresses of Canadian environmental conditions. This work
was accomplished to properly assess the surface, structural, and optical characteristics of solar road
panels and candidate transparent layer materials.
1. Determination of the design requirements and material selection required for the optimal
solar road panel for use in Canadian conditions.
2. Design a prototype solar road panel to be used for laboratory testing and as a platform for
future related research.
Another contribution of this research involves guidance on the structural performance of non-
traditional pavement materials as a form of subgrade or pavement reinforcement. Traditional work in
this field has examined these materials as short-term reinforcement for poor subgrade soils exposed to
heavy vehicle traffic while this research examines how these materials could perform over the long
term on strong subgrades or on typical paved bases for city or municipal traffic loads. The work done
herein is also more focused on finite element modeling of these systems while previous work was
completed primarily through laboratory experiments.
The third major contribution of this work is in determining the performance of transparent
materials as they are subjected to scaling and freeze-thaw cycling. This research is novel in that it
assesses structural, textural, and optical properties of the materials as they are conditioned, a
6
combination that is seldom investigated. Potential applications of this work include material selection
for automotive applications, road lighting infrastructure, and building systems exposed to salt based
corrosion.
Chapter One: An introduction to solar road panels and the scope, hypothesis, and objectives for the
entire thesis.
Chapter Two: An analysis of existing literature on design aspects for pavement and conventional
solar modules and how the two topics converge towards designing solar road panels. This section also
defines the gaps in existing knowledge regarding solar road panel design.
Chapter Three: The detailed design and construction processes for developing the prototype solar
road panel used in testing throughout the rest of this thesis. Also included is a section on lessons
learned through design and construction for the use of future researchers on this topic.
Chapter Four: Development of the testing mechanisms for solar road panel prototypes and materials.
This section discusses the structural and environmental testing conditions that were developed to
execute this research. Emphasis is placed on the development of the standards for both forms of
testing and the design and construction of the fixture used for structural testing.
Chapter Five: Structural testing of the solar road panel prototype and numerical analysis of solar road
panel designs on simulated structured bases. This section analyses the structural feasibility of solar
road panels and draws conclusions about material selection, solar road panel design, and the required
base for panel installation.
Chapter Six: Environmental testing of transparent materials. This section draws initial conclusions on
the applicability of various polymer and glass materials as a transparent layer material for solar road
panels installed in a Canadian climate with typical winter road maintenance practices. This was done
through analysis of their mechanical, optical, and surface properties.
Chapter Seven: The conclusions drawn throughout the thesis are summarized in this chapter and
compiled to define the overall structural feasibility of solar road panels. Recommendations for further
research in the field of solar road panels are also presented.
7
Chapter 2
Design Elements
In order to develop an understanding of the requirements of a solar road panel and the justification for
this research, a thorough review of pavement and solar module design elements is required. This will
highlight the contrast between the designs of the two structures and identify the justification for
thorough research on the design of a combined structure.
The main factors taken into pavement design are the desired design life, traffic loading,
environmental conditions, subgrade soil, drainage, performance of local similar pavements, and
locational constraints (TAC, 2012). Typical pavement design lives for Ontario, marking the point
where the pavement structure needs rehabilitation due to inadequate performance, are between 10 and
20 years for highways depending on the pavement materials used (MTO, 1990). The factors with the
highest impact on the deterioration of a pavement structure are traffic loading and environmental
conditions, as these define the stresses applied to the structure that need to be supported.
Traffic loading is measured in two ways to account for the varying vehicle types that are expected
to travel on the pavement structure. The first measure for this is the annual average daily traffic with
the given percentage of that traffic that is trucks. This method has demonstrated correlations to
pavement performance through older design methods, however newer methodologies require the
8
number of equivalent single axle loads (EASLs) that will travel over the given section (MTO, 1990).
This method converts each individual vehicle into a number of EASLs, which each represent an 80-
kN axle load with dual tires, through a load equivalency factor (LEF) that approximately follows a
fourth power law based on vehicle weight. This equivalency is demonstrated in Figure 2-1 for single,
tandem, and tridem axled vehicles as described by the American Association of State Highway and
Transportation Officials (AASHTO).
Figure 2-1: LEF relation to axle group load of AASHTO standard trucks (MTO, 1990)
The environmental conditions that have the greatest impact on pavement degradation are
temperature and moisture. In Ontario the range of temperatures during the year causes a variety of
issues, as low temperatures in the winter cause the ground to freeze and heave while high
temperatures during the summer weaken asphalt structures allowing rutting and other failures.
Moisture also causes issues because of its ability to infiltrate pavement structures and then freeze and
thaw cyclically through the winter or wear away the surface pavement material. In designing roads
this is taken into consideration by accounting for the expected frost penetration depth for the region.
A map showing the frost penetration design standards for Southern Ontario is shown in Figure 2-2
(MTO, 1990).
Another important design factor for pavement engineers is the surface texture of their structure.
In order for vehicles to safely traverse the structure, the surface must provide ample traction for tires
to grip. With asphalt pavements there is no extra work needed to achieve this, though concrete
structures require additional texturing once they have been laid. This is typically achieved through
dragging a rough material across the surface while the material is curing or mechanically applying a
9
texture after the concrete has hardened. The surface texture is impacted in the short term by weather
conditions and the long term by vehicle traffic. Water and ice accumulation on pavement can make
the surface too slippery to traverse if drainage is inadequate, while vehicle traffic will wears the
surface down and requires maintenance to restore the texture (TAC, 2012).
Figure 2-2: Design depths for frost penetration in Southern Ontario (MTO, 1990)
Between the two, landing mats have a more stringent set of design requirements. This is largely
due to the higher loadings that these mats are subjected to; the extreme weight of military transport
10
aircraft and the high temperature jet exhaust from fighter aircraft. Another important design factor is
that these panels must be light weight, so that they can be rapidly deployed. To accomplish this, the
US Air Force Research Laboratory has set out guidelines that the next generation of landing mats
should weigh approximately 14-kg/m2, handle all aircraft loadings when installed on poor subgrades,
provide adequate performance at a large range of operating temperatures, withstand all shear stresses
from aircraft braking, and have an adequate anchoring system to firmly mount the panels (Foster &
Anderson, 2003).
The most widely used early landing mats were the British Class 60 Trackway, AM-2 (Rollings,
1975), and XM19 landing mats (US DoD, 1987). All of these mats are constructed from aluminum
alloys, which have the benefit of a very high strength to weight ratio, and are typically able to
withstand 1,000 aircraft movements when placed on subgrades with a strength from 1.3 to 8.0 CBR
(Ulery & Wolf, 1971). Given the rapid pace of military operations this represents a short service life,
which then accentuates that the extruded aluminum mats are very expensive when compared to other
strong materials. Modern advancements in material technologies have led multiple companies to
develop their own landing mat systems, though all are still based on high strength extrudable alloys.
An example of one such system is shown in Figure 2-3.
Figure 2-3: Faun Trackway aircraft landing mat system (FAUN, 2012)
Road mats are designed for a much lower strength operation than landing mats are, so many of
the materials used in them are lower strength as well. Naturally aluminum and other metal alloys still
provide strong performance in this application, but the lower loading from vehicles allows other less
11
expensive materials to be considered. The typical application of these mats is the reinforcement of
poor quality subgrades for the transport of military vehicles, and the mat structures are typically
fiberglass reinforced polymer (FRP) hexagonal mats, plastic hexagonal mats, aluminum meshes, and
plastic meshes (Webster & Tingle, 1998). An example of one such system being installed can be
found in Figure 2-4.
Figure 2-4: Faun Trackway heavy ground mobility system installation (FAUN, 2012)
The testing performed on road mats is largely done on the lowest strength subgrades that the
materials would be required to reinforce. Testing of road mats installed over sand and driven on by a
4,536-kg [5-ton] military truck showed that the plastic mesh mats are prone to extreme rutting,
developing 84-mm [3.3-in] ruts within 20 vehicle movements. Under the same conditions the plastic
hexagonal, FRP, and aluminum hexagonal mats developed ruts of 71-mm [2.8-in], 46-mm [1.8-in],
and 28-mm [1.1-in] after 5,000 vehicle movements, demonstrating that these materials are much more
suited to the application (Webster & Tingle, 1998).
An additional study found that the design of the road mat structure was very important to the
performance of the panel. This was done again on a sand subgrade with a fully loaded six-wheel truck
supplying the testing load. The majority of the designs using high-density polyethylene (HDPE) as
the structural material experienced extensive rutting and structural damage over 2,000 vehicle
movements, though one panel using this same HDPE material only experienced minor delamination
in the same trials. This study also demonstrated that multiply FRP and aluminum structures provide
12
adequate performance on poor subgrades, though at a higher cost and weight than the HDPE solutions
(Rushing & Tingle, 2009).
One of the longest standing substitutes to concrete bridge deck construction is steel plate bridge
decks. Being an excellent structural material, steel shows great performance in terms of loading
capacity and stiffness when used in bridge deck applications. Examples of this type of bridge decking
include the George Washington Bridge; the Lions Gate Bridge; and the Golden Gate Bridge, as
shown in Figure 2-5 (ASCE, 2013).
Various high strength aluminum alloys have also been used as bridge deck materials, though
predominantly in pedestrian, portable, or expedient repair applications due to being a lower strength
and higher cost material than more traditional alternatives. When expedient construction or bridge
weight concerns are important, extruded aluminum has proven to be a suitable alternative from a
strength and durability perspective. An example of an aluminum deck vehicle bridge can be found in
Figure 2-6 (Walbridge & de la Chevrotiere, 2012).
13
Recent research demonstrates that extruded and pulltruded FRP panels are well positioned to
withstand vehicle loading as an open supported bridge deck. Work completed at the University of
Kentucky demonstrated that when loading FRP composite bridge deck panels as per AASHTO MS
22.5 (HS25) specified wheel loading, the typical safety factor against failure was between 3 and 8 for
a variety of commercially available products. The main failure modes from this testing were
debonding and flexure shear of the specimens (Alagusundaramoorthy, Harik, & Choo, 2006).
Typically, glass flooring is used in indoor environments as a prominent design feature. Many
cases have seen installations in walkways, bridges, staircases, and cantilevered floor sections. The
most notable of these are often in observation decks, such as the one in the CN Tower in Toronto
(Torontoist, 2013), or on scenic walkways, like the one that loops over a section of the Grand Canyon
in Arizona (Grand Canyon National Park, 2013). Both of these are shown in Figure 2-7.
One of the main concerns in the design of glass floors is ensuring for a large safety factor in the
design. A fear of heights is common, which can make walking over a glass surface a very nerve
wracking experience. To design these sections so that they are safe enough for people to walk on, the
standard practice is to make the glass walking surface a three-pane glass structure (Alsop & Saunders,
14
1999). The three panes are laminated together using a typical resin and the glass is thick enough so
that if any one of the three panes were to break the other two would be able to support the design
load. This structure is demonstrated in Figure 2-8.
Figure 2-7: Glass Floors at the CN Tower (left) and Grand Canyon (right)
The main reason why the majority of these installations are indoors is because the rubber-glass
friction coefficient drops substantially when the surface is wet. Even at outdoor installations, people
are typically not allowed to walk over the surface during inclement weather due to this reduced
friction. This can be accommodated for by adding a texture to the surface, however this also reduces
the transparency of the floor and reduces some of the architectural merit of the installation.
15
2.1.3 Structural Testing of Pavement
A wide variety of approaches are used to determine the structural capacity of pavements. Of most
importance to this research is how controlled tests are performed on structural pavement sections in a
laboratory or research environment.
The objective of structural testing is to determine how a structure will perform under various
loading conditions that best emulate the real loading case for the structure. In terms of pavements, the
most ideal test scenario allows vehicles to drive over a test section that has been instrumented to
measure the strain caused within the structure. This is typically accomplished through two
approaches, instrumenting new pavement sections of existing road infrastructure and constructing
dedicated pavement test tracks. Both types of facilities allow researchers to better understand how
specific pavement materials perform in the environment of the site; the key advantage to instrumented
roads being that the site is often a better replica of that region’s conditions than a test track while a
test track has the advantage of a more controlled load scenario than open roads have. Due to the cost
of the infrastructure required for both, these are often only implemented when determining how a
new-to-the-region pavement structure will perform. An example of and instrumented road section can
be found in Figure 2-9.
Figure 2-9: Highway 401 test section instrumentation schematic (El-Hakim, 2009)
Full pavement structure testing has also been completed in lab environments using confined
pavement structures at a much smaller scale than test tracks, as shown in Figure 2-10. The concept
being that when a load is applied to a pavement structure, only a portion of the total base and
16
subgrade materials are deformed. If a confined specimen is able to contain enough material so that the
deformation only occurs within a section of the contained material then real world results can be
emulated in a lab. This is most commonly done in the analysis of geosynthetic reinforcing layers in
flexible pavements. Typical configurations involve the design of a large steel or concrete box that
contains between 0.25 and 4.5 cubic metres of pavement structural materials depending on the
designed testing regime (Tingle & Jersey, 2005). Loads are typically applied through hydraulic
actuators at low frequency in order to simulate vehicle traffic (Tingle & Jersey, 2005). Larger scale
versions of this testing have been done in concrete pits, which would also allow for vehicles to drive
over the test installation (Pokharel et al., 2011). The main disadvantages of this testing approach are
the cost of developing such a setup and the limited size of the test specimens.
Figure 2-10: Schematic diagram of confined pavement test apparatus (Perkins, 1999)
In lieu of being able to do testing of full pavement structures, the normal method for structurally
testing pavement materials is to test samples of individual materials to measure comparative
properties. These tests often allow for samples to be tested under a wide variety of simulated
conditions in a controlled laboratory environment. Quite often the performance characteristics
determined from these tests can be used as an input for mechanistic pavement design systems; while
the lab results do not directly imply performance in the field, correlations have been developed that
make these results valuable. These tests can include dynamic modulus, fatigue beam, and moisture
susceptibility tests for asphalt samples; compressive strength, flexural strength, and durability testing
for concrete samples; and soundness, absorption, abrasion, and gradation testing for aggregate
samples.
17
2.1.4 Numerical Analysis in Pavement Design
While the majority of pavement design is accomplished through empirical and mechanistic relations,
specialized applications occasionally make use of finite element analysis (FEA) to determine the
strain development that is expected within the pavement and subgrade materials for a given site.
There are a number of reasons for why FEA is not widely used including the variability of pavement
and soil material properties, the expense of viscoelastic material modeling, the variability introduced
through constructing the site, and the ability to adequately model all pavement design factors within
FEA software. As a result of this, FEA tends to be used in niche applications of pavement design.
Often this can be helpful if dealing with difficult or unknown conditions or when trying to develop a
better mechanistic understanding of the failure mechanisms of a pavement structure.
Some of these cases though do provide validated, simplified models for the use in approximating
pavement structural performance. The first major assumption is that the materials behave only in the
elastic regime, which is a valid assumption for determining a static response from pavement loading
but does not account for dynamic effects. These models also often assume homogenous material
properties within a given layer, which assumes a lot of ideal construction practices and material
specifications. A sampling of these models has been summarized below in Table 2-1.
Additionally, work has been completed previously at CPATT on the FE modeling of steel
reinforcing plates for weak subgrades. This work, using the ABAQUS software package, assessed
various thicknesses of steel plate reinforcement for a sliding scale of effective subgrade modulus and
soil bearing capacity. The output of this work demonstrated the type of steel reinforcement required
for various base conditions and, most importantly for this research, demonstrated the applicability of
the ABAQUS software package to pavement structure FE analysis (Mak, 2012).
Single crystalline silicon solar cells are thin wafers of a single crystalline silicon ingot that have
been cut to a prescribed size and doped on the top and bottom of the cell to promote electron
transmission across the wafer. The top of the cell is heavily doped with a negative charge while the
bottom of the cell is lightly doped with a positive charge. When photons enter the cell they travel until
their energy displaces an electron, which then diffuses to the negatively charged top of the cell while
the resulting hole drifts to the base. This release generates the electrical potential across the cell while
repeated incident photons generate the electron flow between the layers. When the cells are linked
together in a circuit, the electron flow passes from cell to cell and by combining enough cells this
generates usable power (Luque & Hegedus, 2003; Neamen, 2003).
Typical solar cells used in utility grade solar modules have a high individual current capacity but
small voltage potential. To accommodate this, solar cells within a module are linked together in series
19
to allow the voltage to build while keeping the current constant. Depending on the size of the module
there can be multiple strings of series solar cells linked in parallel to achieve the desired power output
characteristics. In typical module fabrication there are 72 cells combined in two parallel strings of 36
(Luque & Hegedus, 2003).
The energy conversion potential within the solar cell is highly dependent on the materials used to
make the cell. Each material type has a specific bandgap potential that can be achieved, and this is the
potential that must be exceeded by an incident photon to release an electron (Neamen, 2003). Without
achieving this minimum requirement no energy generation can occur and all of the photon’s energy is
lost as heat. Similarly, any excess energy from the photon that is not used to release an electron will
be lost as heat within the solar cell. For a silicon solar cell, this bandgap energy requirement is 1.1
eV; a level of energy that exists in the upper infrared region of radiation allowing the solar cells to
generate current from that point, through the optical spectrum, and onwards though with additional
thermal losses (Luque & Hegedus, 2003).
Thermal cycling is a large issue in the design of solar modules for a number of reasons. First it is
important to ensure that the module will withstand the thermal expanding and contracting that the
materials will endure over the course of a year (Askeland & Phule, 2006). This is especially important
as the crystalline solar cells used within the modules are very brittle and cannot endure a large lateral
tensile stress (Luque & Hegedus, 2003). Another issue with thermal performance is that photovoltaic
solar cells lose efficiency as their temperature increases on the order of 0.5% per degree centigrade
(Skoplaki & Palyvos, 2009). This is because temperature impacts how the semiconducting cell works;
it increases the developed short circuit current due to a decrease in the bandgap but also lowers the
open circuit voltage due to variation in the intrinsic carrier concentration (Neamen, 2003). It is
therefore important to make careful material selections when designing a solar module to account for
thermal expansion and the quick removal of the excess thermalization losses from the PV conversion.
20
Mechanical loading plays an important role in the design of solar modules. The majority of
typical solar modules are installed on a stationary or solar-tracking structural harness to allow for
optimum exposure of the panels to solar beam radiation. The nature of these harnesses leaves the
solar modules exposed on the top and bottom sides to wind and snow loading, depending on the
climate they are installed in. Normal condition loading, including from environmental factors, is not
an issue for most solar modules however impact or otherwise-transient loading can create issues that
cause premature failure in the solar modules (Chung, Chang, & Liu, 2008). These need to be
accounted for in the design by ensuring that the panel is sufficiently rigid to accept loads from hail,
bird collisions, and like loadings appropriately.
UV radiation poses a challenge to the material selection for solar modules. While most materials
are sufficiently protected for short term exterior use, solar modules are designed for an exterior use
lifetime in excess of twenty years. Combine this with that the solar module must be oriented towards
incoming solar radiation to be effective and this results in a lot of exposure to UV radiation sources.
The materials used in solar modules must be appropriately selected so that they do not degrade during
these operational conditions (Holley Jr, Agro, Galica, & Yorgensen, 1996).
Humidity and atmospheric conditions pose another large design challenge for solar module
designers. Due to the electronics encased within the module there must be ample weatherproofing to
make sure that moisture does not get into it and degrade the system. The same sort of degradation can
come from particulate matter in the air etching the glass surface, or wearing away at the housing.
Depending on the mounting strategy used, moisture infiltration can also come from direct contact
below the panel. This is especially the case for ground or roof mounted units. Therefore proper
materials should be chosen so that there will not be excessive moisture seeping through the panel,
weatherproofing completed for the same effect, and that there will be effective resistance against
other atmospheric particulate matter (Jorgensen et al., 2006).
The glass texturing can be done with a number of performance objectives in mind. While a lot of
research focuses primarily on the light entrapment, surface texturing can also increase the overall
efficiency of a panel throughout the day and concentrate radiation to the useful portions of the solar
22
cell. Through applying structured textures, it is possible to improve the performance of solar glass to
light that approaches the panel at a large incident angle. In one case it was found that a 35% to 55%
efficiency improvement was found through applying a triangular or sinusoidal surface texture to the
glass when light was at an incident angle of 80° to the normal of the surface. This greatly impacts one
of the challenges in solar module design, which is that the best performance is achieved with an
incident angle near zero and that performance tapers on either side of this mark. While this would be
of little use for solar modules that track the sun’s apparent movement, the majority of solar module
installations are stationary and this technology would have a great positive impact on their
performance (Sánchez-Illescas et al., 2008).
While the majority of surface area on a solar module is exposed to the PV solar cell, there are a
few areas that are not; gaps between the cells and the bus-bars and fingers on the solar cell. Any light
that is incident on these areas is not used to excite electrons, so focusing light away from these areas
and onto the main semiconducting cell has efficiency benefits. It has been demonstrated that by
specifically texturing the glass above these areas that the shadow losses in a solar module can be
reduced by 50%, though due to the constraints of ray tracing this is a benefit that can only be
achieved through modules that have full or partial solar tracking (Bergamin & Sammaraee, 2010).
While the texture of the glass has an impact on the reflectiveness of the module, another
important factor is that an anti-reflective coating is often applied to the surface of the glass exposed to
the atmosphere. These coatings are typically layers of SiO2 or Si3N4 which are thinly deposited onto
the glass surface and produce a refractive index gradient at the coating-glass transition that maximizes
transmission. Through most manufacturers this form of textured, coated glass is referred to as solar
glass, as these specific properties are mostly coveted for only solar panel applications (Deubener et
al., 2009).
As noted earlier, the glass is an important element from a structural perspective. With the EVA
and Tedlar being comparably flexible materials to glass and a minimal aluminum frame with respect
to the size of the panel, the glass is required to impart its rigidity to the rest of the panel. With the
environmental loads that solar modules are expected to endure this is not much of an issue, as glass is
a very strong and rigid material to low pressure loading, impact loads can be very detrimental to the
performance of solar modules. Most consumer solar modules are considered with relatively thin glass,
though aftermarket or hobbyist panels are made with thicker layers to provide greater strength though
at a higher cost (Luque & Hegedus, 2003).
23
2.2.3.2 Encapsulating Layers
The majority of solar modules use an EVA compound as the laminating material. This layer
performs a number of tasks for the solar module including protecting the electronics in the module
from moisture, blocking UV radiation from the cells and backsheet, buffering the solar cells from
loads placed on the glass or backsheet, and adhering the structure together (El Amrani et al., 2007). In
mass produced modules, this layer is installed as two plastic layers, one on either side of the solar
cells, so that when the module is put through lamination the EVA melts and forms around the cells
and adheres the panel together (Luque & Hegedus, 2003). In hobbyist projects this can also be
completed using a poured resin, as laminating a solar module without the proper equipment can be
challenging.
The laminating material must also meet a number of specifications, including minimum standards
of electrical, optical, mechanical, and chemical performance (Lange, Luo, Polo, & Zahnd, 2011).
Equally important to this are the reliability and cost performance of the laminate through the entire
assembly process of the solar module (Agro & Tucker, 2004). The specific material choice is made
by the manufacturer and can vary depending on the specific properties that they are looking for; this
can include benchmarks on curing time, moisture transport properties, and overall assembly
procedure. Other materials often used as laminates are silicone, polyvinyl butyral, thermoplastic
polyurethane, and ionomer (Agro & Tucker, 2004; Rose, Jester, & Bunea, 2008).
24
2.2.3.4 Backsheet Layer
The last component in a solar module is the Tedlar backsheet layer. This layer is responsible for
further protecting the solar cells, particularly from any conditions applied to the back side of the
panel, as well as ensuring the stability of the whole module. Tedlar is one of the leading materials
used as a backsheet, though other options such as ethylene propylene diene monomers have been
considered for use as new combinations of materials can help streamline the assembly process by
eliminating need for the second EVA layer (Kempe & Thapa, 2008). Typical properties desired in a
backsheet are high UV resistance and minimal moisture transport as this layer is directly exposed to
the environment and must maintain protection for the EVA and electronics layers (El Amrani et al.,
2007).
2.3.1 State-of-the-Practice
As identified in Section 1.1 there are two organizations working on the development of solar road
panels: Solar Roadways and TNO.
The panel Solar Roadways has developed is a 3.66-m [12-ft] square panel and consists of a
transparent surface layer made from textured glass, an electronics layer that houses circuit boards
supplying the electrical links between the solar cells and other systems, and packaging which
weatherproofs the panel. In addition to simply having solar cells strung together in this system, the
25
Solar Roadways prototype includes LED lighting built into the cell compartments to provide
messaging and lane markings from the road surface. It also includes resistance heaters to deal with
any potential snow or ice issues that may develop on the surface during winter operation. An image of
the first prototype from Solar Roadways can be found in Figure 2-12 (Solar Roadways, 2013).
Figure 2-12: Solar Roadways prototype solar road panel (Solar Roadways, 2013)
2.3.1.2 TNO
TNO, the Netherlands Organization for Applied Scientific Research, is based in Delft and is a
knowledge organization used by local companies and government agencies to develop innovative
technologies to solve technical challenges in the Netherlands and abroad. Solar road panel
development is only a small part of this organization’s energy efficiency systems research. They’ve
partnered with Ooms Avenhon Groep, a civil infrastructure engineering group, and Imtech, an
electrical and mechanical engineering consulting group, to develop solar road panels within the
context of the Dutch environment (TNO, 2013).
The first design of TNO’s SolaRoad concept is being developed for use in cycling paths;
infrastructure that the Netherlands has an abundance of. The panels they are building are 1.5-m by
2.5-m and consist of a glass surface layer, crystalline silicon solar cells, and a concrete block housing.
The trail installation in North-Holland was expected to be installed in the summer of 2012, but to date
this installation has not been completed. An image of a SolaRoad prototype is found in Figure 2-13
(TNO, 2013).
26
Figure 2-13: TNO SolaRoad prototype panel (Ooms, 2013)
27
In addition, very little consideration has been made for how these panels will operate during
winter conditions. This is reasonable design for TNO, seeing as the Netherlands only sees 25 snowy
days on average per year (KNMI, 2013) compared to 64 in Waterloo, Ontario (Environment Canada,
2013) and much higher through other portions of Canada, but is a lacking element in the design from
Solar Roadways.
As identified earlier, the design from Solar Roadways does include a resistance heater in the
panel to melt any snow or ice that may accumulate however this is not a total solution to winter
climate and maintenance issues. It solves the issue of damaging the panels through snow plowing
operations, but salting is still a major part of winter maintenance operations. Even if it was found that
the surface of the solar road panels does not need to be salted, large volumes of salt would be brought
onto the solar road panel surface by cars that have been exposed to salt on other roads. Salting
operations are a large attributor to the damage of civil infrastructure, so it is expected that this needs
to be studied for solar road panel performance as well.
28
Chapter 3
Solar Road Panel Design and Construction
With a thorough understanding of the design of pavement structures and traditional solar panels
having been established, this section details the design and construction processes used for
developing the solar road panel prototypes used in later testing and analysis sections of this thesis.
3.1 Design
The design process for the solar road panel prototype was completed across three main phases; design
requirement development, material analysis, and component system design. The material analysis
focused on determining the materials which are best suited for use in a solar road panel while the
system design takes the materials and the design concept and produces a functioning, manufacturable
prototype.
The second major structural requirement is that the surface of the panel must provide adequate
friction for vehicles to safely travel across it. This is a major challenge as the surface must also be
transparent enough to allow solar radiation to reach the solar cells embedded within the structural
layers and any sort of texturing could impair light transmission if not designed properly (Northmore
& Tighe, 2012b).
Since modern high-efficiency solar cells are made from brittle silicon wafers, a solar road panel
design must allow load to be bypassed around the cells. This can be accommodated by cantilevering
29
the transparent layer over the solar cell compartments, however the transparent layer would then need
to be strong enough to not deflect onto and load the solar cells (Northmore & Tighe, 2012b).
The panel also needs to be designed in such a way that it is weatherproof. This is important for
both structural and electrical reasons as water and contaminant penetration would degrade the
structural integrity of the composite panel and damage the embedded electronics (Northmore &
Tighe, 2012b).
Shading is a major issue for photovoltaic panels because cell area that is not exposed to radiation
is not able to operate as a part of the electrical circuit and this quickly degrades the performance of
the panels. Since the design requires the solar cells to be recessed from the transparent layer, care
needs to be taken to ensure the ledges of the structural layers do not cause internal shading on the
solar cells. Additionally, debris collecting on the surface would need to be removed through street
sweeping, rubber removal, or other processes if it begins to impact the overall radiation reaching the
solar cells (Northmore & Tighe, 2012b).
In addition to the solar cells being very fragile components, the connecting links between the cells
are also very fragile. The cell interconnections are typically a tin-lead ribbon that is soldered to the
top of one solar module and the bottom of the next one in the series. In a solar road panel these
connections would need to be more robust due to the higher loadings on the panel and the larger
distances between solar cells due to the ribs in the structural layer design (Northmore & Tighe,
2012b).
As was shown in Figure 1-2, the structural layers of the solar road panel are the optical and base
layers within the conceptual design model. The main requirements for the material selection for these
layers are that the material should be able to provide adequate structural performance for the panel, as
defined within the design requirements, and that the material be suitable to easy in-house prototype
construction. This second requirement is of key importance due to the use of this prototype as a
current and future research platform.
With these requirements in mind, the candidate materials identified for the structural layers were
A36 steel, 6061-T6 aluminum, and fiberglass reinforced polymers (FRPs). The mechanical properties
of these materials are identified in Table 3-1 with the same properties of traditional pavement
materials used in Ontario for comparison.
Aluminum – 6061-T6
Structural 276 68.9 2.70
(ACI, 2013)
GPO-3 Laminate
Structural Fiberglass (Rochling, 55 12 1.90
2013)
HDPE – Glass Fiber Filled
Structural 36.9 7.38 1.25
(ACI, 2013)
ABS – Glass Fiber Filled
Structural 120 5.55 1.56
(ACI, 2013)
Concrete Pavement (ARA,
Pavement 32 29.6 2.32
2011)
Asphalt Pavement
Pavement N/A 2.76 2.46
(AASHTO, 1993)
In terms of material performance one of the major objectives is to minimize deflections within the
panel due to external loading in order to protect the solar cells. To this end A36 and 6061-T6 are the
best options due to their high compressive yield strength and Young’s modulus; meaning they can
withstand large loads and require large loads before substantial deformation occurs to the material.
31
The yield strengths for the FRP materials are more comparable to the ultimate strength of typical
concrete pavements, however the Young’s modulus only being two to three times that of a typical
asphalt pavement indicates that these materials are more prone to high strains under low loads given a
comparable cross section to the A36 and 6061-T6 options.
Typically the major benefit of the comparatively high strengths of metals over polymers is that
you are able to use less material to obtain the same level of performance, thus having an overall lower
cost and lower weight component. While this was demonstrated in several of the landing mat designs
covered in the literature review, minimizing the material used like this may not be possible in the
prototype design in order to accommodate the electronics, glass, and housing with readily available
components.
Another challenge with the structural layers is choosing a material that is cost effective for
building a prototype out. This creates a large difference between the available material options as the
ideal way to construct the details for the optical layer out of a metal is through a casting operation
while the ideal method for the fiberglass is through customized multi-ply construction. Both of these
processes are very complicated and expensive to accomplish, though for the purposes of developing a
one-off prototype it is simpler to go with a multi-ply fiberglass approach. The simpler alternative to
these is purchasing sheets of the respective material and then cutting them down to the required sizes,
which is not structurally optimal due to the epoxied ribbing but allows for simple construction of the
layers out of any desired material.
One major area where these materials differ is on environmental resistance. Steel and aluminum
would both require coatings to protect them from rusting and from being conductors of the current
flowing through the panel from the photovoltaic system. The two glass-filled polymers both have
typically poor environmental resistance as the polymer matrix degrades in corrosive environments.
The GPO-3 material is designed to be an electrical insulator and also be inert in challenging
conditions.
32
Table 3-2: Transparent layer material mechanical properties
These materials all demonstrate higher compressive strengths than typical concrete pavements,
which indicates that they should all be able to operate as a transparent layer within a solar road panel
through diligent design. The large structural difference comes through the Young’s Modulus of the
materials, as tempered glass is far more rigid under loading than concrete while acrylic and
polycarbonate will be nearly as flexible as an asphalt pavement is currently. This could cause issues
in designing the transparent layer for the polymer materials as they are cantilevered over the solar
cells and large deflections should be avoided as they may cause damage to the solar cells.
Another large difference between these materials is the way in which they are most likely to fail
under loading. As was identified in the literature review, polymer materials under vehicle loading
typically demonstrate plastic deformation through rutting and shoving of the top layers of the cast
material. Also, the optical grade versions of these polymers, which are required in order to maximize
the solar energy that is able to reach the photovoltaic cells, are typically specified for temperatures
above 0°C, meaning that they should not perform as well as expected under typical Canadian winter
conditions.
The tempered glass, on the other hand, does not fail through plastic deformation like the polymer
options though this means there is less indication of performance loss before failure. In order to safely
design a tempered glass panel it must conform to typical glass flooring standards, so the structure
must use multiple redundant panes of tempered glass which are laminated together, for reasons
outlined in the literature review. The tempering process also means that should a pane
catastrophically fail it would break into very small shards instead of large sheets of glass while the
33
lamination also helps bind these broken shards to the other layers of glass in the panel instead of
spreading into the rest of the environment.
The last set of important differences between these materials is in cost and texturing. The
tempered glass is substantially more expensive a design option than the cast polymers, especially
since this would have to be outsourced for the construction of the CPATT prototypes. Also, all of
these materials can have surface textures applied to them through a number of processes during
casting or in post-processing through etching.
Monocrystalline silicon photovoltaic cells are available in a variety of sizes to meet various needs
in custom OEM products. The typical size used in utility power generation applications is 150-mm
square solar cells, as these can be produced efficiently with relatively high energy conversion rates.
Since 150-mm is a large area to cantilever the transparent material over, the decision was made to use
the next size down of high efficiency solar cells; 125-mm square solar cells. These still leave a lot of
the surface area of the panel available to generate electricity while leaving plenty of space for load
transfer around the solar cells. Additionally, due to packaging requirements for the scaled prototype,
the 125-mm cells proved a better fit that the 150-mm cells as demonstrated further in this section.
Figure 3-1 shows the dimensions of the solar cell selected for this project, where all of the specified
dimensions are in millimetres.
34
Figure 3-1: 125 mm square monocrystalline silicon solar cell (RMSolar, 2012)
This solar cell selection allowed the DMSolar DMS-125M-280 photovoltaic cell (RMSolar,
2012) to be used in the design of the prototype panel. This cell has a maximum power point voltage
and current of 0.521-V and 5.342-A respectively. Due to the smaller size of the cells it is possible to
fit 25 of them into each square metre of solar road panel, resulting in a total output voltage of 13.025-
V and a peak power output of 69.58-W.
35
Figure 3-2: Schematic solar road panel interconnection scheme (Northmore & Tighe, 2012b)
With the material selected, the most important design property to determine from a structural
standpoint is the thickness of the transparent layer. Due to safety considerations it is important to have
multiple panes of glass laminated together to support the loading, and in this case a two-pane
configuration was selected such that should one pane break the other would be able to support the
vehicle load by itself. Under this condition it is known that the design stress for tempered glass
members is 42-MPa (Alsop & Saunders, 1999).
36
With these conditions it was possible to determine the bending stress that would be found in a
cantilevered section of the glass. It was assumed that the cantilevered sections would be 140-mm
square, large enough for a 125-mm solar cell and interconnection space, and that the highest load
condition would see an even distributed load from a passing vehicle tire, assumed to be 480-kPa of
pressure. This information was applied to correlations developed for stress relations (Roark & Young,
1975), using low bending theory, and the results for varying pane thicknesses are shown in Table 3-3.
Table 3-3: Maximum transparent layer bending stress as a function of glass pane thickness
Thickness (mm) 6 8 10 12 14 16 18 20
Maximum Bending
75.11 42.25 27.04 18.78 13.80 10.56 8.345 6.760
Stress (MPa)
Maximum
0.5415 0.2284 0.1170 0.06769 0.04262 0.02856 0.02006 0.01462
Deflection (mm)
As demonstrated, all of these configurations give very minimal deflections even when using thin
layers of tempered glass, which validates the use of the low bending theory assumptions. For
individual panes of glass the minimum usable configuration is the 10-mm design, so two 10-mm
panes were determined to be used in the solar road panel prototype.
In order to make the glass structure usable in a solar road panel, a texture must be applied to the
surface and accommodation must be made for the panel to be packaged together by a frame. Since the
primary focus of this research is on the structural characteristics of a solar road panel and not the
functional ones, a simple etching pattern was chosen from the catalogue of the supplier of the glass
structure. This texture will add some tractive capacity to the panel while also not degrading the
optical quality too severely.
In order to frame the panel, it was determined that the best approach to ensuring the glass is
secured into the panel was to design a ledge into the overall glass panel structure. This was done by
using two 10-mm glass panes cut to different square sizes and centred upon each other during
lamination. An image of the designed glass layer can be found in Figure 3-3.
37
laser cut, or water jet cut in order to produce the required elements in the design while the fiberglass
can be cut to size using standard equipment. For this main reason the prototype structural layers were
made from GPO-3 rated fiberglass.
When designing a road-testable prototype, the frame of the panel is going to have a significant
impact on the overall structural performance due to load transfer characteristics. For the purposes of
this study, where the prototype panel is being constructed at a smaller scale than a full-sized road
panel would be, the frame should have minimal impact on the performance of the transparent and
structural layer materials. The frame material should also be readily available and easily workable; to
this end it was determined that the frame should be made out of 6063T5 aluminum channel members.
With the material selected, the options for the overall panel thickness were narrowed down.
Typical 6063T5 channel is available in 25.4-mm [1-in], 38.1-mm [1.5-in], 50.8-mm [2-in], and 76.2-
mm [3-in] channel widths, which is the governing dimension for the panel thickness. With the glass
requiring 20-mm of thickness, it was determined that 50.8-mm is the minimum usable channel size to
allow for multiple layers of fiberglass to compose the optical and base layers. In terms of structural
performance a greater thickness of fiberglass would provide more resistance to bending, however due
38
to practical limitations on fiberglass availability the 50.8-mm channel thickness was chosen over the
76.2-mm channel.
Overall the optical layer of the solar road panel prototype needs to be thin in order to minimize
solar cell shading, as outlined in the design requirements. This layer also needs to allow for cutouts to
be made for cell interconnections, making it easier to design a two-layer structure for the optical
layer; one with the cutouts for interconnections and one without to support the transparent layer.
This meant that the optical layer would be designed as two layers of 6.35-mm cast fiberglass
sheet. The pattern used for this layer is shown in Figure 3-4 and accommodates the 125-mm solar
cells with 6.35-mm of space around the edge to allow for the interconnections to be routed.
39
In order to make the cutouts in the optical layer for the solar cells, the grid pattern was broken
down into smaller rectangular bars that could easily be manufactured with a chop-saw that was
available for use. At full scale production this would be completed through laser or water jet cutting.
40
3.1.7 Prototype Model
The overall panel was then assembled as shown in Figure 3-7 and detailed drawings for each
component can be found in Appendix A.
3.2 Construction
The construction of the prototypes was completed at CPATT at the University of Waterloo between
the lab facilities on campus and at the Region of Waterloo Emergency Services Training site. Since
the focus of the prototype is for determining the structural panel performance the construction of the
electrical subsystem elements has been omitted.
41
3.2.2 Fiberglass Fabrication
As outlined in the Bill of Materials, the fiberglass was all purchased in 6.35-mm and 12.70-mm plates
and then cut to size in-house for assembly into the solar road panel prototype.
With the bulk of the material trimmed away, the second stage of fabrication was to sand the edges
of the panel down to the proper, square dimensions using a palm sander and coarse sanding pads. The
faces of the two pieces to be epoxied together were also sanded to improve bonding.
The final stage was to epoxy the two base plate materials together to form one large block of
fiberglass material with the designated lip for the housing. This was accomplished as shown in Figure
3-10.
42
done with a table saw, laser cutter, or water jet cutter but none of these were readily available with
adequate ventilation.
The first stage of this fabrication was done by using a circular saw to cut the large fiberglass
sheets into portions that could be further cut-up by the chop-saw. All of the sheets required were
divided into sections for the three different piece sizes that make up the optical layers and taped off to
designate cutting lines. This overall process is demonstrated in Figure 3-11.
With the fiberglass sheets trimmed, the next step was to use the chop saw to cut the lengthwise
dimensions of the fiberglass pieces. This was accomplished by setting up the chop saw to the
43
designated specimen width and then sequentially trimming the required width one piece at a time, as
shown in Figure 3-12.
With the correct width of each specimen achieved, the third step was to square one end of each
member so that the pieces could be trimmed to the correct length. This was not required for members
that were adjacent to the edge of the finished cast panel, as it was assumed that the acquired fiberglass
sheets were square enough for our purposes.
With three of the edges squared, the next step was to mark each of the pieces for the correct
length so that the last end could be trimmed off and the parts would be the correct dimensions. This
was done in the lab with all of the pieces marked as shown in Figure 3-13.
The fifth step was to then trim off these ends to finish the fiberglass pieces. This process was
accomplished in a similar fashion to what was shown in Figure 3-12.
With all of the pieces cut to the appropriate size, the sixth step was to measure and sort the pieces
by the critical dimension. In the case of the short ribs this was the length while this dimension was the
width of the long rib and edge members. These dimensions were identified as critical because they
were what would collectively add up to the overall panel dimension in the direction with the most
cuts, meaning that there were more opportunities for compounding dimensional errors to add-up.
44
Figure 3-13: Fiberglass ribs marked for trimming
Then, the pieces were all arranged and taped in the pattern required for the optical layer. After the
taping, the pieces were untapped individually so that all of the connection pointes could be sanded
manually and epoxied using a standard 2-phase epoxy. The pre-epoxied layout of these pieces is
shown in Figure 3-14 while the epoxied joints are demonstrated in Figure 3-15.
45
Figure 3-15: Optical layer epoxied joints
The last step of fabricating the optical layer is to sand down the epoxy on the top and bottom of
the panel as well as around the edges of the fiberglass to make the layer conform with the required
dimensions. The end result of the sanding is as demonstrated in Figure 3-16.
46
The next step involved cutting and applying a plastic edge trim to the edges of the aluminum
frame members. This was completed to protect the structural materials of the solar road panel
prototypes from contact with the frame members.
The last step involved cutting and applying weather-resistant foam to selective internal portions
of the aluminum frame. This was done to create a flexible buffer between the structural materials and
the frame so that hard contact would be minimized. The foam inserts also helped make-up gaps
between the frame members and the structural materials. The end result of the edge seal and foam
application to a frame member is demonstrated in Figure 3-17.
Figure 3-17: Housing frame with edge trim and foam inserts
47
research showed that multi-ply areas of fiberglass were the optimal option for the study, but after
several attempts it was determined that precast fiberglass laminate would have to be used due to the
challenges of working with multi-ply fiberglass.
It was also noted, approximately a year after the glass was acquired from the supplier, that their
recommended glass structure for this application had changed from being two panes of tempered
glass laminated together into using a traditional bulletproof glass type structure. This tends to use
thinner panes of glass and a larger resin layer to protect from impact loading, though it is also
unknown how the thicker resin layer would impact on the rutting and deformation performance of the
transparent layer.
It was also learned, while cutting the base layers of fiberglass, that the optimal method for using a
skill saw to cut through such large blocks of material is to alternate cutting two close parallel lines.
Trimming at the edges of the fiberglass plates was not challenging, as the material was able to bend
away to make room for the saw blades to cut through. However, whenever an interior cut had to be
made, the circular-saw would bind too easily if a relief line was not being cut right next to it. This
helped speed up the process of cutting and reduced the number of saw blades that needed to be
purchased.
48
Chapter 4
Testing and Analysis Methodology
With the prototype panel now designed and constructed, the testing and analysis methodology can
now be outlined. This work is largely divided into three categories: structural testing, FE analysis, and
environmental testing.
This is largely broken into four segments: determining the testing objectives, test frame design,
load apparatus design, and instrumentation.
The testing should specifically allow for variable loading and variable load application, to ensure
that performance of the composite panel is truly consistent. Testing will be performed for static
response and within the elastic range for all materials in question, as the static, low load response will
be sufficient to determine the comparative response of a panel with in-situ loading.
49
Figure 4-1: Structural testing rig schematic
50
uses the base of the testing rig and strong-floor of the structures lab as structural support, making it a
simpler and more efficient design. There are no downsides to testing the panel upside down, as
gravitational forces can be accounted for within the FE model easily.
Figure 4-2: Deflection response of simulated panel to a centred 4,448-N [1,000-lbf] load
51
Figure 4-3: Structural testing rig base plate
Figure 4-4: Deflection response of 101.6-mm x 51.2-mm HSS under 6,672-N [1,500-lbf]
distributed load
With the material selected, it was cut and welded into a ring large enough for the panel to support
itself against. This simulates a simply supported condition that can be simulated in FE software for
validation of the panel model. The final product of this loading ring is shown in Figure 4-5.
Legs were also designed to connect the loading ring to the base plate while giving enough room
for the load apparatus and instrumentation to be installed. Also, since this frame is expected to see
52
limited use after this testing, the legs were designed to be removable so that the frame can easily be
stored post-testing. The legs designed for this apparatus are shown in Figure 4-6.
The final, constructed apparatus can be found in Figure 4-7 while the detailed design drawings for
the structural rig can be found in Appendix C.
53
Figure 4-7: Structural testing rig
The load being applied to the panel by a 25.4-mm [1-in] square steel surface covered in a
medium-stiffness rubber. The square applicator was chosen to simplify the modeling required to
validate the panel design; square areas are easier to model than circular ones. The rubber was added
to ensure a smooth contact area with the textured glass panel; accidental point loading would damage
the glass and cause premature failure of the prototype.
4.1.4 Instrumentation
In order to measure the output of the testing a variety of instrumentation is being used. This includes a
load cell to validate the load supplied to the panel, strain rosettes adhered to the glass and fiberglass
members, and LVDTs to measure the displacement of the bottom fiberglass layer.
The load cell being used to validate the loading in the panel is a StrainSert FL5U-2SPKT, as
shown in Figure 4-10. This was calibrated using a 13.4-kN [3000-lbf] load ring on a range from 0 to
8.90-kN [2000-lbf].
55
Figure 4-10: Structural testing load cell
The instrumentation measuring the response of the panel to the loading is being instrumented
over an eighth of the panel as demonstrated in Figure 4-11. This was done to minimize the
instrumentation requirements while allowing for symmetry effects to be assumed from the strain and
displacement responses from loading in a variety of locations around the panel. As a result, loading
on reference point C-2 could be simulated through the entire panel by loading points C-2, D-3, C,-4,
and B-3, for example.
56
The strain rosettes being used for the testing are all Vishay Model C2A-06-250LR-350 rosettes.
These rosettes are designed to be attached to both glass and fiberglass specimens and provide highly
accurate planar strain measurements. In order to be read by the NI data acquisition system available
from CPATT, the 350-ohm strain rosettes had to be shunted down to 120-ohm using a 182.6-ohm
resistor being mounted in parallel across it at the datalogger terminals.
The transducers being used to measure the displacement of the fiberglass base plate are HP
24DCDT-050 units. These LVDTs allow for up to a 1.27-mm [0.050-in] displacement, which is well
above the maximum expected displacement of 0.84-mm as determined from the initial FE analysis as
shown back in Figure 4-2.
The transducers were mounted to the load ring of the structural testing frame using magnetic
based retort stands. An image of the rosette and transducer positioning on the fiberglass base plate
during testing is shown in Figure 4-12.
To collect all of the data simultaneously, a NI SCXI-1000 data acquisition system with multiple
modules was used. This system was connected to a computer in the structures lab and was accessed
using NI LabVIEW. The entire setup for the structural testing is shown in Figure 4-13.
57
Figure 4-13: Structural testing setup
58
The fiberglass rib portion of the structural layers is modeled as a homogenous block of material
with the grid of squares cut-out. This is more realistic to how a future state prototype would be
fabricated, as this layer would be a customized multiply fiberglass layup, but may not accurately
represent the prototype constructed with epoxied joints.
The prototype transparent layer has a textured surface but this was modeled as flat for the FE
analysis. This is a reasonable approximation, as the soft rubber pad on the load applicator will evenly
apply the load across the contact area.
The boundary conditions of the structural testing were approximated using a flat plate steel ring,
with the same dimensions as the load ring shown in Figure 4-5, with an encastre boundary condition
applied to the top face. This implies that the ring will not deflect at all during testing, which is a
reasonable approximation for the highly rigid load ring.
To govern the contact between the layers of the panel model, surface-to-surface contact models
were generated in the standard solver using a simple normal contact property. The slave component
of the model is adjusted to remove overclosure prior to running the solver as this adjusts the three
shell elements to be directly adjacent and remove displacement distance added from the shell
thickness.
Due to challenges in modeling the interaction between the rigid transparent layer and the more
flexible structural layers some additional solver tools were required. The normal behavior of the
contact was driven by hard contact but with a standard linear penalty model applied to allow for some
error at individual nodes. Similarly, a contact control was added that allows for automatic
stabilization of the results using the default parameters provided by Abaqus. These techniques were
both prescribed by the Abaqus user manual (Abaqus, 2013) to solve convergence issues with the
modeling technique.
59
The two main parameters of the model validation are the meshing and the material properties.
The mesh has the greatest effect on the accuracy of the FE model, with finer meshes producing more
accurate results at the cost of increased computational time and memory allocation. The objective of
mesh optimization is to achieve a mesh that simulates the results of the structural testing with at least
95% accuracy. This is typical of FE analysis, as increasing accuracy requires exponentially more
computational time. Most studies found used an objective accuracy between 90% and 95% (Mak,
2012)(Z. Wu, Chen, & Yang, 2011).
Due to the geometry of the members structured meshes are easily generated for the transparent
and base layers, which tend to provide more accurate results. The mesh strategy used for the optical
layer was a free mesh, due to the continuous nature of the member with irregular cutouts. The
parameters being varied for the mesh validation are outlined below in Table 4-1, with the default
variables for the parametric study highlighted in bold.
The material properties assumed for the materials in the prototype are as outlined in Table 4-2.
The values for tempered glass are typical material property ranges as per the literature (ACI, 2013;
Alsop & Saunders, 1999); the inherent imperfections in glass manufacturing leaves large margins for
mechanical properties. The fiberglass laminate used in the prototype meets the NEMA GPO-3
standard, which has the supplied elastic modulus value from the manufacturer (Rochling, 2013).
Poisson’s ratio is not a commonly required property of thermoset polymers, so a range of potential
values was once again determined from the literature (ACI, 2013; Rochling, 2013). In order to satisfy
these ranges three material property conditions were created to represent the average, most flexible,
and most rigid combinations. These conditions will be modeled and compared to the results of the
structural testing.
60
Table 4-2: Prototype material property ranges and analysis categories
The static load condition is based on the heaviest wheel load applicable by Canadian regulations,
CAN/CSA-S6 CL-625-ONT (CSA, 2006), which is in this case the heaviest single wheel fourth axle
load. The fatigue load was determined based on the single wheel load equivalent of the dual wheel
ESAL load. This was developed in accordance with the geometric relations used to convert dual tire
loads into single tire loads for the analysis of concrete pavement sections (Huang, 2004), and since
the relations are all geometric it was assumed to be applicable to loads on non-concrete pavement
structures. A tire pressure of 600-kPa was assumed in determining this area.
The loads are applied to the model through a pressure application over the contact area. Abaqus
allows a total force to be distributed across the selected area, so this technique was used to apply a
ramped, static, general load to the panel model. This is demonstrated in Figure 4-14 on a sample
61
structural base. The load area is designated on the panel model by partitioning the transparent layer so
that there is a volume with a designated upper surface area that matches the tire contact area.
Gravitational loads were also included in the analysis.
In order to properly assess the performance of the solar road panel it needed to be loaded in four
distinct locations as shown in Figure 4-15: centre, transverse edge, longitudinal edge, and corner.
These conditions cover the extremes of typical panel stresses from tire loading, with the total load
being applied to either one, two, or four panels.
Figure 4-15: FE tire load application locations, direction of travel up the page
62
4.2.2.2 Structural Bases
The structural bases that were selected for analyzing the panel performance on are the typical
pavement structure guidelines for minor arterial roads with 1000 annual average daily truck traffic.
This level of traffic was chosen because it provides a reasonable example of where the panels would
be installed for initial trials, in lower volume and loading applications. The purpose of the FE study is
also to determine comparative performance of solar road panels on a variety of structural bases, and
medium volume road bases will provide a starting point for this comparison.
The structures for concrete (PCC) and asphalt (HMA) roads are taken directly from the
StreetPave report (ARA, 2011), which outlines the standards used for pavement infrastructure in
Ontario. Models for granular and subgrade structural bases are based on the HMA road design with
the additional layers removed as required. For this the HMA base was chosen over the PCC base
because of the increased equivalent thickness of the asphalt base, which implies greater strength. The
high strength subgrade was chosen as the structural subgrade for the models. The layer thicknesses
for the four structural bases are outlined below in Table 4-4.
The material properties for each of these material layers are defined below in Table 4-5. These
values are largely derived from Ontario’s default parameters for the AASHTOWare pavement design
tool (MTO, 2012), the Ontario provincial standards for granular materials (OPSS, 2003), the
StreetPave report (ARA, 2011), the Canadian Pavement Asset Design and Management Guide (TAC,
2012), and the AASHTO Guide for Design of Pavement Structures (AASHTO, 1993). These
documents represent the standard design practice for Ontario pavement structures, so no variability of
these values is considered in the study.
63
Table 4-5: FE structural base material properties
64
Table 4-6: FE boundary conditions per load application model
The mesh strategy in the structural bases is different from the panel model in that biased seeding
techniques are being used to optimize the analysis. Since the load application is concentrated over a
small portion of the overall model, a higher density of nodes is required at the load area to accurately
model stresses, strains, and displacements than in the far corners of the encastred boundaries. This is
accomplished by determining the number of nodes along a given edge using the average mesh size
and seeding the individual edges of the model with that number of elements but with a single bias
towards the loaded corner and a bias ratio of five, the Abaqus default. This seeding process is used for
all structural base edges in the x-y plane and the z-direction edges for the subgrade, but double bias
65
seeding is applied for all other z-direction edges to ensure that both upper and lower face contacts are
modeled accurately. This is demonstrated for the default subgrade layer in Figure 4-16 below.
The parameters being optimized through the validation process are the maximum stress, strain,
and translation within the panel model. For the sake of validation, the panel model is simplified to a
1-m by 1-m by 25.4-mm layer of tempered glass. The goal of the validation process is to achieve a
level of 95% accuracy within the model for each of these properties based on the parameters varied
from Table 4-7. The validation was completed in a cascading fashion, starting with the subgrade
depth and proceeding down the list identified in the table. This was done so that a value could be
locked in for each property to converge to the most accurate solution.
66
number of ESALs that the panel can withstand on each structural base before failure of any
component.
Glass specimens fail through fracture methods due to the brittle nature of the material. The
method by which this occurs is well documented in literature (Alsop & Saunders, 1999) making this
analysis method straight forward. The assumption is that there are micro-cracks in the surface of the
material 0.7-μm in length (Budynas & Nisbett, 2008). As the panel is cyclically loaded the crack
slowly propagates through the material. This propagation occurs at a constant rate while load is being
applied to glass, so sensitivity will also be analyzed for the speed of traffic over the panel. Other
factors including the notch parameters have been identified in literature (Budynas & Nisbett, 2008)
and will vary as the crack length increases, making this an iterative solution. In order for fatigue
failure to occur at all, the developed tensile strains must exceed the 69-MPa compressive edge stress
developed through the tempering process.
Fiberglass laminate fails through traditional fatigue theory methods, as S-N curves have been
developed that accurately predict the performance of various laminates to cyclic loading. In this case
the curve has been developed as shown below, in Figure 4-17, for short glass fiber reinforced
matrices; which is a best available estimate of the fatigue performance of the GPO-3 used in the
prototype panel. Determining the number of available load cycles is based on the fatigue load induced
stress to ultimate stress ratio of the material and then applying the given formula.
Concrete pavements fail through a number of mechanisms as outlined in the equations shown in
Figure 4-18; where Nf is the maximum number of stress cycles, σ is the cyclical stress applied to the
concrete, and Sc is the compressive strength of the concrete. Similarly to the equations for GPO-3, an
endurance limit exists at a maximum stress of 0.45 times the compressive strength of the concrete at
which point unlimited fatigue cycles may be applied. The compressive strength assumed for these
calculations will be 32-MPa, a traditional design value for Ontario pavements (ARA, 2011). These
conditions are also sensitive to the joint placement as typically higher stresses are realized at the
edges and corners of concrete slabs.
67
Figure 4-17: S-N curve for cyclic loading performance of short fiber reinforced matrices
(Demers, 1998)
Asphalt pavements fail through two primary mechanisms; fatigue cracking and rutting. For
fatigue cracking, the relation between design life and horizontal strain developed at the bottom of the
asphalt layer is shown below in Figure 4-19; where Nf is the allowable number of load cycles, εt is the
horizontal strain at the bottom of the asphalt layer, and E1 is the elastic modulus of the asphalt
material. The elastic modulus value used in this empirical equation must be input in imperial units, so
a value of 400,000-psi was used.
68
Figure 4-19: Asphalt pavement fatigue life equation (Huang, 2004)
Rutting, a failure that occurs in both asphalt and granular structures, is a function of the vertical
compressive strain that is developed at the bottom of the lowest granular layer. The relation for this is
shown in Figure 4-20; where Nf is the allowable number of load cycles, εt is the vertical compressive
strain at the bottom of the granular layers, and E1 is the elastic modulus of the asphalt material.
Figure 4-20: Asphalt and granular rutting life equation (Huang, 2004)
For the purposes of this testing it is important to determine if salt brine application has an impact
on these materials that should be further investigated and optimized. For this purpose, the four-
percent by weight solution of anhydrous calcium chloride was seen as an inadequate scaling solution.
In order to determine whether or not the scaling has an impact on the samples, it was determined that
a 25-percent by weight solution of sodium chloride should be used. This is a common salt that is used
as a chemical deicer, making it an ideal candidate, and at this concentration is near the maximum that
can be obtained at room temperature.
This does add another difference between this and traditional scaling resistance tests. A 25-
percent by weight solution of sodium chloride would not actually freeze at the temperatures we could
cycled the samples to, though the samples still reach a temperature of -16°C as verified by an infrared
thermometer. This should have little impact on the study as the typical purpose for freezing the brine
solution is to propagate the ice into cracks in the material. Due to the smoothness of our samples there
should be no significant cracks for ice to propagate into, thus making the higher solution strength a
greater asset in determining corrosion potential and effects.
70
The method of application for the salt brine also proved challenging due to the nature of the
specimens being tested. The standard calls for a constant solution depth on the surface of the
specimens to be maintained, however the number of specimens and their irregular sizes being tested
made this challenging. To accommodate this, it was determined that the specimens could be
submerged or suspended in containers of the brine solution. This still maintains an adequate level of
brine being applied to a specific surface of the specimen, and each specimen could be identified for
weather it was submerged or suspended. The submerged samples were typically glass while the
polymers were suspended due to the differences in density of the materials.
This cart was then rolled in and out of the CPATT Lab’s walk-in freezer at intervals allowing at
least two hours of time for the samples to freeze and two hours of time for the samples to thaw, as
these were the intervals determined during checks in the first cycling to ensure that the samples met
their required temperatures as verified by an infrared thermometer.
71
As this cart was also used for freeze-thaw cycling of pervious concrete samples, the tracking
sheet used to count the testing cycles performed on those samples was modified for use with these
optical test specimens.
In order to minimize the cost of the apparatus required for this test, it was decided that an existing
base would be used for the testing which had the capacity for additional brackets to be fitted to it to
72
meet the ASTM D790 specifications. The base, which is used for similar flexural testing, is owned by
Dr. Marianna Pollak within Civil and Environmental Engineering and is shown below in Figure 4-23.
Three pieces of hardware were then designed to make this base work with the Instron machine
from Mechanical Engineering and the ASTM standard we are following; a collar to mount the base to
the Instron, the support nose brackets, and the load nose assembly. Figure 4-24 shows the finished
components as manufactured by RJB Machining, and touched up as required, while detailed drawings
for them can be found in Appendix D.
Figure 4-24: 3-point bending test support noses, base collar, and load nose (left to right)
In order to accommodate for the thinner materials, it was determined that a smaller span than
101.6-mm [4-in] should be used in testing. The span based on the actual material thicknesses was
determined to be 95.25-mm [3.75-in] as this would allow for all of the specimens to be tested within
the bounds of sample measurement for the ASTM D790 standard.
To accommodate the test specimens, a platform with adjustable bumpers, as shown installed in
Figure 4-27, was designed to be installed around the existing formwork for the pendulum. This
platform also allowed for compatibility between ongoing friction testing in the lab, so that the
pendulum did not have to be constantly readjusted between testing concrete cylinders for other
research and the optical materials for this analysis.
The data obtained from this testing is in terms of British Pendulum Numbers, as the values
obtained were too low for comparison in terms of Skid Number; a value more commonly used in
pavement evaluation.
Tests were conducted at 5-nm intervals from 250-nm up to 1200-nm. The 1200-nm ceiling for the
tests was chosen to ensure that the entire spectrum of absorption for solar PV cells was covered
(Luque & Hegedus, 2003) and to reduce time from running the test to 2500-nm as was available; the
76
longwave radiation testing, testing over 800-nm, is highly time intensive due to the sensor used for
data collection.
For the testing, 50.8-mm [2-in] square samples were used to ensure that material was covering the
entire opening for light transmission. The samples were cleaned with a glass cleaner prior to testing to
remove any residue that collected during the scaling process. For transmission testing the samples
were installed in front of the integrating sphere of the Cary 5000, as shown in Figure 4-29, so that
electromagnetic radiation could be passed through the sample and into the integrating sphere. For the
reflection testing, the samples were installed behind the integrating sphere as shown in Figure 4-29.
Figure 4-29: Sample positioning for transmissivity and reflectivity testing (left to right)
77
Chapter 5
Structural Prototype Testing and Analysis
This chapter outlines the results from the structural testing and finite element analysis that was
described in the methodology.
Since the panel orientation for testing was upside down, as shown in Figure 4-13, the orientation
of all of the strain and deflection response data is with respect to the locations shown in Figure 5-2.
This figure still shows the load cases, however since the base layer of the panel is facing upwards in
this configuration these labels only represent where the maximum strains and deflections should be
noted during the testing.
78
Figure 5-2: Structural testing result orientation
Due to the orientation of the rosettes and LVDTs, as outlined in Figure 4-11, there are no values
collected for A-2, A-4, B-1, B-5, D-1, D-5, E-2, or E-4. This is represented in the figures below as
having no strain or deflection at these points.
300
250
200
150
100
E
50
C
0
1 2 A
3 4 5
This same phenomenon applied to all four data sets on the glass panel, including the eccentric
load placements for the side and diagonal load cases. This is demonstrated below in Figure 5-4 for the
2.22-kN diagonal load test.
79
Tensile Strain [Microstrain]
120
100
80
60
40
E
20
C
0
1 2 A
3 4 5
Also notable from Figure 5-4 is that the measured strain response is fairly symmetrical about the
diagonal axis from A-5 to E-1. Given that four tests with 2.22-kN loads at B-2, B-4, D-2, and D-4
were amalgamated to make this profile, the symmetry represents a high level of accuracy within the
testing regimen and constructed prototype.
In some cases for the strain response of the base layer strains were observed at the edges of the
panel that were higher than at the expected peak location. This is shown in Figure 5-5 for the 2.22-kN
centre load case. Note that the highest strain would be expected at location C-3, however higher
values are found at every point around the edges. This is believed to be a result of boundary effects on
the panel, as repeated testing delivered the same results.
Tensile Strain [Microstrain]
60
50
40
30
20
E
10
C
0
1 2 A
3 4 5
80
5.1.2 Deflection Response
Due to the orientation of the panel and the sensitivity of the transducers being used, deflection
measurements were only taken for determining the differences in deflection between two load cases.
As a result, this testing was only accomplished for the centre load condition for the deflection
difference between 2.22-kN and 4.44-kN.
The results from this testing followed the trends that were expected, as shown in Figure 5-6,
where the largest deflection between the 2.22-kN and 4.44-kN loads occurred at the centre of the
panel and tapered towards the edges.
Deflection Difference [mm]
1.5
0.5
E
C
0
1 2 A
3 4 5
Figure 5-6: Deflection difference from testing between 2.22-kN and 4.44-kN centre loads
However, it is important to note that the deflections measured here are much larger than were
expected based on the initial FEA shown in Figure 4-2. In this figure an expected maximum
deflection of 0.84-mm is demonstrated for a 4.44-kN centre load, while the test data shows a
deflection of 1.4-mm between the 2.22-kN and 4.44-kN load conditions.
Extensive reworking of the laboratory test apparatus occurred in an effort to ensure that the
testing was appropriate for measuring the deflection responses. As noted earlier, the objective of the
testing is to drive simulations that can predict the in-field strain and deflection responses of this panel
to vehicle loading. The displacement transducers were recalibrated to ensure their accuracy using thin
plates of known thicknesses. The frame members were also removed, as shown in __, as it was
believed that the foam inserts and edge seal may be affecting the response being measured.
When the tests were rerun under the new conditions it was found that the deflections measured
were of a similar magnitude. Since there is a discrepancy between these measurements and the
81
original predicted values, the model development will focus on the measured strain values and the
deflection will be analysed with respect to these models.
Model Parameter
Layer Integration Layers Mesh Size [mm]
Transparent Layer 3 6.35
Optical Layer 3 25.4
Base Layer 3 6.35
82
Figure 5-7: 2.22-kN diagonal load simulation average material property glass strain contours
Figure 5-8: 2.22-kN diagonal load simulation average material property base strain contours
83
5.2.3 Model Validation
To compare the strain data between the structural testing and the simulations the absolute difference
between the two models was calculated. This result for the 2.22-kN diagonal load with average
material properties is shown in Figure 5-9. Note that the differences in strain are larger on the
boundaries than they are on the inside of the model; this is likely due to difficulty in accurate
boundary condition simulation.
Absolute Strain Difference
20
[Microstrain]
15
10
5 E
C
0
1 2 A
3 4 5
Figure 5-9: 2.22-kN diagonal load absolute strain difference for the average material property
model
The results shown in Figure 5-9 were then averaged to provide an average absolute error metric
which was used to determine the optimal material property configuration. The results of this are
shown in Figure 5-10 and demonstrate that, across the entire model, the rigid material property
configuration produces the lowest average absolute strain error.
This figure, however, does not take into account that the boundaries of the model have larger
errors due to the boundary conditions. To eliminate these known errors from the decision making, the
boundaries were removed from the data set by simple exclusion; the new data set did not include
values from grid locations with A-, E-, -1, or -5 labels.
New average absolute strain errors were determined as shown in Figure 5-11. This figure shows
that while the average properties are the best performer for the diagonal load case, that the overall
best material property configuration for matching the structural testing is still the rigid configuration.
84
18
16
[Microstrain]
12
10
8 Flexible
6 Average
4
2 Rigid
0
2.22-kN 2.22-kN Side 2.22-kN 4.44-kN
Centre Diagonal Centre
Load Cases
Figure 5-10: Average absolute strain error between testing and simulations
10
9
Average Absolute Error
8
7
[Microstrain]
6
5
Flexible
4
3 Average
2
Rigid
1
0
2.22-kN 2.22-kN Side 2.22-kN 4.44-kN
Centre Diagonal Centre
Load Cases
Figure 5-11: Average absolute strain error between testing and simulations for non-boundary
locations
85
The other deficiency of the model that has been developed is its ability to accurately predict the
maximum glass strain under the 4.44-kN centred load test. As noted in Figure 5-12, the model under
predicts the maximum strain seen in the transparent layer by 41-microstrain. The majority of this
difference is likely due to a mechanical defect in the glass specimen, as unsupported glass should not
strain that much under similar load and boundary conditions.
Absolute Strain Difference
50
40
[microstrain]
30
20
E
10
C
0
1 2 A
3 4 5
Figure 5-12: 4.44-kN centre load absolute strain difference in the transparent layer for the rigid
material property model
Table 5-2. This information shows that the values measured are all consistently larger than the
values that were predicted. In this case there are two groupings of values that are off by similar
margins, with C-3, D-3, and D-2 in one group between 300% and 400% and E-3 and E-1 in the other
around 600% to 650%. This is similar to the results found from the strain model validation as E-3 and
E-1 are both boundary locations and their test values differ from the model by a larger margin than
the other central nodes.
86
Deflection Difference [mm]
0.3
0.25
0.2
0.15
0.1
E
0.05
C
0
1 2 A
3 4 5
Figure 5-13: Deflection difference from modelling between 2.22-kN and 4.44-kN centre loads
The results show that there is a clear difference between the deflections that are being measured
and those that are being predicted, though the similarity of the deflection profile between the
measured and predicted results lends confidence that the modelling procedures accurately depicting
the performance of the panel in the laboratory testing conditions. Further study should focus on the
deflection performance of these panels, particularly for in-situ applications.
Base Structure
Parameter PCC HMA Granular Subgrade
Subgrade Depth [m] 0.5 1.0 1.0 1.5
Base Length [m] 4.0 3.0 2.0 1.5
Base Width [m] 4.0 3.0 2.0 1.5
HMA Average Mesh Size [m] - 0.025 - -
PCC Average Mesh Size [m] 0.025 - - -
Granular A Average Mesh Size [m] 0.050 0.050 0.025 -
Granular B Average Mesh Size [m] - 0.050 0.050 -
Subgrade Average Mesh Size [m] 0.100 0.100 0.100 0.100
The most important detail to note from Figure 5-14 is that all of the stresses found within the
transparent layer are far below the tensile and compressive yield strengths of glass, meaning that the
loads being applied can be safely supported. Also of importance is the ranking order of the pavement
structures, with the bases demonstrating the least to most stress, in order, were the PCC, HMA,
Granular, and Subgrade bases. This result was expected as the PCC base contains the most rigid
materials while each subsequent layer becomes more flexible, providing less resistance to deflection
and less load spreading.
The unexpected result from Figure 5-14 is that the centre load case produces the highest stress
while the corner case is the lowest for each structural base. The opposite is true in typical pavement
applications due to low load transfer between panels limiting the availability of load spreading.
88
However, Figure 5-15 demonstrates that the maximum stress locations in the solar road panel
prototype are more dependent on the optical layer grid pattern.
Max Transparent Layer Stress [MPa] 50
45
40
35
30
25
20
15
10
5
0
Centre Longitudinal Transverse Corner
Tire Load Locations
Figure 5-15: Transparent layer stress contours under static, centre load with HMA base
89
While the maximum stress in Figure 5-15 still occurs in the bottom right corner of the member
directly under the applied load, a second maximum stress location appears directly above it and this
space corresponds to the adjacent cutout in the optical layer. These higher stresses are a result of the
glass being cantilevered over the solar cells at these points, and this proves to be the main structural
criteria for the transparent layer as assumed in section 3.1.4. This phenomenon is further illustrated in
Figure 5-16 where the maximum stress from the corner load case occurs inset from the corner of the
panel due to the location of the first optical layer cutout.
Figure 5-16: Transparent layer stress contours under static, corner load with HMA base
Figure 5-17 demonstrates the maximum stress in the transparent layer as a result from the fatigue
load application.
Under the fatigue load case the pattern of the transparent layer stress changes dramatically from
the static load case. This is because the load being applied is lower, but has a larger width on the glass
surface due to the dual tire configuration of an ESAL. As a result of the lower load and larger
application area, the maximum stresses are much smaller than under the static load case. The stress
contour from the fatigue, centre load with an HMA base can be found in Figure 5-18, and this image
demonstrates again that the maximum stresses occur in the cantilevered glass sections.
90
20
Figure 5-18: Transparent layer stress contours under fatigue, centre load with HMA base
91
5.3.2.2 Base Layer Stress
The second benchmark is the stress that is developed in the base layer. This characteristic is also
important to determine the structural capacity of the panels to vehicle loading as values under 55-MPa
are required to avoid yielding, and the results found are shown in Figure 5-19.
10
Max Base Layer Stress [MPa]
9
8
7
6
5
4
3
2
1
0
Centre Longitudinal Transverse Corner
Tire Load Locations
These results for maximum base layer stress follow the same pattern as was noted for the
maximum optical layer stress, which was expected. These maximum stresses developed under the
ribbed sections from the optical layer, as shown in Figure 5-20, with the values decreasing as you get
farther from the loaded area. It is important to note that the maximum stresses achieved are well
under the 55-MPa yield strength of GPO-3, so these components also will not fail due to static
loading.
The base layer stress for the fatigue load case, as shown in Figure 5-21, also produces similar
results to the static load case but with much lower peaking under the transverse load condition.
92
Figure 5-20: Base layer stress contours under static, centre load with HMA base
5
Max Base Layer Stress [MPa]
0
Centre Longitudinal Transverse Corner
Tire Load Locations
93
5.3.2.3 Base Layer Strain
The next benchmark is the strain developed in the base layer. This is important as the mechanical
limits for solar cells are typically defined in terms of strain limits, so ensuring that this value is under
6,250-microstrain to prevent damage to the solar cells is critical to the design (Ritchie, 2003). The
results for the maximum base layer strain under the static load case are found in Figure 5-22.
800
Max Base Layer Strain [micro-strain]
700
600
500
400
300
200
100
0
Centre Longitudinal Transverse Corner
Tire Load Locations
This figure follows the exact same pattern as the maximum base stress figure, as is expected due
to the linear relationship between stress and strain during elastic deformation. It also shows that the
maximum strains developed in the base layer are far below the 6,250-microstrain limit that the solar
cells can be subjected to. Figure 5-23 shows the locations of these maximum strains and it is
important to note that the maximum strains are located under the optical layer ribs and not in the solar
cell cutouts, so the actual strains subjected to the solar cells will be far lower than noted in Figure
5-22.
The lower stress state of the fatigue load case produces significantly reduced strain profiles, as
shown in Figure 5-24.
94
Figure 5-23: Base layer strain contours under static, centre load with HMA base
300
Max Base Layer Strain [micro-strain]
250
200
150
100
50
0
Centre Longitudinal Transverse Corner
Tire Load Locations
95
5.3.2.4 Structural Base Deflection
The final benchmark being assessed is the maximum deflection measured within the structural base.
This value is representative of the deflection caused in the panel. The results of this are shown in
Figure 5-25.
5.0
Max Structural Base Deflection [mm]
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Centre Longitudinal Transverse Corner
Tire Load Locations
The structural base deflections produce similar trends to the structural base strains shown in
Figure 5-22, which again is logical due to the typical performance of plates used in pavement
reinforcement. The profile of this deflection through the specimens is found in Figure 5-26, and this
shows the expected maximum deflection under the loaded section with the deflection reducing
quickly as you move away from this point.
Figure 5-26: HMA layer deflection contours under static, centre load with HMA base
Figure 5-27 shows the structural base deflections determined from the fatigue load case. Once
again these values are noticeably lower than those for the static load.
96
2.0
Figure 5-28 clearly shows that the transparent layer in the prototype would not form fatigue
cracks in these configurations. As a result, it is possible to reduce the thickness of the glass panes
used in testing, in particular when the panel is installed on a PCC, HMA, or Granular structural base.
As a result of this, no fatigue crack growth analysis was completed.
97
however, is only valid when this stress ratio is over 0.3 as components have demonstrated infinite
fatigue lives and lower stresses. As a result the first check to perform on the optical and base layers is
to see if they surpass this limit, which is 16.6-MPa for GPO-3 laminates. This is shown below in
Figure 5-29 for the base layer stresses, which in every case were higher than those in the optical
layers.
70
Max Transparent Layer Stress [MPa]
60 PCC - Static
50 PCC - Fatigue
HMA - Static
40
HMA - Fatigue
30
Granular - Static
20 Granular - Fatigue
10 Subgrade - Static
Subgrade - Fatigue
0
Center Longitudinal Transverse Corner Endurance Limit
Tire Load Locations
18
16
Max Base Layer Stress [MPa]
PCC - Static
14
PCC - Fatigue
12
HMA - Static
10
HMA - Fatigue
8
Granular - Static
6
Granular - Fatigue
4
Subgrade - Static
2
Subgrade - Fatigue
0
Center Longitudinal Transverse Corner Endurance Limit
Tire Load Locations
98
Figure 5-29 also clearly shows that the fiberglass components were designed under the endurance
limit of GPO-3, so in the configurations with this prototype design they will not fail either due to
cyclic loading.
16
Max PCC Layer Stress [MPa]
14
12
10
8
6
4
2
0
Centre Longitudinal Transverse Corner
Tire Load Locations
For the HMA bases there are two critical performance parameters, the horizontal strain at the bottom
of the HMA layer and the vertical strain at the bottom of the granular layers. The horizontal strain has
a direct empirical relation to fatigue cracking life, where lower developed strain results in longer
pavement life. The results noted from this analysis are shown in Figure 5-31 and are compared to the
strain developed with the same structural base model without the panel. This figure shows that under
the different load conditions the horizontal strain either remains the same or decreases versus the
99
control, so installing a solar road panel on an asphalt pavement structure increases the fatigue
cracking life of the asphalt pavement base.
290
Max HMA Layer Horizontal Strain
280
270
[micro-strain]
260
250
240
230
Center Longitudinal Transverse Corner
Tire Load Locations
In terms of the vertical compressive strain on the Granular B layer in the HMA base an even greater
performance improvement is realized by installing the solar road panel. Once again the relation
identified in Figure 4-20 indicates that lower strains result in higher rutting resistance, and in each
case the solar road panel installation reduces the vertical compressive strain realized in the Granular B
layer.
520
Max Granular B Layer Vertical Strain
510
500
[micro-strain]
490
480
470
460
Center Longitudinal Transverse Corner
Tire Load Locations
Figure 5-32: HMA base maximum vertical compressive strain in the Granular B layer
100
The vertical compressive strain on the Granular B layer in the granular base, as shown in Figure 5-33,
followed the same trend as identified within the HMA base; adding the solar road panel to the
structure increases the rutting resistance of the base in all load conditions.
840
Max Granular B Layer Vertical Strain
820
800
780
[micro-strain]
760
740
720
700
680
Center Longitudinal Transverse Corner
Tire Load Locations
Figure 5-33: Granular base maximum vertical compressive strain in the Granular B layer
101
traditional paving materials they use in the structural design, which reduces costs and potentially
improves the overall sustainability of pavements. The concrete base provides the largest opportunity
for optimization, so this will likely be the preferred base choice for solar road panel installations.
102
Chapter 6
Environmental Material Testing Results and Analysis
103
the raw extension and load output for the first tested glass sample, where the peak represents the load
at which the sample failed through brittle fracture.
300
250
Compressive Load (N)
200
150
100
50
0
0 0.5 1 1.5 2
Compressive Extension (mm)
Since glass is a Hookean material, the straight sloping profile until failure was expected for each
sample. What was not expected was the dip in the compressive load at approximately 95-N of
loading. This dip was apparent in all of the testing data for all of the materials, so it was determined
that the dip was caused by settling that was occurring in the testing rig repeatedly at this load. In order
to determine the flexural modulus of the materials accurately, the starting point for the flexural
measurement was taken after the dip occurred with the ending point being taken from later in the
straight portion of the curve in a consistent manner for each material. All of the calculations based on
this data were performed as outlined in ASTM D-790.
104
700
600
400
300
200
100
0
0 2 4 6 8 10 12 14 16
Compressive Extension (mm)
The data output from the testing is summarized in Table 6-1. This data demonstrates that the
average flexural modulus remained relatively consistent despite the scaling and freeze-thaw effects.
These results are further backed by the very low relative standard deviations, demonstrating the
resilience and consistency of the material tested.
The one high RSD, 7.0% for the 25th cycle of the scaled set, is due to a sample that had an
abnormally high modulus value from testing. Analysis of the results and specimen could not
determine a cause for this variation so the sample was still included in the results. Removal of this
sample from the testing pool only reduced the average modulus by 0.08 GPa and the RSD by 2.0%.
105
nature, so the material accepts the load elastically until it reaches its brittle failure point and then
shatters. The typical glass specimen failed under flexural load as demonstrated in Figure 6-4.
Figure 6-4: Typical glass 3-point bending test failure (Specimen G-15-US-3)
As shown in Figure 6-4, there was a dominant side from which the failure cracks originated. This
was consistent across all of the samples as there was a marginal alignment error between the load
nose and the samples under the bending test.
One of the challenges with working with glass is that there is a high variability in the distribution
of particles in the glass matrix. This results in some unpredictability in the properties of the materials
as the particle distribution also affects where the dislocations are in the material that are where failure
cracks will propagate from. This is apparent in the initial results from the 3-point bending testing,
which are demonstrated in Table 6-2.
Freeze-Thaw Cycles
Conditioning Measurement
0 5 10 15 25 50
Average Flexural Modulus (GPa) 47.5 49.3 47.2 42.2 48.9 45.8
Normal
RSD (%) 6.3 4.2 6.2 43 4.8 19
Average Flexural Modulus (GPa) 47.5 47.5 51.2 38.1 50.5 47.6
Scaled
RSD (%) 6.3 3.7 2.6 22 3.8 2.3
While most of the data set had low RSD values, the 15th and 50th normal cycles and the15th
scaled cycle stand out for their very high RSD values of 43%, 19%, and 22% respectively. While
there were no discrepancies in the failure mechanisms of the samples in G-15-S, both of the other sets
had an odd sample failure that implied an unusual defect in the sample being tested. Figure 6-5
demonstrates these two failure mechanisms, where sample G-15-N-1 broke in a near perfect shear
failure and sample G-50-N-1 broke in a more dramatic bending failure than the typical sample
identified in Figure 6-4.
106
Figure 6-5: Irregular failures of samples a) G-15-N-1 and b) G-50-N-1
With these samples eliminated as outliers, the updated results from the flexural modulus testing
are shown in Table 6-3. While the RSD for 15th scaled cycle remains high, the RSD values for 15th
and 50th normal cycles have been reduced to 3.9% and 2.3% respectively. This indicates a much more
accurate sampling of real flexural modulus data.
107
800
700
Compressive Load (N)
600
500
400
300
200
100
0
0 5 10 15 20 25 30 35
Compressive Extension (mm)
The figure demonstrates that the glass specimens are much more fragile than both of the polymer
options, with failure occurring at around 50-MPa of flexural stress and at well under 1-mm of
displacement. However, it is important to note that minimal deflection was outlined as a design
requirement for a solar road module so that the solar cells would not be damaged from vehicle traffic.
The stress at which the glass fails is an equivalent stress to the acrylic and polycarbonate samples at
approximately 4-mm and 5-mm respectively.
108
120
100
60
40
20
0
0 5 10 15 20 25 30
Compressive Extension (mm)
Not only does the lower rigidity of the polymer options impact deflection over the solar cells, it
also places a greater emphasis on the rigidity of the panel being based in the structural layer materials.
In the current prototype design this is inverted as the rigidity is largely based on the tempered glass
optical layer with the fiberglass structural layers acting as support. In a design with acrylic or
polycarbonate as the optical layer, aluminum, steel, or some other structural metal would likely be
required for the structural layers to make the entire panel rigid enough to minimize stresses on the
solar cells.
2480
2460
Flexural Modulus (MPa)
2440
2420
2400
2380
2360
2340
0 10 20 30 40 50
Freeze-Thaw Cycles
Figure 6-8 demonstrates that the scaled specimens perform better than the normal specimens,
which is the opposite of what was believed would occur. The general drop in performance and then
steady increase of both sets is likely due to an adjustment of the polymer chains within the material
that causes an immediate weakening and then steadily strengthens the material over time. The
difference in the magnitude of this change between the normal and scaled samples is due to the way
the normal samples were conditioned; in air as opposed to submerged. It is believed that the salt brine
provided greater thermal insulation to those specimens during the cycling, so they saw less dramatic
temperature induced polymer chain relocations. Further study would be required to validate this
hypothesis.
Also notable from Figure 6-8 is that the ‘0’ freeze-thaw cycle control set appears to be an outlier
from determining a strong linear correlation between freeze-thaw cycles and flexural modulus for
both conditions. This is highlighted below in Table 6-6 which shows the results of determining a
linear regression fit to both the total and ‘0’-excluded polycarbonate data sets. While the complete set
110
of data does not demonstrate a statistically significant regression, both of the conditions with the ‘0’
set excluded are statistically significant. The ‘0’-excluded data set and regressions are shown in Table
6-6.
Table 6-6: Polycarbonate 3-point bending regression analysis
Data Set
Condition Intercept Variable P-Value R-squared
Control All2390 0.65 0.57 0.09
Scaled 2416 0.66 0.40 0.18
‘0’ Excluded Control 2354 1.75 0.02 0.88
Scaled 2394 1.33 0.04 0.79
2480
2460
Flexural Modulus (MPa)
2440
2420
2400
2380
2360
2340
0 10 20 30 40 50
Freeze-Thaw Cycles
Figure 6-9: Polycarbonate flexural modulus results, excluding '0' cycle data
For the most part the RSD values are low, though for the 10th and 15th cycle data these values are
consistently above 10% and in some cases come as high as 30%. This is related to the fact that it is a
111
small range and any variation in the BPN will have a large impact due to the already low average
BPN values, though further studies should include a larger sample to minimize this variation.
Figure 6-10 shows the comparative results of the friction testing between the three material sets.
This chart shows that the performance of all of the materials was very similar, with a slightly higher
BPN being noted for the polycarbonate samples over the first 15 cycles.
16
British Pdneulum Number (BPN)
14
12
10
8
6
4
0 10 20 30 40 50
Freeze-Thaw Cycles
112
A two-factor ANOVA was run for each of these materials and the results of this can be found
below in Table 6-8. While in this analysis there was no statistical difference between the two types of
conditioning, statistical differences were found for the freeze-thaw effect in all cases. It should be
noted that the conditioning of the polycarbonate samples almost demonstrated a significant
difference.
The trend established from the results of the acrylic testing is shown in Figure 6-11, where both
trendlines have a clear increase of friction over increasing freeze-thaw cycling but that there is
essentially no difference between the two conditioning methods. This increase, as there is no change
between conditioning methods, is suspected to be related to temperature effects resulting in variations
in the microtexture on the surface of the material.
18
16
British Pendulum Number
14
12
10
8
6
4
2
0
0 10 20 30 40 50
Freeze-Thaw Cycles
113
The glass friction results are demonstrated in Figure 6-12 and show the steady increase of friction
with freeze thaw cycling. While in this case the control samples tend to have higher friction readings,
there was no statistically significant trend identified from the data.
14
British Pendulum Number
12
10
8
6
4
2
0
0 10 20 30 40 50
Freeze-Thaw Cycles
The polycarbonate friction results, as shown in Figure 6-13, demonstrate that while the scaled
samples continually demonstrate higher friction than the normal samples, the results of the two are
still too similar to identify a difference between them. Overall there is still an increase in friction with
increasing freeze-thaw cycling. Similar to the acrylic samples, the increasing friction is likely due to
an impact of temperature cycling on the surface microtexture of the material.
16
British Pendulum Number
14
12
10
8
6
4
2
0
0 10 20 30 40 50
Freeze-Thaw Cycles
The very low RSD values demonstrated in Table 6-10 show that the values obtained from the
testing are an accurate representation of the actual data. The highest RSD values are for the
absorptivity measurements, though these high values are a result of summation errors from
determining the absorptivity values.
115
Table 6-11 shows the results of a two variable ANOVA study on the transmissivity, reflectivity,
and absorptivity properties of the acrylic samples. This study found that conditioning was a
significant factor for both transmissivity and absorptivity while freeze-thawing was only significant
for transmissivity, though it was close for absorptivity as well.
The results of Table 6-11 are further outlined below in Figure 6-14, which plots the average
transmissivity values and the linear regression fits to this data. This figure also shows that the scaled
samples showed a lower drop in transmissivity than the normal samples did, and this is believed to be
due to the temperature effects outlined for the polycarbonate specimens in the three point bending
testing analysis. The high variability of the scaled transmission values merits further analysis.
89.8
89.6
89.4
Transmissivity (%)
89.2
89.0
88.8
88.6
88.4
88.2
88.0
87.8
87.6
0 10 20 30 40 50
Freeze-Thaw Cycles
116
The graph for the average reflectivity values, as shown in Figure 6-15, shows the high variability
of the results that was indicated by the ANOVA analysis.
7.5
Reflectivity (%) 7.5
7.4
7.4
7.3
7.3
7.2
7.2
0 10 20 30 40 50
Freeze-Thaw Cycles
The results of the absorptivity calculation, as shown in Figure 6-16, show the close regression fits
for increasing absorptivity with increasing freeze-thaw cycles. The result that the normal samples
have a higher absorptivity than the scaled samples is believed to be the result of a polymer chain
realignment as outlined in the transmissivity section. Overall the increase is small though, registering
at approximately one percent over the 50 freeze-thaw cycles for each set of specimens.
6.0
5.0
Absorptivity (%)
4.0
3.0
2.0
1.0
0.0
0 10 20 30 40 50
Freeze-Thaw Cycles
117
6.4.2 Glass Results and Analysis
The average results for transmission and reflection for the glass samples are shown in Table 6-12
along with the calculated absorption coefficient and the calculated RSD values for each parameter.
The overall results are similar to those found for the acrylic samples; the test values are relatively
consistent across the range of testing and the RSD values are all low except for the absorptivity
measurements where compounding errors are present.
Table 6-13 shows the results of a two variable ANOVA study on the transmissivity, reflectivity,
and absorptivity properties of the glass samples. This study found that there were no significant
correlations between conditioning method or freeze-thaw cycling and any optical property of the
material. This is due to the overall consistent response of the optical properties during testing.
Figure 6-17 demonstrates the profile obtained from the transmissivity testing. Both the normal
and scaled specimen sets produced relatively flat regression fits across the 50 cycles, though the
scaled set saw slightly greater variability.
118
80.8
80.6
80.4
Transmissivity (%)
80.2
80.0
79.8
79.6
79.4
79.2
79.0
78.8
78.6
0 10 20 30 40 50
Freeze-Thaw Cycles
Figure 6-18 shows the reflectivity results from the glass samples. The scaled specimens produced
an almost perfectly flat curve fit while the control set of data shows a decrease, but by only 0.08%
across the 50 cycles. This demonstrates that in both cases the control and scaled average values are a
good representation of each set of data.
7.85
7.80
Reflectivity (%)
7.75
7.70
7.65
7.60
0 10 20 30 40 50
Freeze-Thaw Cycles
119
Figure 6-19 shows the profiles that were obtained from determining the absorptivity of the glass
specimens. While this graph shows the scaled samples as more absorptive, the previous ANOVA
analysis determined that there is statistically no impact from the conditioning method.
14.0
13.5
Absorptivity (%)
13.0
12.5
12.0
11.5
0 10 20 30 40 50
Freeze-Thaw Cycles
120
Like the acrylic and glass specimen sets before it, the polycarbonate samples demonstrate
consistent test results and low RSD values.
Table 6-15 shows the results of a two variable ANOVA study on the transmissivity, reflectivity,
and absorptivity properties of the glass samples. This study found that the only significant effect is of
conditioning on the reflectivity data, but the effect of conditioning on absorptivity and the freeze-thaw
cycling on both transmissivity and absorptivity come close to being significant.
Figure 6-20 shows the results from the transmissivity testing on the polycarbonate specimens.
This data shows the same fluctuation that the three point bending results did in Figure 6-8, where the
transmissivity increases by a substantial margin at first before dropping off with repeated freeze-thaw
cycles. This is also believed to be caused by a realignment of the polymer chains in the material.
83.0
82.5
Transmissivity (%)
82.0
81.5
81.0
80.5
80.0
0 10 20 30 40 50
Freeze-Thaw Cycles
121
As noted from the ANOVA analysis, there is little difference between the two conditioning
methods until the 50th cycle, causing the lack of statistical difference between these data sets. It is
clear though that there is a trend occurring across different freeze-thaw cycles, however the relative
plateau of results between 5 and 15 cycles causes the evaluation to fall just outside of statistical
significance.
Figure 6-21 shows the results of the reflectivity testing, and demonstrates that there is a clear
difference between the two conditioning methods. The normal samples show an overall lower
reflectivity than that of the scaled samples, which is likely connected to a combination of the polymer
chain realignment and the slightly greater impact of the salt brine on the surface microtexture. This
trend was also noted on the friction results from the polycarbonate samples, as noted in Figure 6-13.
9.0
8.9
Reflectivity (%)
8.8
8.7
8.6
8.5
8.4
0 10 20 30 40 50
Freeze-Thaw Cycles
Figure 6-22 demonstrates the absorptivity of the polycarbonate samples and note the difference
between the two conditioning methods and the freeze-thaw cycling. The scaled set of samples is less
affected overall than the control ones, though these differences are due to the summation of
measurement errors.
122
samples were, overall, slightly higher than the polycarbonate results. The transmissivity is the most
important factor in determining the efficiency of the total solar road panel efficiency, as a higher
value means more incident radiation on the panel will reach the solar cells, making acrylic the best
option from this perspective.
11.5
11.0
10.5
Absorptivity (%)
10.0
9.5
9.0
8.5
8.0
0 10 20 30 40 50
Freeze-Thaw Cycles
90
88
Transmissivity (%)
86
84
82
80
78
0 10 20 30 40 50
Freeze-Thaw Cycles
123
Figure 6-24 shows the comparison between the average reflectivity values of the specimens used in
the optical testing. From this chart it is apparent that the polycarbonate specimens have the highest
reflectivity, followed by the glass specimens, and the acrylic specimens have the lowest. In terms of
transportation engineering reflectivity is a very important safety property; roads that are too reflective
produce glare that can distract drivers and cause accidents. From this perspective both the acrylic and
glass materials provide good options in comparison to polycarbonate.
9.0
8.5
Reflectivity (%)
8.0
7.5
7.0
0 10 20 30 40 50
Freeze-Thaw Cycles
Finally, Figure 6-25 shows the average calculated values for absorptivity of the materials used in the
environmental testing. Here it is clear that the glass specimens are the most absorptive, followed by
the polycarbonate samples, and that the acrylic samples are far less absorptive. The energy absorbed
by these materials is going to correspond directly to the temperature of the optical layer of a solar
road panel, so lower values are more ideal. That being said, the structural layer beneath the optical
layer will be absorbing far more radiation and will heat the optical layer through conduction. The
benefit of choosing a material with low absorptivity should be studied further.
124
16
14
12
Absorptivity (%)
10
8
6
4
2
0
0 10 20 30 40 50
Freeze-Thaw Cycles
Freeze thaw cycling was observed to have a much bigger impact on the performance of these
materials. In every case friction increased as a result of freeze-thaw cycling, the flexural modulus of
polycarbonate increased, and the transmissivity of acrylic decreased. Again glass proved to be the
most inert to the conditioning.
As a result, it was found that a properly designed glass surface would make for the best possible
transparent layer of a solar road panel.
125
Chapter 7
Conclusions and Recommendations
7.1 Conclusions
Overall, this research proved the hypothesis and completed the objectives outlined; it is possible to
build a solar road panel that can withstand the structural and environmental loads that Ontario
pavements are subjected to. It was also found that the prototype designed for this research meets all of
these criteria with no adjustments.
This combination of material places the emphasis for the rigidity of the panel on the transparent
layer. Other configurations, with different transparent layer materials, would likely need rigid
structural layers in order to minimize the strains that the solar cells are subjected to. These other
configurations would also suffer from rutting issues similar to those of asphalt pavements, though the
rutting would occur over the solar cell compartments. This would likely greatly increase the
roughness of pavements, damage vehicles, and increase the susceptibility of the surface to
hydroplaning.
The prototype designed for this research demonstrated an infinite design life under both fatigue
and static load conditions, which indicates that optimizations can be made to reduce the volume of
material needed in each panel and this will directly lower the cost of the structures. This is especially
126
true when a concrete structural base is considered, as the threshold stresses for failure of the concrete
are also far from being reached from these load conditions.
It was also found that this panel was safely designed to ensure that the solar cells will not fail due
to strain cycling. This finding clears one of the main concerns of the traditional solar module industry,
which is highly focused on the strain induced microcracking of solar cells due to the flexural
performance of their encapsulating solar module. Especially in the cases with concrete or asphalt
bases there is room for design optimization before the strain fatigue criteria would be met.
Greater impacts were found from the freeze-thaw cycling of the materials. In particular it was
found that this conditioning increased the friction available on the surface of all of the specimens in a
linear fashion over the 50-cycles of testing. This would be a major benefit to solar road panels as
these transparent surfaces have very poor friction characteristics without textures and coatings, so
improvements that can be gained through normal operation are highly beneficial.
Overall it was found that glass proved to be the best transparent material to deal with the
environmental conditions subjected to Ontario’s pavement infrastructure. It’s inertness to the
environmental conditioning found in this study confirmed the prototype design decision made to use
this material for its structural benefits over acrylic and polycarbonate.
7.2 Recommendations
The recommendations consist of two main categories; future research that continues the work
outlined in this thesis and additional areas of study in order to make solar road panels a viable
product.
Structural testing of solar road panel prototypes should be expanded to include in-situ testing. If
frictional issues cannot be resolved easily this could be mitigated by testing the panels in a single
wheel path, allowing traction to be supplied by all of the wheels not loading the panel. This testing
would allow for better analysis of fatigue effects on the edges in the optical layer and validate the
work shown here in the fatigue and static simulations in general.
Work should be conducted on the wet and dry friction characteristics of coated and uncoated
glass surfaces. It is highly likely that in order to develop enough friction for safe travel on a tempered
glass surface that both macro- and micro-textural improvements will be required; these are developed
through surface texturing and surface coating. Different types of high-impact resistant glasses should
also be examined for their applicability to this research.
The pavement load case analysis performed in this research should be updated to include the
viscoelastic characteristics of asphalt and granular materials as structural bases. It is likely that the
deterioration of these layers over time will increase the stresses seen in the solar road panel and, as a
result, will decrease the design life of the panel.
The research suggested above on the frictional characteristics of textured and coated glasses
should also include an optimization for the optical properties of the panel as a whole. The texturing
and coating have the potential to increase the reflectivity of the surface, but also the air pocket
between the glass and the solar cell encapsulant should be considered to determine what the net
electricity generation of these panels could be.
Research should also be done on the costs of solar road panels and their related infrastructure,
with a focus on the return of investment from installing the infrastructure. Aspects of this should
include panel design, structural base design, associated grid connecting hardware, and revenue from
selling the electricity or offsetting buying electricity from the grid.
128
References
AASHTO. (1993). AASHTO guide for design of pavement structures. (). Washington, DC: AASHTO.
ACI. (2013). Matweb material property data. Retrieved May 13, 2013, from
http://www.matweb.com/index.aspx
Adams, W. M. (2006). The future of sustainability: Re-thinking environment and development in the
twenty-first century. ().IUCN.
Agro, S., & Tucker, R. (2004). Development of new low-cost, high-performance, PV module
encapsulant/packaging materials: Annual technical progress report. ( No. SR-520-35683).
Golden: NREL.
Alagusundaramoorthy, P., Harik, I. E., & Choo, C. C. (2006). Structural behavior of FRP composite
bridge deck panels. Journal of Bridge Engineering, July/August, 384.
Alsop, D. J. A., & Saunders, R. J. (1999). Structural use of glass in buildings. London: Institute of
Structural Engineers.
ARA. (2011). Methodology for the development of equivalent pavement structural design matrix for
municipal roadways. ( No. 000830). Toronto, Ontario: RMCAO.
Askeland, D., & Phule, P. (2006). The science and engineering of materials (5th ed.). Toronto:
Thomson.
ASTM. (2009). Standard test method for determining solar or photopic reflectance, transmittance,
and absorptance of materials using a large diameter integrating sphere. ( No. ASTM E1175-
87).American Society for Testing and Materials.
ASTM. (2010). Standard test methods for flexural properties of unreinforced and reinforced plastics
and electrical insulating materials. ( No. ASTM D790-10).American Society for Testing and
Materials.
129
ASTM. (2012a). Sandard test method for solar absorptance, reflectance, and transmittance of
materials using integrating spheres. ( No. ASTM E903-12).American Society for Testing and
Materials.
ASTM. (2012b). Standard test method for scaling resistance evaluation of concrete surfaces exposed
to deicing chemicals. ( No. ASTM C672-12).American Society for Testing and Materials.
ASTM. (2013). Standard test methods for total normal emittance of surfaces using inspection-meter
techniques. ( No. ASTM E408-13).American Society for Testing and Materials.
Bergamin, L., & Sammaraee, T. (2010). Waved glass: Towards optimal light distribution on solar cell
surfaces for high efficient modules. Solar Energy, 84(1), 90-100.
Budynas, R. G., & Nisbett, J. K. (2008). Shigley's mechanical engineering design (8th ed.). New
York: McGraw Hill.
Caliendo, C., & Parisi, A. (2010). Stress-prediction model for airport pavements with jointed concrete
slabs. Journal of Transportation Engineering, July, 664.
Cho, Y. H., McCullough, B. F., & Weissmann, J. (1996). Considerations on finite-element method
application in pavement structural analysis. Transportation Research Record: Journal of the
Transportation Research Board, 1539, 96.
Chung, K., Chang, K., & Liu, Y. (2008). Reduction of wind uplift of a solar collector model. Journal
of Wind Engineering and Industrial Aerodynamics, 96(8), 1294-1306.
Costello, S. B., Bargh, L. S., Henning, T. F. P., & Hendry, M. (2013). Proposed new performance
indicator - vehicle operation cost index (VOCi) due to road roughness. Transportation Research
Board 92nd Annual Meeting, Washington, D.C.
Demers, C. E. (1998). Fatigue strength degredation of E-glass FRP composites and carbon FRP
composites. Construction and Building Materials, 12(1998), 311.
Deubener, J., Helsch, G., Moiseev, A., & Bornhöft, H. (2009). Glasses for solar energy conversion
systems. Journal of the European Ceramic Society, 29(7), 1203-1210.
130
El Amrani, A., Mahrane, A., Moussa, F., & Boukennous, Y. (2007). Solar module fabrication.
International Journal of Photoenergy, 2007
Enfiinty. (2011). Europe’s first “green” train leaves the station thanks to Belgium’s solar tunnel.
Retrieved July 5, 2012, from http://www.enfinitycorp.com/downloads/news-releases/europes-
first-green-train
Environment Canada. (2013). Canadian climate normals 1971-2000: Waterloo wellington airport.
Ottawa: Environment Canada.
Foster, D., & Anderson, M. (2003). Rapid forward deployment made easier with composite airfield
matting. AMPTIAC Quarterly, 7(1), 17-22.
Grand Canyon National Park. (2013). Grand canyon sky walk. Retrieved August 1, 2013, from
http://www.grandcanyon-nationalpark.org/grand-canyon-skywalk/
Greene, J., Toros, U., Kim, S., Byron, T., & Choubane, B. (2010). Impact of wide-base single tires on
pavement damage. Transportation Research Record: Journal of the Transportation Research
Board, 2155, 82.
Holley Jr, W., Agro, S., Galica, J., & Yorgensen, R. (1996). UV stability and module testing of
nonbrowning experimental PV encapsulants. Photovoltaic Specialists Conference, 1996.,
Conference Record of the Twenty Fifth IEEE, 1259-1262.
Hsieh, H. H., Lin, F. M., & Yu, S. P. (2011). Performance of low series-resistance interconnections
on the polycrystalline solar cells. Solar Energy Materials and Solar Cells, 95(1), 39-44.
Huang, Y. H. (2004). Pavement analysis and design (2nd ed.). New Jersey: Pearson Prentice Hall.
IEA. (2011). IEA statistics: CO2 emissions from fuel combustion: 2011 edition. (). Paris: International
Energy Agency.
ISG. (2013). ISG glass flooring system details. Retrieved September, 5, 2013, from
http://www.structuralglass.com/eledet/ISG-GF-A.pdf
131
Jorgensen, G., Terwilliger, K., DelCueto, J., Glick, S., Kempe, M., Pankow, J., . . . McMahon, T.
(2006). Moisture transport, adhesion, and corrosion protection of PV module packaging
materials. Solar Energy Materials and Solar Cells, 90(16), 2739-2775.
Kempe, M. D., & Thapa, P. (2008). Low-cost single-layer replacement for the back-sheet and
encapsulant layers. Society of Photo-Optical Instrumentation Engineers (SPIE) Conference
Series, , 7048 12.
KNMI. (2013). Long term average 1981-2010 - average yearly precipitation. De Bilt: Royal
Netherlands Meteorological Institute.
Lange, R. F. M., Luo, Y., Polo, R., & Zahnd, J. (2011). The lamination of (multi) crystalline and thin
film based photovoltaic modules. Progress in Photovoltaics: Research and Applications, 19(2),
127-133.
Luque, A., & Hegedus, S. (2003). Handbook of photovoltaic science and engineering. Chichester:
Wiley.
Mak, T. L. A. (2012). Modular road plate system. (Master of Applied Science, University of
Waterloo).
MTO. (1990). Pavement design and rehabilitation manual. Toronto: Ontario Ministry of
Transportation.
MTO. (2012). Ontario's default parameters for AASHTOWare pavement ME design interim report.
(). Toronto: Ontario Ministry of Transportation.
Neamen, D. A. (2003). Semiconductor physics and devices: Basic principles (3rd ed.). New York:
McGraw-Hill.
Northmore, A., & Tighe, S. (2012a). Developing innovative roads using solar technologies. CSCE 9th
International Transportatoin Specialty Conference, Edmonton.
Northmore, A., & Tighe, S. (2012b). Innovative pavement design: Are solar roads feasible? 2012
Conference of the Transportation Association of Canada, Fredericton.
NRC. (2012). Photovoltaic potential and solar resouce maps of canada. Retrieved February 8, 2013,
from http://pv.nrcan.gc.ca/
132
OPSS. (2003). Material specification for aggregates - base, subbase, select subgrade, and backfill
material. ( No. OPSS 1010). Toronto: Ontario Provincial Standard Specification.
Perret-Aebi, L. E., Li, H. -., Chapuis, V. & Heinstein, P. (2011). PV module design. Retrieved May
30, 2012, from http://pvlab.epfl.ch/pv_module_design
Pokharel, S. K., Han, J., Manandhar, C., Yang, X., Leshchinsky, D., Halahmi, I., & Parsons, R. L.
(2011). Accelerated pavement testing of geocell-reinforced unpaved roads over weak subgrade.
Transportation Research Record: Journal of the Transportation Research Board, 2204, 67-75.
Ritchie, R. O. (2003). Feilure of silicon: Crack formation and propogation. 13th Workshop on
Crystalline Solar Cell Materials and Processes, Vail, Colorado.
RMSolar. (2012). Mono 125S0R2 solar cell. Retrieved March 14, 2012, from
http://www.rmsolarpanel.com/html/49/2010-12-13/content-34.html
Roark, R. J., & Young, W. C. (1975). Formulas for stress and strain (5th ed.). New York: McGraw-
Hill.
Robinson, M. C. (2005). Landing mat development at WES. In B. M. Fowle (Ed.), Builders and
fighters (pp. 195-206) University Press of the Pacific.
Rochling. (2013). Grade UTR Arc/Track & flame resistant laminate. (). Cleveland: Rochling Glastic
Composites.
Rollings, R. S. (1975). Comparison of the british class 60 trackway and AM-2 mat for bomb damage
repair applications. ( No. AFWL-TR-75-149). Washington, D.C.: NTIS.
Rose, D., Jester, T., & Bunea, G. (2008). Automated manufacturing of high efficiency modules.
Rushing, T. S., & Tingle, J. S. (2009). Full-scale evaluation of mat surfacings for roads over sand
subgrades. Journal of Terramechanics, 46(2009), 57-63.
Sánchez-Illescas, P., Carpena, P., Bernaola-Galván, P., Sidrach-de-Cardona, M., Coronado, A., &
Álvarez, J. (2008). An analysis of geometrical shapes for PV module glass encapsulation. Solar
Energy Materials and Solar Cells, 92(3), 323-331.
Skoplaki, E., & Palyvos, J. (2009). On the temperature dependence of photovoltaic module electrical
performance: A review of efficiency/power correlations. Solar Energy, 83(5), 614-624.
133
Solar Roadways. (2013). Solar roadways - A real solution. Retrieved February 13, 2013, from
http://solarroadways.com/main.html
Statistics Canada. (2012). Electric power generation, by class of electricity producer, annual
(megawatt hour). ( No. Table 127-0007).Statistics Canada.
TAC. (2012). Pavement asset design and management guide. Ottawa: Transportation Association of
Canada.
Tingle, J. S., & Jersey, S. R. (2005). Cyclic plate load TEsting of geosynthetic-reinforced unbound
aggregate roads. Transportation Research Record: Journal of the Transportation Research
Board, 1936, 60-69.
TNO. (2013). SolaRoad combines road and solar cells. Retrieved February 13, 2013, from
http://www.tno.nl/content.cfm?context=thema&content=inno_case&laag1=895&laag2=912&ite
m_id=1234
Transport Canada. (2008). Transportation in canada 2007: An overview. ( No. TP 14816E). Ottawa:
Transport Canada.
Ulery, H. H., & Wolf, D. P. (1971). Thickness requirements for soils beneath landing mats; bare base
support. ( No. S-71-3). Washington, D.C.: NTIS.
US EIA. (2011). International energy outlook 2011. (). Washington, D.C.: U.S. Energy Information
Administration.
Walbridge, S., & de la Chevrotiere, A. (2012). Opportunities for the use of aluminum in vehicular
bridge construction. (). Montreal: MAADI Group.
Webster, S. L., & Tingle, J. S. (1998). Espedient road construction over sands using lightweight
mats. ( No. N-1472). Washington, D.C.: NTIS.
134
World Commission on Environment and Development. (1987). Our common future. (). Oxford:
Oxford University Press.
Wu, G., & Yu, X. (2012). Thermal energy harvesting across pavement structure. Transportation
Research Board 91st Annual Meeting, Washington DC.
Wu, Z., Chen, X., & Yang, X. (2011). Finite element simulation of structrual performance on flexible
pavements with stabilized Base/Treated subbase materials under accelerated loading. ( No.
FHWA/LA.10/452). Baton Rouge: LTRC.
Xia, K. (2010). A finite element model for Tire/Pavement interaction: Application to predicting
pavement damage. International Journal of Pavement Research and Technology, 3(3), 135.
135
Appendix A
Solar Road Panel Prototype Design Drawings
136
137
138
139
140
141
142
143
144
145
Appendix B
Solar Road Panel Prototype Bill of Materials
146
147
Appendix C
Structural Testing Frame Design Drawings
148
149
150
151
152
153
154
155
Appendix D
Three-Point Bending Apparatus Fixture Drawings
156
157
158
159
160
161
162
163
Appendix E
Prototype Model Validation Results
164
165
Appendix F
Pavement Load Case Model Validation Results
166
167
168
169
170