RENEWABLE ENERGY AND CONSERVATION MEASURES FOR
NON-RESIDENTIAL BUILDINGS
By
Andrew James Grossman

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Mechanical Engineering – Master of Science
2014

ABSTRACT
RENEWABLE ENERGY AND CONSERVATION MEASURES FOR NON-RESIDENTIAL
BUILDINGS
By
Andrew James Grossman
The energy demand in most countries is growing at an alarming rate and identifying economically
feasible building retrofit solutions to decrease the need for fossil fuels so as to mitigate their
environmental and societal impacts has become imperative. Two approaches are available for identifying
feasible retrofit solutions: 1) the implementation of energy conservation measures; and 2) the production
of energy from renewable sources. This thesis focuses on the development of retrofit software planning
tools for the implementation of solar photovoltaic systems, and lighting system retrofits for mid-Michigan
institutional buildings.
The solar planning tool exploits the existing blueprint of a building’s rooftop, and via image
processing, the layouts of the solar photovoltaic arrays are developed based on the building’s
geographical location and typical weather patterns. The resulting energy generation of a PV system is
estimated and is utilized to determine levelized energy costs.
The lighting system retrofit analysis starts by a current utilization assessment of a building to
determine the amount of energy used by the lighting system. Several LED lighting options are evaluated
on the basis of color correlation temperature, color rendering index, energy consumption, and financial
feasibility, to determine a retrofit solution.
Solar photovoltaic installations in mid-Michigan are not yet financially feasible, but with the
anticipated growth and dynamic complexity of the solar photovoltaic market, this solar planning tool is
able to assist building proprietors make executive decisions regarding their energy usage. Additionally, a
lighting system retrofit is shown to have significant financial and health benefits.

To my Grandpa, Sheridan Norburg, and my parents, Phil and Nancy Grossman, for all of their love and
support

iii

ACKNOWLEDGEMENTS

I would like to thank, first and foremost, the students at SPARTA who helped make this
possible with all of their hard work and dedication. A special thanks to Kathryn Meersseman for
helping me with the edits of countless papers, including this thesis. I would also like to thank my
advisors Dr. Richard Lunt and Dr. André Bénard; I could not do it without them. Huge thanks to Dr.
Albert Cohen and Tom McCollum for the economic analysis help. Thank you to Dr. Dirk Luchini

Colbry for help with the extensive image processing. I would lastly like give a special shout out
to beer. Thank you for getting me through graduate school. I owe you.

iv

TABLE OF CONTENTS

LIST OF TABLES ...................................................................................................................................... vii
LIST OF FIGURES ..................................................................................................................................... ix

CHAPTER 1 INTRODUCTION ............................................................................................................. 1
CHAPTER 2 PHOTOVOLTAIC OVERVIEW .................................................................................... 3
2.1 Solar Photovoltaic History ........................................................................................................... 3
2.2 Solar Cells and the Photovoltaic Effect ...................................................................................... 4
2.3 Performance of Solar Cell ............................................................................................................ 5
2.3.1 Maximum Power Point ....................................................................................................... 5
2.3.2 Module Efficiency ............................................................................................................... 5
2.3.3 Fill Factor ............................................................................................................................. 6
2.3.4 Temperature Effect .............................................................................................................. 6
2.3.5 Irradiance Effect .................................................................................................................. 7
2.3.6 Optical Loss.......................................................................................................................... 7
2.3.7 Degradation .......................................................................................................................... 8
2.3.8 Dirt, Dust, Snow, and Wind ............................................................................................... 9
2.4 Common Photovoltaic Technologies .......................................................................................... 9
2.4.1 Crystalline Silicon Technology ....................................................................................... 11
2.4.1.1 Monocrystalline Silicon Technology .................................................................... 11
2.4.1.2 Polycrystalline Silicon Technology ...................................................................... 11
2.4.2 Thin Film Technology ...................................................................................................... 12
2.4.2.1 Amorphous and Microform Silicon Technology ................................................ 13
2.4.2.2 Cadmium Telluride Technology ............................................................................ 13
2.4.2.3 Copper Indium Selenium Technology .................................................................. 14
2.5 Photovoltaic Systems .................................................................................................................. 15
2.6 Review of Photovoltaic Modeling Tools ................................................................................. 16
2.6.1 Pre-Feasibility Tools ......................................................................................................... 16
2.6.1.1 Financial Analysis Tool for Electrical Energy Projects (FATE2-P)................. 17
2.6.1.2 RETScreen International ........................................................................................ 17
2.6.2 Sizing Tools........................................................................................................................ 17
2.6.2.1 Hybrid Optimization of Multiple Energy Resources (HOMER) ...................... 17
2.6.2.2 Hybrid Designer ....................................................................................................... 17
2.6.2.3 PVSYST.................................................................................................................... 18
2.6.3 Simulation Tools................................................................................................................ 18
2.6.4 Shortcomings of the Current Solar Photovoltaic Modeling Tools .............................. 19
2.7 United States Solar Photovoltaic Market ................................................................................. 19
2.7.1 Incentives and Initiatives .................................................................................................. 21
2.7.2 Overview of the Bottom-Up Installed Price Benchmark ............................................. 22
2.7.2.1 Bottom-Up Installed 2010 Benchmark Prices ..................................................... 25
v

CHAPTER 3 FORMULATION OF THE PHOTOVOLTAIC SYSTEM COMPOSITION ......... 27
3.1 Image Processing of the Blueprint ............................................................................................ 29
3.1.1 Extracting the Roof Regions ............................................................................................ 30
3.1.2 Depth Map .......................................................................................................................... 31
3.1.3 Roof Shade Modeling ....................................................................................................... 31
3.1.4 Pseudo-3D Mapping ......................................................................................................... 33
3.2 Layout of the Modules ................................................................................................................ 34
3.3 Initial Energy Comparison & Inverter Characteristics ........................................................... 38
3.4 Electrical Connections ................................................................................................................ 39
3.5 Sizing of the Balance of Systems .............................................................................................. 39
3.6 Simulation of the Energy Generation Capabilities ................................................................. 41
3.7 Case Studies ................................................................................................................................. 43
3.7.1 Case Ia: Validation of the Solar PV System Design Tool ........................................... 44
3.7.2 Case Ib: Energy Capabilities of the Solar PV System Design Tool ........................... 46
3.7.3 Case II: Complex Roof Geometries of the Solar PV System Design Tool ............... 47
3.8 Discussion .................................................................................................................................... 53
3.9 Summary and Conclusions......................................................................................................... 54
CHAPTER 4 FLUORESCENT TO LED LIGHTING RETROFIT ................................................. 55
4.1 Phase I: Preliminary Assessment .............................................................................................. 56
4.2 Phase II: Data Procurement ....................................................................................................... 57
4.2.1 Building Schedule & Utility Billing History ................................................................. 57
4.2.2 Building Spaces & Inventory Survey ............................................................................. 58
4.2.3 Energy Audit ...................................................................................................................... 59
4.3 Phase III: Engineering ................................................................................................................ 60
4.3.1 Characteristics of Light..................................................................................................... 61
4.3.2 Retrofit Options ................................................................................................................. 64
4.4 Phases IV-VI ................................................................................................................................ 65
4.5 Case Studies ................................................................................................................................. 66
4.5.1 Lighting Quality................................................................................................................. 66
4.5.2 Financial Feasibility Study ............................................................................................... 69
4.6 Discussion .................................................................................................................................... 71
4.7 Summary and Conclusions......................................................................................................... 71
APPENDICES .......................................................................................................................................... 72
Appendix I: Solar Photovoltaic Time Periods ............................................................................... 73
Appendix II: Lighting Power Density for Common Spaces Around a University ................... 75
Appendix III: Lighting Technologies.............................................................................................. 76
Appendix IV: Economic Fundamentals .......................................................................................... 82
BIBLIOGRAPHY .................................................................................................................................... 87

vi

LIST OF TABLES

Table 2.1. Overall crystalline silicon degradation rate per year........................................................... 9
Table 2.2. Simulation software tools available .................................................................................... 18
Table 2.3. DOE SunShot Initiative benchmark and target prices ...................................................... 21
Table 3.1. PV system component derate factors for DC to AC conversion at STC ........................ 38
Table 3.2. Installed module characteristics under standard testing conditions ................................ 44
Table 3.3. Design comparison of an existing solar PV system at the MSU Surplus Store vs. the
solar PV design tool ................................................................................................................................. 45
Table 3.4. Comparison at the MSU Surplus Store of the solar PV system design for highest
energy generation per module and highest energy generation per region ........................................ 47
Table 3.5. Total area of each region and area of each region after shade is included ..................... 48
Table 3.6. The layout of the modules on the roof the MSU Psychology Building, for the highest
energy generation per module on a yearly basis, and highest energy generation per region on a
yearly basis ................................................................................................................................................ 50
Table 3.7. Comparison at the MSU Psychology Building of the solar PV system design for
highest energy generation per module and highest energy generation per region ........................... 50
Table 3.8. Economic evaluation of the solar PV system installations for the highest energy
generation per module and highest energy generation per region ..................................................... 51
Table 3.9.Amortization payment summary for the solar PV system installations that correspond
to the highest energy generation per module and highest energy generation per region ................ 53
Table 4.1. Overview of the six steps for a successful lighting system retrofit................................. 56
Table 4.2. Indicators for determining the viability of an economically feasible retrofit ................ 57
Table 4.3. Walk-through times used to quantify energy usage .......................................................... 59
Table 4.4. IESNA preferred color temperatures .................................................................................. 63
Table 4.5. CCT and CRI parameter comparison of the current F32T8 lighting technology .......... 67
Table 4.6. CCT and CRI parameter comparison of the perspective LED technology .................... 67
vii

Table 4.7. A comparison of the two lighting technologies that are being assessed for the financial
feasibility of a retrofit .............................................................................................................................. 69
Table 4.8. Estimated hourly usage on an annual basis ........................................................................ 69
Table 4.9. Financial characteristics for the lighting retrofit of the three building spaces .............. 70
Table 4.10. Economic feasibility comparison of the two lighting technologies .............................. 71
Table A.1. Time periods for BoM and expected periods for replacement ....................................... 73
Table A.2. Time Periods for Labor Costs and Expected Periods for Replacement......................... 73
Table A.3. Indirect capital costs time period ........................................................................................ 74
Table A.4. Discrete indirect costs .......................................................................................................... 74
Table A.5. Module time periods............................................................................................................ 74
Table A.6. The ASHRAE/ IESNA/ ANSI 90.1 LPD for the common space types around a
university ................................................................................................................................................... 75

viii

LIST OF FIGURES

Figure 2.1. The light that encounters the cell creates electron-hole pairs, which are separated by p
and n-type layer barrier, thus creating a voltage that moves a current through an external circuit.
For interpretation of the references to color in this and all other figures, the reader is referred to
the electron version of this thesis. ............................................................................................................ 4
Figure 2.2. Typical I-V and P-V curve, as well as the key points of these curves for a solar cell
[10] ............................................................................................................................................................... 5
Figure 2.3. The fill factor, defined by the green area and divided by the gray area [10] ................. 6
Figure 2.4. The effects of the ambient temperature on a solar cell’s current and voltage [10] ....... 7
Figure 2.5. The effects of the solar irradiance on a solar cell’s current and voltage [10] ................ 7
Figure 2.6. The brown area of the PV cell is caused by UV radiation degrading the ethyl vinyl
acetate layer [12] ........................................................................................................................................ 8
Figure 2.7. Various solar cell technologies and their respective record for efficiency [17] .......... 10
Figure 2.8. Single junction PV module Layers[10] ............................................................................. 10
Figure 2.9. Monocrystalline solar cell [20] ........................................................................................... 11
Figure 2.10. Polycrystalline solar cell [20] ........................................................................................... 12
Figure 2.11. Layer structure of an amorphous solar cell deposited on a glass substrate [18, 24] . 13
Figure 2.12. Layer structure of a cadmium telluride solar cell [18, 24] ........................................... 14
Figure 2.13. Layer structure of a copper indium gallium selenide solar cell [18, 24] .................... 14
Figure 2.14. Components of a grid-tied PV system, coupled with the power flow ........................ 15
Figure 2.15. The 2000-2013 U.S. Solar PV installations and weighted average system price [33]
.................................................................................................................................................................... 20

Figure 2.16. The 2011 and 2013 state-by-state solar PV installations by capacity [16, 33] .......... 20
Figure 2.17. 2010 commercial rooftop PV system element breakdown [40] ................................... 26
Figure 3.1. Flow chart illustrating the methodology employed for a solar PV system design that
can be tailored to a building rooftop ...................................................................................................... 28
ix

Figure 3.2. Image corresponding to connected components matrix, where each shade represents a
different object region. There are sixty-three unique object regions depicted ................................. 30
Figure 3.3. Various sun angles used to calculate the shade ................................................................ 32
Figure 3.4. Images representing the matrices incorporated in 2.5D mapping. The lighter shades in
c and d correspond to larger pixel-by-pixel values .............................................................................. 34
Figure 3.5. Representation of a viable panel region ............................................................................ 35
Figure 3.6. Description of the vectors employed in the layout of the modules ............................... 36
Figure 3.7. Illustration of how two string subintervals may be connected in series ....................... 36
Figure 3.8. Michigan State University Surplus Store characteristics. The current PV System is
located on green highlighted roof section in 3.8a. The average monthly energy demand is in
section 3.8b. .............................................................................................................................................. 44
Figure 3.9. Monthly breakdown of the existing PV system compared to the solar PV planning
tool .............................................................................................................................................................. 46
Figure 3.10. Characteristics of the Michigan State University Psychology Building. The labels 17 in 3.10a correspond to the regions employed in the layout of the modules. The average monthly
energy consumption from 2007-2013 is shown in 3.10b. ................................................................... 48
Figure 3.11. Economic evaluation of the solar PV system installations for a.) highest energy
generation per module and b.) highest energy generation per region ............................................... 52
Figure 4.1. Data input form for each space ........................................................................................... 58
Figure 4.2. Data input form for each fixture for each room ............................................................... 59
Figure 4.3. Flowchart for determining the average daily energy usage of each fixture ................. 60
Figure 4.4. 1931 CIE chromaticity chart [63] ...................................................................................... 61
Figure 4.5. Color temperature chart courtesy of Sylvania Lighting .................................................. 62
Figure 4.6. The IES eight pastel reference colors used for quantifying the chromaticity shift [64]
.................................................................................................................................................................... 63

Figure 4.7. 2010 commercial building sector estimate of installed lamps technology ................... 64
Figure 4.8. The light quality tests for the current F32T8 bulbs. The CCT was measured off the
spectra in a, and in b, the CCT was declared ........................................................................................ 67
Figure 4.9. The light quality tests for the perspective LED retrofit kit. The CCT was measured
off the spectra in a, and the CCT was declared in b ............................................................................ 68
x

Figure 4.10. Side-by-side color comparison, where the current F32T8 lighting technology is
shown in a, and the perspective LED lighting technology is shown in b ......................................... 68
Figure 4.11. Economic evaluation of the three building spaces. The breakeven time period is
where the current lighting technology intersects with the proposed LED technology ................... 70
Figure A. 1. Typical incandescent light bulb ................................................................................ 76
Figure A. 2. Typical halogen light bulb ........................................................................................ 77
Figure A. 3. Typical compact fluorescent light bulb .................................................................... 79
Figure A. 4. Typical fluorescent tube light ................................................................................... 80
Figure A. 5. Typical LED diode ................................................................................................... 81

xi

Chapter 1 INTRODUCTION
The energy demand in many prospering countries is growing at an alarming rate. At the present
time, about 80 percent of the world’s electricity and heating needs come from fossil fuels such as coal and
natural gas. It is well known that these fossil fuels are a large contributor to air pollution, which ultimately
leads to the effects of global warming. Nonetheless, these fuels are often times the preferable choice as
opposed to renewable sources because of their reliability, present abundance, and well-established
extraction and refining techniques. However, unlike solar and wind energy, a finite quantity of fossil fuel
exists which makes them vulnerable to market fluctuations. In 2013, about 63 percent or roughly 2.2
trillion barrels of the oil were imported in the United States (US). This equated to about $218 trillion in
foreign exchange. The US accounts for 4.5 percent of the world’s population yet consumes 25 percent of
the global oil production [1].
Nuclear power in the US was attributed to roughly 19.4 percent of the US electricity consumption
in 2013. Nuclear power has many advantages such as lower greenhouse gas emissions and high energy
density in comparison to conventional fossil fuels. However, the disadvantages of nuclear power include
radioactive waste by-products, which poses a threat to human health, potential cataclysmic consequences
if system failures occur, and it requires a large amount of water. In March 2011, Japan’s Fukushima
Daiichi nuclear power plant encountered severe damaged from an earthquake and a tsunami. As a result,
the radioactive fallout was not only increased in Japan, but also scientists reported increased radioactivity
in the US. The Fukushima nuclear disaster was second largest in history, only behind the 1986 Chernobyl
disaster [1, 2].

In regards to energy consumption and the environment, buildings play significant role.
For instance, 40 percent of the US’s energy consumption is attributed to the nation’s 120 million
buildings, where 72 percent of that energy is electricity consumption. Furthermore, it is projected
that from 2007 to 2025, the total energy consumption due to buildings is expected to increase by
50 percent. The main reason behind this growth is that new buildings are being constructed
1

before old buildings are retired. Given this expected growth, along with the finite quantities of
fossil fuel available, brings about the possibility of an energy paradigm. Therefore, identifying
economically feasible building retrofit solutions that decrease the need for fossil fuels has
become imperative [3, 4].
The United States Department of Energy (DOE) defines retrofit as a process of
improving and/or upgrading a building’s infrastructure such that significant energy cost savings
are achieved. Typically, a building retrofit is separated in two categories: 1) energy conservation
measures, and 2) energy production by means of renewable sources [3]. Accordingly, the focus
of this thesis is separated into two sections, where the first section focuses on the development of
a solar photovoltaic (PV) planning tool that takes a ubiquitous approach towards utilizing
rooftops, while the focal point of the second section presents a methodology for analyzing a
lighting system retrofit while working on a limited budget.
The next chapter will give an overview of the basic PV characteristics, as well as an
examination of the current PV system modeling tools. The subsequent section discusses a design
methodology for developing a PV planning, which is tested and verified on two case studies. The
last chapter will discuss a method for a lighting system retrofit.

2

Chapter 2 PHOTOVOLTAIC OVERVIEW
The intent of this chapter is to provide a brief overview of solar photovoltaics. It begins
with a discussion regarding its history, which is followed by a summary of the PV effect in a
single-junction solar cell. Next, the performance parameters of a solar cell are reviewed, along
with the common PV technologies. The subsequent sections present a compendium of PV
systems and modeling tools, which is concluded by a summary of the PV market.

2.1 Solar Photovoltaic History
Since the 7th century, people have investigated techniques for harnessing and converting solar
energy into useful applications. Initial applications entailed creating fire by concentrating the sun’s energy
via glass and mirrors. Today, solar energy is still utilized to light fires, but since the discovery of the PV
effect in 1839, by French physicist Edmond Becquerel (1820-1891), solar energy is used for an
assortment of applications ranging from charging an iPod to powering vehicles and buildings. When
Becquerel was 19, he recognized that when two electrodes are exposed to light, electricity is generated,
where the amount generated is proportional to the sun intensity. His experimental setup consisted of a
black box that contained an acidic solution and housed the electrodes that were either coated with AgCl or
AgBr [5, 6]. In 1887, Heinrich Hertz (1857-1894) discovered that ultraviolet light influences the lowest
voltage capable of inducing a spark to jump between two electrodes [7]. However, the PV effect remained
an anomaly of the science community for the next three quarters of a century. Through the 1940s and
1950s, research was conducted to develop a process that will create highly pure crystalline silicon. This
process is known as the Czochralski process, which led to the cultivation of commercializing PV cells. In
1954, Scientists at Bell Laboratories depended on the Czochralski process to develop a crystalline silicon
PV cell, with an efficiency of six percent [7]. Today, the main research focus for solar cells are crystalline
silicon, thin-film, multi-junction, organic and dye-sensitized [5, 6].

3

2.2 Solar Cells and the Photovoltaic Effect
The PV effect, depicted in Figure 2.1, is the physical process by which a solar cell converts
sunlight into electricity. The semiconductor of a silicon solar cell consists of mainly two layers, n-type
layer and p-type layer, and the barrier where these two layers meet is the pn junction. The n-type layer is
negatively charged and is formed by doping with impurity atoms that have an excess valence electron,
which is typically phosphorous. The p-type layer is positively charged and is formed by doping with
impurity atoms that have one less valence electron, which is typically boron [8, 9].
When sunlight collides with the silicon solar cell, one of three things may occur: 1) the photon
passes right through the silicon, 2) the photon is reflected at the surface of the silicon, or 3) the photon is
absorbed by the silicon. If the photon is absorbed and has enough energy to free an electron, the negative
electron charge flows from the n-type layer, via an external electric circuit, to a load such as a light bulb
and then to the p-type layer, where they recombine with holes. This flow of electrons is how the electrical
current is produced within the solar cell. The charge separation between the n and p-type layer at the
junction creates a voltage, which drives the current through the external circuit [8, 9].

Figure 2.1. The light that encounters the cell creates electron-hole pairs, which are separated by p and ntype layer barrier, thus creating a voltage that moves a current through an external circuit. For
interpretation of the references to color in this and all other figures, the reader is referred to the electron
version of this thesis.
4

2.3 Performance of Solar Cell
The maximum power point, fill factor, and efficiency of a solar cell are influenced by a copious
amount of factors during conversion. These factors are often the result of many climatic conditions such
as irradiance, ambient temperature, snow, wind, and dust.

2.3.1 Maximum Power Point
Figure 2.2 depicts a typical current-voltage (I-V) and power-voltage (P-V) curve for a solar cell,
where at the “knee” of the curve denotes a solar cells maximum power point Pmp. This is found by
multiplying both the short circuit current at the maximum power point Imp with the voltage at the
maximum power point Vmp. The electric output of a module is dependent upon the module efficiency and
technology [10].

Figure 2.2. Typical I-V and P-V curve, as well as the key points of these curves for a solar cell [10]

2.3.2 Module Efficiency
The efficiency of a PV cell ηPV is defined as the ratio of the energy output from the solar PV cell
with respect to the solar global irradiance G at standard testing conditions (STC) entering the cell, i.e.
(1)
where A is the is the area of the solar PV cell. The STC are an irradiance 1000 W/m2, an air mass (AM) of
1.5 spectral distribution, and a 25˚C solar PV cell temperature. The efficiency is a vital performance
factor for the situation when space is limited and a certain amount of power is needed [8].
5

2.3.3 Fill Factor
Figure 2.3 depicts I-V curve, where the fill factor FF is the green area. The fill factor is defined as
a ratio of the maximum power of the solar cell to the product of the open circuit voltage VOC and short
circuit current ISC, in other words
(2)
An ideal solar cell would produce a perfectly rectangular I-V curve, where the short circuit current and
open circuit voltage coincide with the current and voltage at the maximum power point. The importance
of the FF is it’s comparative abilities. If comparing two solar cells with the same ISC and VOC, the module
with the higher FF will produce more power [10].

Figure 2.3. The fill factor, defined by the green area and divided by the gray area [10]

2.3.4 Temperature Effect
Figure 2.4 depicts how the ambient temperature affects a solar cell’s performance. When the
ambient temperature is greater than the STC temperature, 25˚C, the short circuit current slightly increases,
whereas the open circuit voltage is dramatically reduced. In other words, the current increase is
proportional to a voltage decrease. Overall, when the temperature is above the STC temperature, the
overall power output of the module is reduced [11].

6

Figure 2.4. The effects of the ambient temperature on a solar cell’s current and voltage [10]

2.3.5 Irradiance Effect
Figure 2.5 depicts how the solar irradiance affects a solar cell’s current and voltage. The short
circuit current of a solar cell is linearly proportional to irradiance, whereas the open circuit voltage of a
PV module is logarithmic with respect to irradiance [11].

Figure 2.5. The effects of the solar irradiance on a solar cell’s current and voltage [10]

2.3.6 Optical Loss
There are two main types of irradiation that can enter a solar cell, direct and diffuse,
where direct irradiance is mainly responsible for generating power. The diffuse irradiance is less
intense compared to the direct irradiance because it encounters obstructions or absorptions before

7

reaching the module. Some of these obstructions may be due to atmospheric conditions,
surrounding landscape features, while other obstructions or absorptions may be due to the
module’s glass cover, anti-reflective solar cell coating layers or ethyl vinyl acetate encapsulant [12].
2.3.7 Degradation
The main cause of a PV module degrading overtime is due to UV radiation diminishing the ethyl
vinyl acetate encapsulant and the back sheet layers. These layers are discussed in Section 2.4.

Figure 2.6. The brown area of the PV cell is caused by UV radiation degrading the ethyl vinyl acetate
layer [12]
The UV radiation causes these materials to lose elasticity, and results in moisture reaching the PV cells.
This moisture causes an electrical resistance within the connections, and thus, the output voltage and
current decreases. Other degradation factors include lamination disintegration of the backing material,
solder-joint wear, hot spots, and discoloration as depicted in Figure 2.6. NREL has classified degradation
into five categories:
1. Degradation of packing materials
2. Degradation of cell/module interconnects
3. Degradation caused by moisture intrusion
4. Degradation of the semiconductor device
5. Loss of adhesion
Further, the degradation of a PV module differs by the type of technology and region, depending on the
climate and weather. Overall, crystalline silicon has an average annual module degradation rate of 0.360.64% and the degradation of the entire PV system ranges from 0.23-0.90%, where a breakdown is shown
Table 2.1 [12, 13].
8

Table 2.1. Overall crystalline silicon degradation rate per year
Technology
Configuration
Degradation Rate per Year
Module
0.36-0.47%
Monocrystalline technology
System
0.23-0.90%
Module
0.61-0.64%
Polycrystalline Technology
System
0.59-0.60%

2.3.8 Dirt, Dust, Snow, and Wind
Typically, wind increases the performance of a PV module because the convection reduces the
operating cell temperature. However, dirt, dust and snow can cause optical losses and partial shading.
When a module experiences partial shading, the shaded modules act as a load by consuming the power
from other modules. This load consumption causes harmful excessive heat, which results in failure before
the expected lifetime [14]. On the other hand, if the module is not covered with snow, the diffuse solar
energy may reflect of the surrounding snow, which may cause more irradiance entering the module, thus
increasing the energy production [13, 15].

2.4 Common Photovoltaic Technologies
Although a substantial amount of research is currently being performed on a multitude of PV
technologies, as portrayed in Figure 2.7, the single junction PV technology is currently the most cost
effective, and accounted for 87% of the global PV market sales in 2011 [16].

9

Figure 2.7. Various solar cell technologies and their respective record for efficiency [17]
The main layers of a single junction PV module are depicted in Figure 2.8. The front surface the
module is primarily low iron glass. This type of glass is used because it has low reflectivity, high
transmissivity, and self-cleaning properties, as well as its impermeability to water and gas, and it is
relatively inexpensive. The next layer is a transparent ethyl vinyl acetate film that encapsulates the solar
cells within the module. This film is used because its properties have a strong bond to the solar cells and it
is stable at different operating temperatures. The back surface is a thin polymer sheet, typically tedlar.
Tedlar is used because its properties have low thermal resistivity and low water/water vapor permeability.
The solar cells, which was described in Section 2.2, are responsible for producing power [18].

Figure 2.8. Single junction PV module Layers[10]
10

2.4.1 Crystalline Silicon Technology
Crystalline silicon PV cells have many benefits including being the highest rated efficiency in
comparison to other single pn junction modules on the market. Additionally, silicon is a very abundant
element, and because this is a mature technology, a considerable amount of research has been performed
regarding its reliability and robust design [19, 20].

2.4.1.1 Monocrystalline Silicon Technology
Of all the crystalline silicon solar cell technologies on the market, monocrystalline is not only the
oldest, but is also the most efficient with an average efficiency range of 16-18%, and a maximum reported
efficency of 25% [20]. The solar cell has a thickness of about 0.2-0.3 mm, and has many shapes such as
round, semi-round, and square. The square cells offer the highest packing density in the module but are
the most expensive, whereas round cells are the cheapest because less material is wasted during
production. The color of cell is either black if it contains an anti-reflective coating, shown in Figure 2.9,
or gray if it does not contain an anti-reflective coating [19, 20].

Figure 2.9. Monocrystalline solar cell [20]

2.4.1.2 Polycrystalline Silicon Technology
Polycrystalline solar cells were developed as a means to reduce production costs, and in 2008,
they accounted for 48% of the world’s solar cell production. The silicon wafers within the module are
fabricated ingots, which were casted from molten silicon. Polycrystalline silicon PV technologies have a
typical efficiency range of 15-17%, but have been proven to be as efficient as 20.4% [20]. The color of

11

cell is either bluish-black, if it contains an anti-reflective coating, shown in Figure 2.10, or can be green,
gold, silver, brown, or violet if it does not contain an anti-reflective coating [19, 20].

Figure 2.10. Polycrystalline solar cell [20]

2.4.2 Thin Film Technology
Over the last few years, thin film technology has received a considerable amount of attention due
to its wide range of applications, aesthetic qualities, and functional value. The 2011-2016 market
predictions anticipate an annual growth of 27.5%, which may be associated with the fact that this
technology requires less energy and production materials to manufacture them in comparison to
crystalline technology, thus making them cheaper [21, 22].
There are many attributes regarding thin film photovoltaics. For instance, this technology can be
utilized on any surface, including curved surfaces, because it can be deposited on flexible substrates. The
common applications are roof shingles and calculators, as well as less common applications, such as
windows, metal, plastic and paper [22]. Another notable quality is that this technology can be designed
for specific angle dependence properties that increase the responsivity at off-normal solar incidence
angles. Furthermore, thin film photovoltaics experience a performance increase at high ambient
temperatures, whereas crystalline technology encounters a performance decrease [22].
The typical semiconductor materials are amorphous and microform silicon (a-si and μ-si),
cadmium telluride (CdTe), and copper indium selenide and copper indium gallium selenide (CIS and
CIGS) [22].

12

2.4.2.1 Amorphous and Microform Silicon Technology
Amorphous and microform silicon technology is the first thin technology, and commonly used in
small electronic devices such as hand-held calculators. Unlike traditional bulk-crystalline silicon
technology, the silicon atoms are in a non-order arrangement and the active material is 1/300th of the
thickness. However, the efficiency of amorphous and microform silicon ranges from 12 to 13%, and they
have a shorter lifetime in comparison to crystalline silicon technology [22, 23]. The layer structure of an
amorphous solar cell is shown in Figure 2.11, where the transparent conducting oxide is usually SnO2.

Figure 2.11. Layer structure of an amorphous solar cell deposited on a glass substrate [18, 24]

2.4.2.2 Cadmium Telluride Technology
Cadmium telluride technology is the first and only thin film technology that surpasses crystalline
technology in price per watt peak, and is the second most widely used solar cell material. The highest
efficiency reported has been 20.4 percent, whereas the typical efficiency ranges from 9 to 11% [22].
Although cadmium telluride technology is known for its ease of manufacturing and efficiency, there are
many drawbacks that hinder large-scale commercialization. For instance, cadmium is an abundant but
toxic element, which consequently affects the disposal and long-term safety. Furthermore, tellurium is a
rare earth element [22, 23]. The layer structure of cadmium telluride solar cell in depicted in Figure 2.12.

13

Figure 2.12. Layer structure of a cadmium telluride solar cell [18, 24]

2.4.2.3 Copper Indium Selenium Technology
Copper indium selenium technology is the most efficient thin film solar cell at 20.8%, which is
partially due to it having a relatively high responsivity at low and indirect sun light [22]. In comparison,
crystalline silicon technology has better heat resistance, and unlike cadmium telluride technology, this
technology does not require any toxic materials. In regards to large-scale commercialization, the copper
indium selenium technology materials are much more difficult to manufacture, thus hindering its
introduction into the general market [22, 23]. Figure 2.13 shows the layer structure of a copper indium
gallium selenide solar cell.

Figure 2.13. Layer structure of a copper indium gallium selenide solar cell [18, 24]

14

2.5 Photovoltaic Systems
There are two main types of PV systems: 1) off-grid and 2) grid-tied. Off-grid PV systems are
typically installed in remote areas and are a stand-alone source of electrical energy. In addition, off-grid
PV systems can be coupled with another energy source such as wind or a diesel generator, which is
known as a hybrid system. Grid-tied PV systems are connected to the utility power distribution and may
be used to supply power to the utility power grid or supplement the utility power needs of the building.
Furthermore, grid-tied PV systems have the advantage of lower capital cost because these systems avoid
the need for energy storage components. In addition, avoiding the need for energy storage components
eliminates energy storage losses, thus making grid-tied PV systems more effective [8, 18, 24, 25].

Figure 2.14. Components of a grid-tied PV system, coupled with the power flow
In Figure 2.14, the typical components of a grid-tied PV system are shown, along with arrows
indicating the direction of the power flow. A PV module is comprised of solar cells connected in series to
build voltage and modules are further connected in series and parallel connections. The characteristic
properties of a module depend on the type and quantity of solar cells, and how the solar cells are
electrically connected. In a conventional PV system, modules are connected in series to form strings, such
15

that the operating voltage of the system is increased. The power produced from the strings is transmitted
through fuses to a power inverter. Combiner boxes are typically used to converge the power from the
strings onto a single interconnection to consolidate wiring. In doing so, the current from the system is
increased based on the amount of strings entering/exiting the combiner box. An inverter converts the DC
power of the modules into AC power, which is then transmitted to the power grid, or load application [8,
18, 24, 25]. This is further explained in Chapter 3.

2.6 Review of Photovoltaic Modeling Tools
Designing and simulating a solar PV system is rather straightforward. Energy flows from the
modules, through a series of components, to the source needing power. Currently, numerous software
tools are available for analyzing and dimensioning a PV system. These software packages range from
exceptionally basic to exceedingly sophisticated, and may be classified into three categories: 1) prefeasibility, 2) sizing, and 3) simulation [26, 27].
This section presents an overview of the software tools that are currently available on the market,
however, because there are a large number of tools, this list is not exhaustive, the tools are not ranked or
rated, but distinguishing attributes are noted.

2.6.1 Pre-Feasibility Tools
A pre-feasibility tool, such as FATE2-P and RETScreen, automates calculations that typically an
engineer would do by hand. The main attribute of these tools is their ability to perform a first time
assessment at a specified location for possible energy generation and levelized cost of energy. Often
times, these tools are implemented as spreadsheets, because minimal iterative work is required and
calculations may be automated. Advanced pre-feasibility tools have macros or code embedded within the
spreadsheet to provide a more detailed analysis. Generally, energy planners, financiers, vendors, and other
people who need a quick evaluation use these tools [26, 27].

16

2.6.1.1 Financial Analysis Tool for Electrical Energy Projects (FATE2-P)
FATE2-P is a financial analysis spreadsheet based tool, developed by NREL and Princeton
Economic Research Inc. This tool has the ability to take into account financial consideration such as debt,
tax credits and government incentives [26, 27, 28].

2.6.1.2 RETScreen International
RETScreen International is a decision support, renewable energy awareness, and financial
incentive spreadsheet based tool. At the core of this tool, it standardizes multiple renewable energy
designs, such that it may output the plausible energy generation, greenhouse gas reduction, and levelized
cost of energy [26, 27, 29].

2.6.2 Sizing Tools
Based on the specified energy requirement of a PV system, a sizing tool will determine the most
advantageous size of the various components. This could be constituted upon the size of the components
that yield the most favorable energy generation, while other tools attempt to determine the size of the
components that will yield the minimal levelized cost of energy. Sizing tools, such as HOMER, Hybrid
Designer, and PVSYST, are often times compiled as software packages that have a user-friendly
interface. Generally, system installers use these tools [26, 27].

2.6.2.1 Hybrid Optimization of Multiple Energy Resources (HOMER)
HOMER is a software program that is capable of modeling hybrid renewable microgrids for a
range of energy sources. It was originally owned by NREL, but is now a private entity. HOMER’s
strength comes from its ability to perform multiple iterations for numerous system configurations and
component sizes [26, 27, 30].

2.6.2.2 Hybrid Designer
Hybrid Designer is a free software program, developed by the Energy Research Centre at Cape
Town University. This program utilizes a genetic algorithm to evaluate multiple system configurations

17

that provide acceptable reliability, in order to determine the system that has the optimal levelized cost of
energy [26, 27, 31].

2.6.2.3 PVSYST
PVSYST is one of the most commonly used software programs for designing a PV system. This
program is able to combine multiple BOS components with a pre-feasibility study, such that hourly
simulations of a complete PV system are performed. In doing so, the user selects components from the
built-in database, and the software calculates the size of each component. However, PVSYST cannot
analyze multiple systems at one time in order to find the case that yields the optimal levelized cost of
energy. Therefore, the user usually performs a sensitivity and risk of investment analysis [26, 27, 32].

2.6.3 Simulation Tools
Currently, copious amounts of software tools that can simulate a PV system are available. The
tools all function in a similar pattern, where the user characterizes each component, various loads,
location, etc. Following, the program will simulate the system and outputs the energy generation and
expected consumption as a function of time. Some tools, such as Hybrid2, PVSYST, and PV-DesignProS also compute a financial analysis [24, 26, 27]. Table 2.2 is a summary of some of the simulation tools
available.
Table 2.2. Simulation software tools available
Software Tool

Source
Consortium formed by NMRC (Ireland), ARMINES (France), ICI (Romania), and
Ashling
IMIO (Poland)
Hybrid2
NREL (USA)
INSEL
University of Oldenburg (Germany)
PV-DesignPro-S Maui Solar Energy Software Corporation (USA)
PVSYST
Geneva University (Switzerland)
RAPSIM
Murdoch University (Australia)
SAU/AREA
Cardiff University (United Kingdom)
SOMES
Utrecht University (Netherlands)
SOLSIM
IAF/EWIS (Germany)
WATSUN-PV University of Waterloo (Canada)

18

2.6.4 Shortcomings of the Current Solar Photovoltaic Modeling Tools
Pre-feasibility tools such as RETScreen are advantageous because they have the ability to
compare multiple renewable energy technologies. However, these programs usually do not have a current
financial and product database, which may cause a dilemma when comparing two or more similar
technologies. Sizing tools such as PVSYST are beneficial because of the considerable modeling
capabilities. However, some of these programs require extensive knowledge about the BOS, and are not
always intuitive, and in some cases, these tools are incapable of incorporating a hybrid PV system. While
there are many simulation tools available that can accurately quantify the plausible energy generation,
these tools do not have the ability to simulate and compare multiple PV system designs at one time.
Furthermore, a program can couple sizing and simulation, such as the idea behind PVSYST, but also
provides the layout is advantageous for engineers, dealers, etc [26, 27].

2.7 United States Solar Photovoltaic Market
The recent trends in the solar PV market are astonishing. Referring to Figure 2.15, the amount of
installations in the United States increased from 4 MW in 2000 to almost 4.8 GW in 2013. This quick
growth in installations has been driven by government policies favoring renewable energy and a sharp
decline in costs. From 2012 to 2013, the capacity-weighted average installed costs decreased from
$3.04/W to $2.59/W, which is a 15% decline. However, this market is extremely dynamic, meaning that
the installed prices have a large variance from project-to-project and state-to-state. For instance, the 2013
common residential prices stretched from just under $3.00/W to just under $7.00/W, whereas nonresidential prices range from $1.70/W to $8.00/W, and utility prices range from $1.68/W to just over
$3.00/W [33].

19

4751

$12.00

$10.00

4000

3369
$8.00
3000
$6.00
1919

2000

$4.00
852

1000
4

11

22

45

58

79 105 160

298

435

0

$2.00

Weighted Average System Price ($/W)

Annual U.S. PV installations (MW)

5000

$0.00

Residential

Non-Residential

Utility

Figure 2.15. The 2000-2013 U.S. Solar PV installations and weighted average system price [33]
While the US PV market has been exponentially growing each year, the number of installations remains
relatively concentrated in a few key states, as represented in Figure 2.16.
2011 Installation By State

2013 Installations By State

Other
States
19%

Other
States
28%

California
29%

New
Jersey
5%
Massachus
etts 5%

Colorado
5%
New
Mexico
6%

Arizona
15%

New
Jersey
17%

North
Carolina
7%

California
55%
Arizona
9%

Figure 2.16. The 2011 and 2013 state-by-state solar PV installations by capacity [16, 33]

20

The solar PV industry consists of two markets, domestic and international, where the
international market had a substantial influence on the reduced solar PV system costs. This began in 2007
when new restrictions pertaining to the production of raw silicon came into effect.

By 2011,

manufacturing worldwide increased by 100-fold, where China and Taiwan were responsible for
producing 61% of the global PV module supply. However, this high level of production also contributed
to numerous bankruptcies and consolidations, which eventually led to an investigation from the
Department of Commerce. The investigation determined that the Chinese Government was subsidizing
the cost of crystalline-silicon modules, meaning that PV modules were sold below production costs. This
is known as ‘illegal-dumping’ and large tariffs were applied to Chinese modules as result [34].

2.7.1 Incentives and Initiatives
The federal, state, and local governments have grown increasingly aware of the economic,
environment, and societal benefits of implementing renewable energy. As a result, many incentives and
programs have been created such as the 2011 DOE SunShot Initiative. The DOE SunShot Initiative is a
collaborative nationwide initiative that aims to reduce the total costs of solar electricity by 75% from
2010-2020, thus making solar energy competitive with traditional technology. The study outlining the
initiative states that solar energy by 2030 may generate 14% of the US electricity needs and 27% by 2050.
This would result in an 8% decrease or 181 million metric tons of CO2 reduced by 2030, and 28% or 760
million metric tons by 2050. If the aforementioned scenario comes to fruition, 290,000 new solar jobs are
projected to be created by 2030, and 390,000 by 2050. The SunShot Initiative price targets and reference
targets are shown in Table 2.3, where these reference targets were calculated by Black & Veatch in 2012
[35].
Table 2.3. DOE SunShot Initiative benchmark and target prices
Benchmark
Reference
SunShot
Market
2010 Price
2020 Price
2020 Target Price
Utility Scale PV ($/WDC)
4.00
2.51
1.00
Commercial Rooftop PV ($/WDC)
5.00
3.36
1.25
Residential Rooftop PV ($/WDC)
6.00
3.78
1.50

21

A common way to facilitate renewable energy projects is renewable energy credits. A renewable
energy credit is much like purchasing clean electricity from the utility company, but this program is
voluntary, meaning it allows a business to choose what projects to support, i.e. wind, solar, geothermal,
etc. When renewable energy sources generate electricity, it is separated into two categories, electricity
produced and environmental benefits. Based on what the user wants, the user either chooses electricity
produced in MWh increments, or environmental attributes [36].
Another program is the Public Utility Regulatory Power Act of 1978 (PURPA), which requires
the local utility company to purchase any excess renewable energy from a grid-tied system, at a rate equal
to what it costs them to produce power. An additional program is third-party power purchasing
agreements, which are designed to create incentives for the owner, consumer, and utility company. The
DOE and the North Carolina Solar Center have created the Database of State Incentives for Renewables
& Efficiency (DSIRE) in order to track the programs that help meet the needs of the owner, consumer,
and utility company [37, 38].

2.7.2 Overview of the Bottom-Up Installed Price Benchmark
NREL and the US DOE published in February, 2012, the Residential, Commercial, and UtilityScale Photovoltaic (PV) System Prices in the United States: Current Drivers and Cost-Reduction
Opportunities. In this report, a range of solar PV costs using a bottom-up price analysis was assessed
using Monte Carlo simulations.

A Monte Carlo simulation is a statistical sampling method to

approximate solutions to quantitative problems. More specifically, a Monte Carlo simulation transcribes
uncertainties from the model inputs, by specifying the inputs as probability distributions, and creates
model outputs based on the inputs [39].
As mentioned previously, there has been a significant decrease in the price of solar PV systems
from either the global market players or governmental incentives. However, primarily due to the fair
market value consideration on system prices, system cost reductions are not recognized, nor recognized in
a timely manner. This is in large part due to the separation between installation costs, component prices,
and system prices. Additionally, fair market value does not provide clear enough prices for understanding
22

the drivers in the system price. Therefore, this section describes overnight capital costs for commercial
rooftop solar PV, developed from a bottom-up price analysis methodology [40].
For the analysis, NREL compiled a list of the estimated bill of materials (BoM) costs, which not
only included material related costs, but also installation labor costs. In the report, it stated that the
estimated direct labor costs were sensitive to changes in independent variables such as module and string
size, module efficiency, and other system design parameters. Furthermore, the list subdivided some
material categories such as wiring, where it included many components such as wire, conduit, and
connectors, but excluded installation labor because it was in another category. Additionally regarding
wiring materials, this category was broken down into two groups, DC and AC. A carpenter or general
laborer may install the DC wiring components, which contains all the components from the module to the
inverter, such as the combiner box and lighting rod, but excludes the inverter. A skilled electrician
typically installs the AC wiring components, which includes all the components from the combiner box to
the application, but excludes the utility substation if needed [40].
The pricing benchmark for rooftop commercial PV systems includes building types such as:
1. Architecturally unique buildings such as churches
2. Big-box offices
3. Office buildings more than two stories
4. Warehouse buildings one-two stories
5. Skyscrapers
Generally, a commercial rooftop has minimal to no slope and surface of asphalt, ballastedmembrane, or standing-seamed metal. The pricing benchmark for commercial rooftop PV systems
includes: architecturally unique buildings such as churches, big-box offices, office buildings taller than
two stories, skyscrapers, and warehouse buildings that are one to two stories. Furthermore, based on a
building’s preexisting features and region-specific system design requirements, a significant variance in
the cost and design of the system arises. An example of a design variance based on a building’s
preexisting features is a roof that has a inverted roof membrane, which requires an anchored-ballasted
23

roof racking structure for the PV system. Region-specific system design requirements may include
meeting municipal and county seismic standards. Alternatively, a region-specific system design
requirement is the load capacity of the building. For example, areas in the U.S. where snow is eminent,
buildings are already designed to be able to handle the load of a PV system; elsewhere areas that do not
typically receive snow have a roof with a much more limited weight bearing capacity. In culmination,
preexisting building features and region-specific system design requirements affect the design and
hardware of the system, and how the system is installed, thus affecting the overall cost of the system [40].
In this report, it was stated that an experienced eight-person crew, using the direct-attach method
for standing metal seam rooftops, can install up to 150 modules per day. In contrast, this experienced
eight-person crew, using the method for flashed point penetration for through-roof systems, where
approximately 0.07 penetrations per module are required, can install up to 25 modules per day. However,
this experienced eight-person crew, using the through-roof rail-type system, can install up to 600 modules
per day. Nevertheless, this experienced eight-person crew, depending on project staging techniques and
local building requirements, using the ballasted mounting system, can install up to 1,000 modules per day.
The ballasted system compared to the standing-seam mounting system, may provide 15-20% labor
savings. However, the ballasted system is generally limited to membrane type roofs, and low to no roof
slope [40].
The labor costs for the installation were calculated from the U.S. national average wage rates and
standard burden rates. The national average for an electrical contractor before insurance and benefits is
$49.00/hour, whereas the general contractor is roughly $33.00/hour. These averages took into account
payroll taxes, retirement, health and liability insurance, and operating overhead costs that were varied by
market sector, installer size and experience. The burden labor rates, included a 0.44% builders insurance,
6.2% federal and state unemployment insurance, 2.02% public liability insurance, 7.65% Social Security
taxes (FICA), and 6.4% worker compensation insurance. It is projected that as more solar PV installations
occur nationwide, large cost reductions in overhead and cost of trips for permits are to follow, as well as
more efficient means of gaining permits. Collaborating installers have reported that the economy-of-scale
24

and standardized ground-mount systems leads to utility scale installations having the lowest project
overhead costs per kW installed, followed by commercial rooftop installation, and then residential
systems [40].

2.7.2.1 Bottom-Up Installed 2010 Benchmark Prices
The Monte Carlo Simulation for commercial solar PV rooftop analyzed a three-story building
with a standing-seam system design. The simulation analyzed a system for 914 crystalline silicon
modules that were 0.992 m by 1.653 m, and had an efficiency of 14.5%. They were installed on a
standing-seam metal roof without through-roof penetration anchors, and were mounted using four clips
per module. The overhead costs used in the system total $12,000, which was the culmination of building
permits, commissioning and engineering expenses. The building permits, which included delays caused
by permit-related activities, and commissioning costs such as utility upgrades to the building’s electrical
panel, totaled roughly $2,000, whereas the engineering was estimated at $10,000. However, some areas
reported a permit fee as high as 10% of the project costs, which is not the case at Michigan State
University [40].
The result of the Monte Carlo Simulation determined the 2010 price benchmark for commercial
solar PV system rooftop of $4.59/W. Figure 2.17 details the 2010 element price benchmark breakdown,
where modules are 45% of the total system price and is the largest contributor. The second and third
largest contributors, 14% of the total system price, are the supply chain costs and installation materials,
followed by the inverter being 8% of the total system price [40].

25

Permitting &
Commissioning,
4%

Installer
Installer
Overhead , Profit, 2%
2%

Sales Tax, 6%

Module
Installation Materials

Supply Chain Costs
Inverter

Labor, 6%

Labor
Module, 45%

Inverter, 8%

Sales Tax
Permitting &
Commissioning
Installer Overhead

Supply Chain
Costs, 14%
Installation
Materials, 14%

Installer Profit

Figure 2.17. 2010 commercial rooftop PV system element breakdown [40]

26

Chapter 3 FORMULATION OF THE PHOTOVOLTAIC SYSTEM
COMPOSITION
With the anticipated growth and dynamic complexity of the solar photovoltaic (PV) market, along
with the large capital cost of PV systems, a need for a planning tool that will assist in the design of PV
systems is desired. A wide variety of planning tools are currently available, as discussed in Section 2.6,
for analyzing a PV system, where typically these tools are classified into three categories: pre-feasibility,
sizing, and simulation. A prefeasibility tool determines if a specified application is financially feasible,
while sizing tools establishes the size of the balance of system (BOS) components, and simulation tools
provide a detailed analysis of the expected energy generation by emulating the PV system [24]. However,
a robust planning tool does not exist that can take into account the complex geometry of a rooftop for
multiple PV modules and tilt angles, while being able to analyze an assortment of component databases
for the BOS, such that a multitude of solar PV system designs are created [24]. Therefore, this section of
the thesis introduces a tool that takes a ubiquitous approach towards analyzing rooftop potential that can
lead to ideal configuration designs and cost-effective building targets.
In reference to Figure 3.1, a procedure is characterized for relating global coordinates to the
blueprint via image processing and pseudo-3D mapping. Then, a scheme for simulating the sun
precession around the building is outlined and utilized to eliminate areas with a substantial degree of
shading. A strategy for parametrically determining the layout of the modules and sizing the BOS, as well
as simulation of plausible energy generation is developed. This design methodology is demonstrated on
two case studies: the first case presents a comparison of an existing PV system to the planning tool, the
second case study portrays the ability of the planning tool designing a PV system on a geometrically
complex rooftop.

27

Figure 3.1. Flow chart illustrating the methodology employed for a solar PV system design that can be
tailored to a building rooftop

28

The procedure for creating the virtual replica of the building begins with an image file format of
the rooftop blueprint. Then binary image processing is employed to extract the various roof regions. This
is followed by specifying the geographical location and orientation of the building, as well as details
pertaining to the individual roof regions. Some of these details aid in the creation of the depth map, while
other specifics develop the components needed for pseudo-3D mapping (see details below), a common
technique in computer graphics that creates a virtual scaled model of the building. Pseudo-3D mapping
was chosen because it is able to associate a scale with Cartesian coordinates. Next, the movement of the
sun is simulated and the areas of the roof that experience a large amount of shading are eliminated. This is
because the shaded modules act as load by consuming the power from other modules. This load
consumption causes harmful excessive heat, which results in failure before the expected lifetime [14].
Following, an iterative procedure is performed for 2 output targets: 1) maximizing the amount of modules
on a roof region for two module orientations and a multitude of tilt angles, and 2) creating optimized BOS
designs based on the electrical specifications of the building. Finally, calculations are performed for
simulating the expected energy generation for multiple PV system designs.

3.1 Image Processing of the Blueprint
The design of a PV system is streamlined by incorporating image processing of the blueprint.
Previously this has not been done, but it has the following advantages: 1) evaluating an unlimited number
PV system components; 2) assessing a multitude of module layouts and orientations; 3) providing a high
degree of dimensional precision for layout of the modules; and 4) substantially reducing analysis time.
Many of the blueprints provided were PDF files corresponding to 11x17’’ sheets. In order to
perform the image processing using MATLAB, the blueprints were rasterized using ImageMagick, from
PDF to PNG file format at 200 pixels per inch [41]. ImageMagick was used because it is a free software
suite for displaying, converting, and editing files with Windows Software Development Kit [42]. The
PNG file format was chosen because it supports lossless data compression [43].

29

3.1.1 Extracting the Roof Regions
A typical blueprint image is described as a 2D array of n rows by m columns of three-byte pixels
that correspond to the primary colors red (R), green (G), and blue (B). However, to extract the roof
regions from the blueprint, the blueprint must be an n x m binary file. This conversion is executed by
initially removing any noise on the image using a Wiener filter, then, the image is converted from color to
grayscale and finally into an n x m binary image of the blueprint [43].
The object regions of the binary image are extracted by implementing a connected component
labeling operation. This process is described in [43, 44] and results in a connected component matrix that
is the equivalent size of the binary image. Figure 3.2 depicts an exemplar connected components matrix
containing sixty-three unique shades that represent the sixty-three unique object regions. Following,
information about each object region is obtained such as the binary area and global pixel locations.

Figure 3.2. Image corresponding to connected components matrix, where each shade represents a
different object region. There are sixty-three unique object regions depicted
After the image has been enhanced and subdivided into informational object regions, details are
specified for each object region and the building as a whole. The entered data includes:

30

1. The size and scale of the blueprint
2. The longitude, latitude, and azimuth of the building
3. A label for each object region, including viable locations for the PV system (some regions
may be deemed unviable due to buildings code restrictions)
4. The region pitch

3.1.2 Depth Map
The depth map of the building DM, referred to as the z-coordinate, is the same n x m dimensions
as the blueprint. It is created by relating the pixel-by-pixel location of connected components matrix, to
the corresponding height of the object region. If an object region has a pitch, the program relates the four
selected corners by a series of vectors, such that four planes are constructed. Then the pixel-by-pixel
location values of each plane are heuristically averaged together, eliminating outlier values, and related to
the depth map. Outlier values may come into existence when the selected corner is not at the outer corner
of the sloped region.

3.1.3 Roof Shade Modeling
The ensuing steps require rotating the binary, connected components, and depth map matrices, as
well as the blueprint image files, such that the building is orientated with respect to its geographical
coordinates. This not only aids in quantifying the location of the sun throughout the day, but also provides
the setup for the layout of the modules.
The seasonal angles required to describe the location of the sun with respect to the design of the
PV system are depicted in Figure 3.3, where Σ is the module tilt angle, δ is the declination angle, θa and θz
are the solar altitude and zenith angles, respectively, and ϕc and ϕs are the sun and module collector face
azimuth angles, respectively. The amount of shade that each region of the building experiences is
calculated using the Winter Solstice Altitude Angle, αW. At this sun angle, the effects due to partial
shading on the modules are reduced. If a module experiences partial shading, an electrical resistance

31

within the module occurs, which result in harmful excess heat, such that failure is encountered before the
expected lifetime of the module [14, 45].

Figure 3.3. Various sun angles used to calculate the shade
The amount of shading that occurs on each object region can be determined in several ways, for
instance, a ray-tracing algorithm. However, implementing a ray-tracing algorithm is computationally
expensive, and surrounding objects such as neighboring buildings are not distinguished on the blueprints
[46]. Instead, the amount of shade that the building rooftop experiences is determined using simple
geometric concepts that are associated with the pixel-by-pixel values within the depth map. Subsequently,
a matrix the same size as the depth map is generated, and locations that experience shade are flagged. The
basic algorithm used is
, where 1 ≤ i ≤ n and 2 ≤ j ≤ m

32

Then, the shade matrix is superimposed with the connected components matrix, and areas containing
shade are eliminated since the cost of maintenance and performance reduction in these regions outweigh
the benefit of additional collection area.

3.1.4 Pseudo-3D Mapping
Although the DM of the building may be viewed as 3D model, it does not have a dimensional
correlation to the physical environment. Thus, due to the module layout employing a systematic
dimensional analysis scheme, discussed in Section 3.2, a scale must be associated with the x, y, and DM
pixel coordinates. This may be achieved a number of ways, for instance, creating a point cloud and
meshing each point with a series of triangles that may be thought of as planes. However, given the postprocessing design of the module layout, incorporating pseudo-3D mapping, sometimes referred to as 2.5D
point mapping is a better fit. The 2.5D point mapping strategy is commonly used in the development of
human face models and video games, where each mapped point specifies a 2.5D projection of a surface,
in this case a building, in 3D space [47, 48].
The 2.5D point mapping technique requires three matrices that correspond to the x, y, and z
coordinates, a flag matrix, and the rotated color blueprint image. The flag matrix is required to account for
missing information, and as stated previously, the z coordinate is the DM. The x and y matrices,
commonly referred to as 2.5D column and row matrices, are n x m arrays the same size as the DM. Each
column or row value in the 2.5D matrices grow in a successive fashion to a specified value, in this case,
the actual size of building [47]. The technique can be visualized as
2.5D Projection Point(i, j) = {Row(i, j), Column(i, j), DM(i, j), r(i, j), g(i, j), b(i, j), Flag(i, j)},
where 1 ≤ i ≤ n and 1 ≤ j ≤ nm.
Figure 3.4 depicts the original RGB blueprint, DM, 2.5D row, and column matrices. The lighter
shades of gray in 3.4c and 3.4d correspond to larger values of the row and column matrices, respectively.
When the three matrices are superimposed with the RGB blueprint image, it culminates into a 2.5D
projection that corresponds to the actual building.

33

a) Original RGB blueprint image

b) Depth map

c) 2.5D row matrix
2.5D column matrix
Figure 3.4. Images representing the matrices incorporated in 2.5D mapping. The lighter shades in c and d
correspond to larger pixel-by-pixel values
Following this procedure, the object regions are separated into a set of binary n x m matrices for
the module layout. For this thesis, the matrices referring to the binary object regions are also called the
viable panel regions.

3.2 Layout of the Modules
The goal of the module layout procedure is to identify a layout that will maximize the number of
modules in each region. While this is a well-studied concept there has been considerably less research
conducted towards characterizing an algorithm that will maximize rectangular shaped objects in an
irregular area [49, 50, 51]. Nonetheless there have been numerous studies aimed at identifying the most
advantageous PV configuration pertaining to the electrical distribution, module tilt angle, sizing ratios of
the PV modules to the inverter, and sizing for load applications [52, 53, 54, 55, 56]. However, these
studies are not as applicable because these configurations require a uniform number of modules connected
34

in series, which may not be the case when rooftop is geometrically complex. Considering that the
modules are maintained at a fixed tilt angle, and have the same characteristic properties such as
dimensions and nameplate capacity, i.e. the PV system only contains identical modules, allows for the
assumption that the total spacing between each string TS is of fixed length. Additionally, because the goal
is to uniformly fit as many modules into a region as possible, results in an algorithm that can perform
numerical approximation of definite integrals, that can be viewed as Left Riemann Sums. Referring to
Figure 3.5, the bounding curves of integration may be viewed as the outer perimeter of each viable panel
region.

Figure 3.5. Representation of a viable panel region
Figure 3.6 describes the vectors used in the module layout, where MP is the module projection on
to the ground and it is a function of module tilt angle and dimension L2. The shade projection SP on to the
ground due to the sun is calculated from the maximum module height above ground H and the high noon
solar altitude angle α. These are
(3)
(4)
(5)

35

(6)
Thus, the total spacing TS between each string is the summation of the module projection and the shade
projection.

Figure 3.6. Description of the vectors employed in the layout of the modules
In reference to Figure 3.7, a string can be split into subintervals of modules connected in series.
For example, each string in a PV system requires twelve modules; however, the layout of the modules has
two subintervals S, that each contain six modules. These modules are connected in series via either
module-to-module wires, or a pass through wire box.

Figure 3.7. Illustration of how two string subintervals may be connected in series

36

The amount of string subintervals that can fit in the viable panel region is determined by
(7)
where A and B are the minimum and maximum column values that correspond to the location of the
viable panel region with respect to the 2.5D column matrix. The amount of subintervals is correlated to
the column starting position on the binary matrix. Due to the binary matrix containing a length per pixel,
there is a possibility that the shade from a subinterval could end in the same column as a string may start.
In that case, a flag is stored that indicates that the column starting position of the binary matrix will be
reexamined.
The algorithm for the layout of the modules references three main matrices: 1) binary, 2) 2.5D
row, and 3) 2.5D column. The binary matrix determines where a module can be placed, and the 2.5D row
and column matrices give the dimensions. The layout begins at the binary minimal row and column
position for each viable panel region. Following, each binary row is summed along the column, the binary
distance corresponding to the MP. If the sum of each binary row equals the binary projection, the 2.5D
row and column values that coincide to the start and end position of that row are stored. When the last
binary row of the subinterval is completed, the total distance that corresponds to the stored 2.5D row
values is determined. Then this value is divided by the module length not used in the projection (in the
case of Figure 3.6, this is the L1 length), thus resulting in the amount modules that can fit in a subinterval.
The aforementioned algorithm repeats for each subinterval start position. In addition, the last starting
position of a subinterval plus the MP may not equal the last column of the viable panel region. Therefore,
an iterative procedure where the binary start position for a subinterval is shifted until the last subinterval
starting position plus the MP is equal to the last column of the viable panel region. Each iteration is then
compared, and the iteration that supports the most modules is stored. Next, the orientation of the module
is flipped, i.e. L1 and L2, and the code repeats. Lastly, each module orientation is compared, and the
orientation that can support the most modules is stored for further processing. This algorithm was
developed for a collection modules with a range tilt angles, and four solar altitude angles that correspond
to the Winter and Summer Solstice and the Vernal and Autumnal Equinox at high noon.
37

3.3 Initial Energy Comparison & Inverter Characteristics
The maximum amount of modules that can fit in a region for the four altitude angles, multiple
modules, and tilt angles ranging from 0-90˚ has been determined. Now the planning tool compares each
module layout on the basis of power and energy. A derate factor DF may be applied without having to
design the BOS. These DFs, presented in Table 3.1, account for the losses when converting from DC to
AC at standard testing conditions (STC) [57].
Table 3.1. PV system component derate factors for DC to AC conversion at STC
Range
Component Derate Factors
Typical
Low
High
PV Module Nameplate DC Rating
0.95
0.8
1.05
Inverter and Transformer
0.92
0.88
0.98
Mismatch
0.98
0.97
0.995
Diodes and Connections
0.995
0.99
0.997
DC Wiring
0.98
0.97
0.99
AC Wiring
0.99
0.98
0.993
Soiling
0.95
0.30
0.995
System Availability
0.98
0.00
0.995
Overall Derate Factor
0.769
0.00
0.96
The expected AC power PAC is a function of the system’s DC rating and the inverter derate factor
DFinv. The DC rating is found by multiplying the maximum power of each module Pmp with the total
amount of modules M in the system. The PAC is determined by
(8)
Following, the base energy capability GS t of each layout is calculated. This is a function of the total
quantity of modules in the system, the solar insolation entering the module’s collector face Ic t at a given
time period, the area of the module APV, the efficiency of the module ηPV, and a system derate factor DFsys.
(9)
The module efficiency is determined by the current and voltage at the maximum power point on the I-V
curve, Imp and Vmp, respectively, as well as the area of the module and the solar insolation at STC I0 [8].
This is expressed as

38

(10)
The layout that supports the highest energy generation per module and highest total energy
generation is chosen for matching a preliminary inverter, where this inverter must have the same output
phase ϕ and voltage V of the building [8].

3.4 Electrical Connections
Using Kirchhoff’s Voltage Law in conjunction with the maximum operating voltage of the
inverter Vinv mpp and the module’s open circuit voltage at STC Voc 0, the number of modules in series MS is
found by
(11)
As a result, the number of strings connected in parallel MP is found by
(12)
The maximum operating voltage for the inverter is typically 70% of the maximum rated input DC voltage
[58]. It should be noted that the floor function means to round down to the nearest whole number. In
addition, the amount of strings is reduced if the short circuit current of the system is greater than the input
current of the inverter.
After the number of modules in series and parallel has been determined, the layout of modules
from the preliminary design is modified to match the electrical connections. The layout modifications are
as follows: 1) if a region cannot support a string, it is eliminated; or 2) if a region of x number of strings
has y modules remaining, the y remaining modules are eliminated as well. Due to the potential that
modules from the regions are reduced, the inverter is reexamined for the possibility that a smaller inverter
is available. If a smaller inverter is available, Eqs. (8) - (10), and (12) are repeated.

3.5 Sizing of the Balance of Systems
The BOS formulation being described is in reference to Figure 2.14 Figure 3.1. The wires from
the strings transition via a pass through wire box into regular building wire typically referred to as

39

conduit. The input fuse rating for the box PTBFR is a function of the short circuit current from a module
Isc 0, the amount of strings that are entering the pass through box PTBN, and a safety factor of 125% to be
in accordance with NEC [8, 18, 24, 59], i.e.
(13)
In the case of smaller systems, i.e. systems smaller than 1kW, the pass through wire box can be
used as the combiner box. Consequently, in the case of larger systems, the conduit from the pass through
wire box is fed into a combiner box [8, 18, 24]. The NEC handbook contains tables that determine the
size of the conduit wire based on the amperage and type of conduit, where the type depends on its
location within the system.
The input fuse rating of the combiner box CFR is a function of the short circuit current of the
module, the amount of strings entering/exiting the pass through wire box, a factor of 125% to account for
higher insolation at lower temperatures and another factor of 125% for safe over sizing in accordance
with NEC. The maximum input voltage rating of the combiner box CVR is determined from the total open
circuit voltage of the PV arrays, and a temperature correction factor from Table 690.7 of the NEC
handbook [8, 18, 24, 59], i.e.
(14)
(15)
The maximum output voltage of the combiner box is equal to the maximum input voltage. The maximum
busbar rating, established during UL 1741 testing, defines the maximum continuous output current of the
combiner box. The combiner box’s maximum output current CIm is determined by the number of inputs
CN and the open circuit current [8, 18, 24, 59], as
(16)
Due to the intricacy of the NEC codes, along with the dynamic complexity of the solar PV
market, calculations for ground fault protection devices are not included. Furthermore, many inverters on
today’s market include ground fault protection devices and in some cases, DC array disconnects.
However, in the case that the DC array disconnect is not included with the inverter, the disconnect input
40

fuse rating SIDC must be sized in accordance with NEC 690. This is determined by multiplying Eq. (14)
with the amount of strings connected in parallel, and dividing by the number of combiner box outputs Co
[8, 18, 24, 59].
(17)
To determine the size of the disconnect under normal operating conditions, a temperature correction
factor F is applied as

(18)
where Tmax, normal is the maximum allowed temperature during normal operating conditions, and ΔTmax is
maximum allowed temperature rise specified by the manufacturer. These values are typically 40˚C and
85˚C respectively. Therefore the disconnect rating under normal conditions is then
.

(19)

Additionally, the maximum input voltage for the disconnect may be sized using Eq. (15) [8, 18, 24, 59].
The inverter is initially selected using Eq. (8). Then, based on the manufacturer’s specifications,
the voltage of the PV system must be in agreement with the inverter [8, 18, 24, 59], i.e.
(20)
(21)
These calculations assume that the number of DC outputs from the combiner box, do not exceed the
number of inputs for the inverter. If so, the PV system will need more combiner boxes. The AC
disconnect, according to the NEC, must be sized based on the outputs of the inverter. Equations (13) to
(21) are for PV systems 600V DC or less; larger systems require special attention to NEC 690 [8, 18, 24,
59].

3.6 Simulation of the Energy Generation Capabilities
Two models are described in this section for determining the expected energy generation of a PV
system. In the models, a module degradation factor MDF is included to account for the module

41

deteriorating over time, which is typically 0.5% each year [13]. The common calculation used in both
models for determining the expected energy generation PV system in a given time period GS t is
(22)
The simple model for calculating the expected energy generation of the PV system can be
determined by applying Eqs. (10) and (22), along with solar resource data and information supplied on the
module specification sheet [8].
The more complex model for determining the expected energy generation, and the one utilized
here, is the SNL Photovoltaic Array Performance Model [60]. This model includes the electrical, thermal,
and optical properties and is designed to be used in conjunction with hourly solar resource and
meteorological data. An interesting attribute of this model is that it is applicable for an array, string, single
module, or single solar cell. The temperature on the back-surface of a module Tm and the cell Tcell are
given by
(23)
(24)
The temperature on the back-surface of a module Tm, is a function of the insolation that enters the module
collector face, the wind speed WS at 10 meters above ground level, and the ambient temperature Tamb.
Additionally, two temperature coefficients a and b, couple the affects of wind speed and insolation on the
temperature of the module. The solar cell temperature Tcell is a function of the module’s back-surface
temperature, the temperature difference between the solar cell and module back surface ΔT, as well as a
ratio of the insolation entering the module collector face and the insolation at STC. The current and
voltage derived from the SNL model are as follows
(25)

(26)
(27)

42

(28)
The terms C0 and C1 relate the current at the maximum power point to the effective insolation, whereas C2
and C3 relate the voltage at the maximum power point to the effective insolation. The number of modules
and solar cells electrically connected in series is Ms and Ns respectively, and δ(TCell) is the thermal voltage
per cell at the cell temperature. Additionally, n is the diode factor, and

and

are

temperature coefficients that relate the module’s open circuit and maximum power point voltage as a
function of the effective insolation.

3.7 Case Studies
The solar PV design tool is tested and verified on two case studies. The first case study
corroborates the solar PV design tool by comparing it to an existing rooftop solar PV system at the
Michigan State University Surplus Store. This case study is divided into two parts. The first part provides
a comparison of the module layout and energy generation, based on the existing 44˚ module tilt angle. In
the second part of this case study, the solar PV design tool will exhibit two designs for the layout of the
module: 1) the highest energy generation per module on a yearly basis and 2) the highest energy
generation per region on a yearly basis. The energy generation for each scenario is a function of the solar
insolation, module tilt angle and the module efficiency. These scenarios are different because typically
when the modules have a lower tilt angle, more modules can be placed into an object region as opposed to
a higher tilt angle. Thus, the object region that contains more modules may produce more energy, but at a
potentially higher cost/area. The second case study demonstrates the solar PV design tool’s ability to
develop a PV system when the viable panel regions are irregularly shaped. The module selected for the
design is presented in Table 3.2 and has the same characteristics as the modules on the MSU Surplus
Store.

43

Table 3.2. Installed module characteristics under standard testing conditions
Pmp
210 W
Maximum Power
Voc
33.6 V
Open Circuit Voltage
Vmp
26.4 V
Maximum Power Point Voltage
Isc
8.33 A
Short Circuit Current
Imp
7.95 A
Maximum Power Point Current
L
1,482
mm
Length
1
L2
992 mm
Length

3.7.1 Case Ia: Validation of the Solar PV System Design Tool
The Surplus Store on the campus of MSU is located at 42.43˚N, 84.50˚W, and has 90˚ azimuth.
The yearly average energy demand is 470 MWh, an average base power load of 170 kW and peak load of
250 kW. The roof of this building is shown in Figure 3.8a, where the green highlighted area represents a
902m2 area depicting the location of the PV system. The existing 40.3 kW PV system is comprised of 16
strings with 12 modules per string, which culminates into 192 modules tilted at 44˚. There are also 16pass through wire boxes and one 16-input combiner box. Additionally, the system contains a 50 kW 3ϕ
480/277V inverter, a DC disconnect rated at 600V and 200A, and an AC disconnect rated at 480VAC and
72A. Figure 3.8b depicts the monthly average energy consumed by the building, as well as the amount of
energy the PV system generates. According to this figure, the PV system generates roughly 10% of the
building load. In this case study, the layout of modules from the PV design tool allowed for a four-foot
perimeter around the edge of the roof.

Figure 3.8. Michigan State University Surplus Store characteristics. The current PV System is located on
green highlighted roof section in 3.8a. The average monthly energy demand is in section 3.8b.
44

When comparing the PV system at the Surplus Store to the PV design tool, the solar PV design
tool recommends that 214 modules tilted at 44˚ (orientated such that L1 creates the projection on the
ground) can be fit in the space. Based on the number of modules and the building requirements, the tool
determined that a preliminary inverter should be 3ϕ 480/277V and 44 kW or more. Then using the
iteration described in Figure 3.1, the layout was reduced to 204 modules, comprised of 17 strings with 12
modules per string. Accordingly, the inverter for the system still had to be 3ϕ 480/277V but now has a
minimal power rating of roughly 41 kW. Consequently, the program determined that 17 pass-through
boxes were needed, and a combiner box that has 17 inputs with a DC disconnect requiring a minimum
amp rating of 157A and 600V. The PV design tool established that the AC disconnect should be rated at
72A, which was based on the inverter’s maximum over current protection per phase while at 480V. The
results of this comparison are presented in Table 3.3.
Table 3.3. Design comparison of an existing solar PV system at the MSU Surplus Store vs. the solar PV
design tool
PV System
Existing Solar PV System located on MSU Potential Solar PV System located on MSU
Attributes
Surplus Store
Surplus Store
12 Modules
12 Modules
Module Layout 192 Modules 16 Strings
204 Modules 17 Strings
Per String
Per String
Amount of Pass
Through Wire
16
17
Boxes
Combiner Box
1 Combiner Box
16 Inputs
1 Combiner Box
17 Inputs
Characteristics
Inverter
50 kW
3ϕ
480/277 V
41 kW
3ϕ
480/277 V
Characteristics
DC Disconnect
600 V
200 A
600 V
157 A
Characteristics
AC Disconnect
480 V
72 A
480 V
72 A
Characteristics
The results of the solar tool match relatively well, as far as the module layout and the sizing
multiple components. Nonetheless, the calculation of the expected energy generation should also be
accurate. The initial calculation determined that over the lifetime of the system roughly 225 kWh a year
per module could be produced. Therefore, the energy derivation provided by SNL showed that over the
lifetime of the system, approximately 216 kWh/year-module could be produced. The current system at the

45

Surplus Store produces about 219 kWh/year- module. Figure 3.9. illustrates a monthly comparison of the
existing PV system versus the PV planning tool.

Figure 3.9. Monthly breakdown of the existing PV system compared to the solar PV planning tool

3.7.2 Case Ib: Energy Capabilities of the Solar PV System Design Tool
Continuing with the case study of the MSU Surplus Store, the solar PV design tool will exhibit
two designs for: 1) the highest energy generation per module on a yearly basis and 2) the design for the
highest energy generation per region on a yearly basis.
The situation resulting in the highest energy generation per module yielded that 243 modules
tilted at 33˚ orientated such that L1 created the projection on to the ground could fit in the space. As a
result, based on the number of modules and the building requirements, the PV design tool determined that
a preliminary inverter should be 3ϕ 480/277V and 49 kW or more. Following the iteration described in
Figure 3.1, the layout was reduced to 240 modules, comprised of 20 strings with 12 modules per string.
Accordingly, the inverter for the system still had to be 3ϕ 480/277V, but now a minimal power rating of
roughly 48 kW is required. Then, the program computed that 20 pass through boxes were needed and a
combiner box that was either 20 inputs or two combiner boxes that were each 10 inputs. Subsequently,
the DC disconnect required a minimum rating of either 185 or 93A respectively and 600V. A 72A AC

46

disconnect was established by the PV design tool, which was based on the inverter’s maximum over
current protection per phase while at 480V.
The situation that yielded the highest energy generation per region determined that 603 modules,
tilted at 0˚, orientated such that L1 creates the projection on to the ground could fit in the space.
Accordingly, the inverter for the system still had to be 3ϕ 480/277V but has a minimal power rating of
approximately 121 kW. Using the same iteration scheme, the layout was reduced to 600 modules
comprised of 50 strings with 12 modules per string. As a result, the inverter for the system was 3ϕ
480/277V with a minimal power rating of 120 kW. The program determined that 50 pass through boxes
were needed and a combiner box that was either 50 inputs or two combiner boxes that were each 25
inputs. Subsequently, the DC disconnect required a minimum rating of either 462 or 231A respectively
and 600V. A 72A AC disconnect was established by the PV design tool, which was based on the
inverter’s maximum over current protection per phase while at 480V. The results for both cases are
summarized in Table 3.4.
Table 3.4. Comparison at the MSU Surplus Store of the solar PV system design for highest energy
generation per module and highest energy generation per region
PV System
Highest Energy Generation per Module
Highest Energy Generation per Region
Attributes
12 Modules
12 Modules
Module Layout 240 Modules 20 Strings
600 Modules 50 Strings
Per String
Per String
Tilt Angle
33˚
0˚
Amount of Pass
Through Wire
20
50
Boxes
1 Combiner Box
20 Inputs
1 Combiner Box
50 Inputs
Combiner Box
Characteristics
2 Combiner Boxes
10 Inputs
2 Combiner Boxes
25 Inputs
Inverter
48 kW
3ϕ
480/277 V
120 kW
3ϕ
480/277 V
Characteristics
600 V
185 A
600 V
462 A
DC Disconnect
Characteristics
600 V
93 A
600 V
231 A
AC Disconnect
480 V
72 A
480 V
72 A
Characteristics

3.7.3 Case II: Complex Roof Geometries of the Solar PV System Design Tool
The second case study portrays the ability of the solar PV design tool to account for a
geometrically complex rooftop and perform an economic evaluation. An exemplar rooftop on the campus
47

of MSU is the Psychology Building. This building is located on the North-Side of campus, at 42.73˚N,
84.55˚W, and has a 174˚ azimuth. The yearly average energy demand, shown in Figure 3.10b, is 1,300
MWh. The normal base load is 150 kW with a Summer peak load of 300 kW. Furthermore, the
panelboard for this building is 3ϕ 480/277 V. Figure 3.10a depicts the roof of the Psychology Building,
where the numbers 1-7 indicate the regions deemed viable for the installation of a solar PV system. Table
3.5 portrays the area of each region before and after the effects of shade.

Figure 3.10. Characteristics of the Michigan State University Psychology Building. The labels 1-7 in
3.10a correspond to the regions employed in the layout of the modules. The average monthly energy
consumption from 2007-2013 is shown in 3.10b.
Table 3.5. Total area of each region and area of each region after shade is included

Region
1
2
3
4
5
6
7

Area (m2)

Viable Area(m2)

26.7
339.9
365.9
103.1
489.7
524.3
341.3

16.1
291.0
320.4
93.4
489.7
431.1
239.3

The solar PV design tool will exhibit two designs for the layout of the module: 1) the highest
energy generation per module on a yearly basis and 2) the highest energy generation per region on a
yearly basis. The situation resulting in the highest energy generation per module yielded that 428 modules
tilted at 33˚ orientated such that L1 and L2 created the projection on to the ground could fit in the space.
48

As a result, based on the amount modules and the building requirements, the PV design tool determined
that a preliminary inverter should be 3ϕ 480/277V and 86 kW or more. Again, following the iteration
described in Figure 3.1, the layout was reduced to 384 modules, comprised of 32 strings with 12 modules
per string.
Accordingly, the inverter for the system still had to be 3ϕ 480/277V but now a minimal power
rating of roughly 77 kW is required. The program determined that 32 pass through boxes were needed and
a combiner box that was either 32 inputs or two combiner boxes that were each 16 inputs. Subsequently,
the DC disconnect required a minimum rating of either 296 or 148A respectively and 600V. A 72A AC
disconnect was established by the PV design tool, which was based on the inverter’s maximum over
current protection per phase while at 480V.
The strategy that yielded the highest energy generation per region determined that 1069 modules,
tilted at 0˚, orientated such that L1 and L2 created the projection on to the ground, could fit in the space.
The PV design tool determined that a preliminary inverter should be 3ϕ 480/277V and 214 kW or more.
Again, following the iteration described in Figure 3.1, the layout was reduced to 1020 modules comprised
of 85 strings with 12 modules per string.
Correspondingly, the inverter for the system still had to be 3ϕ 480/277V but now has a minimal
power rating of approximately 204 kW. The program determined that 85 pass through boxes were needed.
Given the combiner box product data base used in the program, a combiner box that could accept 85
inputs did not exist. In order to avoid current mismatch, the module layout was again reduced such that it
now contains 1008 modules comprised of 84 strings with 12 modules per string. Thus, the inverter power
rating was reduced to 202 kW, and the PV design tool determined that 84 pass through wire boxes were
needed. As a result, the program determined that two combiner boxes of each 42 inputs, or three combiner
boxes of 28 inputs. Assuming each combiner box had one output, the DC disconnect required a minimum
rating of either 388 or 259A respectively and 600V. A 72A AC disconnect was established by the PV
design tool, which was based on the inverter’s maximum over current protection per phase while at 480V.

49

Table 3.6 displays the results of the module layout for both scenarios, while Table 3.7 presents the system
characteristics for each design.
Table 3.6. The layout of the modules on the roof the MSU Psychology Building, for the highest energy
generation per module on a yearly basis, and highest energy generation per region on a yearly basis
Highest Energy Generation Per Region
Highest Energy Generation Per Module
Region
Module
First Pass
Second Pass
Module
First Pass
Second Pass
Orientation
Modules
Modules
Orientation
Modules
Modules
1
L1 or L2
3
0
L2
7
0
2
L1
68
60
L1
170
168
3
L2
70
60
L1
172
168
4
L1 or L2
19
12
L1
46
24
5
L1
111
108
L1
287
276
6
L1
102
96
L1
250
240
7
L1
55
48
L1
137
132
Table 3.7. Comparison at the MSU Psychology Building of the solar PV system design for highest energy
generation per module and highest energy generation per region
PV System
Highest Energy Generation per Module
Highest Energy Generation per Region
Attributes
12 Modules
12 Modules
Module Layout 384 Modules 32 Strings
1008 Modules 84 Strings
Per String
Per String
Tilt Angle
33˚
0˚
Amount of Pass
Through Wire
32
84
Boxes
1 Combiner Box
32 Inputs
2 Combiner Box
42 Inputs
Combiner Box
Characteristics
2 Combiner Boxes
16 Inputs
3 Combiner Boxes
28 Inputs
Inverter
77 kW
3ϕ
480/277 V
202 kW
3ϕ
480/277 V
Characteristics
600 V
296 A
600 V
388 A
DC Disconnect
Characteristics
600 V
148 A
600 V
259 A
AC Disconnect
480 V
72 A
480 V
72 A
Characteristics
The economic evaluation for the two situations is a culmination of the NREL 2010 bottom-up
installed price benchmarks, discussed in Section 2.7, and the 4th quarter 2013 US Solar Market Insight
cost variance, which is $1.70/W to $8.00/W with an average of $3.57/W for non-residential installations
[33]. In this analysis, the energy generated G from the PV system is considered the energy cost savings,
where the commodity rate of electricity CRE is calculated as a flat rate, but is adjusted using the national

50

escalation rate of electricity EE, and is discounted with the prime lending rate PLR. The present value
cost savings on a monthly scale is determined by

(29)

where each rate R (EE, MDF, PLR) was converted from annual effective to monthly effective by
(30)
These results are dictated in Table 3.8 and illustrated in Figure 3.11.
Table 3.8. Economic evaluation of the solar PV system installations for the highest energy generation per
module and highest energy generation per region
Financial
Highest Energy Generation Per Module
Highest Energy Generation Per Region
Aspects
$1.70/W
$3.57/W
$8.00/W
$1.70/W
$3.57/W
$8.00/W
Initial Capital
$198,250
$309,800
$574,300
$496,000
$813,400
$1,507,600
Cost
Levelized
Cost of
Energy
Return on
0.049%
-0.186%
-0.4730%
0.029%
-0.25%
-0.53%
Investment

51

Figure 3.11. Economic evaluation of the solar PV system installations for a.) highest energy generation
per module and b.) highest energy generation per region
Using these benchmark prices, an amortization schedule was created for 5, 10, and 15 year loans, where
the initial downpayment is roughly 20% of the total installed cost. The monthly annuity A is determined
by

52

(31)
where the interest rate R is the national prime lending rate. The payment summary is dictated in Table 3.9.
Table 3.9.Amortization payment summary for the solar PV system installations that correspond to the
highest energy generation per module and highest energy generation per region
Highest Energy Generation Per Module
$1.70/W
$3.57/W
$8.00/W

Highest Energy Generation Per Region
$1.70/W
$3.57/W
$8.00/W

$48,250

$64,800

$124,300

$101,000

$163,400

$307,600

$150,000

$245,000

$450,000

$395,000

$650,000

$1,200,000

5 year

$2,708.83

$4,424.42

$8,126.48

$7,133.25

$11,738.25

$21,670.62

10 year

$1,462.48

$2,388.72

$4,387.44

$3,851.20

$6337.41

$11,699.83

15 year

$1050.55

$1715.90

$3,151.65

$2,766.44

$4552.38

$8,404.39

5 year

$12,720

$20,776

$38,160

$33,496

$55,120

$101,760

10 year

$25,894

$42,294

$77,683

$68,188

$112,208

$207,154

15 year

$39,720

$64,877

$119,162

$104,597

$172,122

$317,765

5 year

$162,720

$265,776

$488,160

$428,496

$705,120

$1,301,760

10 year

$175,894

$287,294

$527,683

$463,188

$762,208

$1,407,154

15 year

$189,720

$309,877

$569,162

$499,597

$822,122

$1,517,765

Financial Aspects

Total Paid

Total
Interest
Paid

Payment
(Monthly)

Initial
Downpayment
Loan Amount

3.8 Discussion
The results of the simulations are reasonably accurate on an annual basis; however, the
discrepancies may be due to a variety of reasons such as the existing metered data being rounded, system
shutdowns not being documented, and generally inconsistent weather patterns. Nonetheless, the solar tool
may be improved in a number of ways. For instance, the algorithm for the module layout may be revised
to account for pitched roof sections, or arrays at various module orientations. Whereas the simulation may
be revised by including a detailed system thermal model, an inverter performance model, and losses in the
conduit and connections.
In regards to the economic analysis, the 2013 US Solar Market Insight did not include the cost of
labor, grid-interconnect, and permits, however these values were taken from 2010 bottom-up installed
price benchmark and were adjusted based on their respective inflationary rate. The O&M and replacement
costs were taken from a report that was made by Black & Veatch for NREL [61].As far as this tool having
53

the capabilities to reduce the installed costs, Michigan State University was quoted $50,000 for an
evaluation of three rooftops of similar size to the case studies. However, this quote only examined one PV
system design, whereas this tool can examine multiple PV system designs. In addition, the financial
evaluation for the three rooftops would be an additional $30,000. That equates to almost $27,000 per roof.
Therefore, this tool is a very cost-effective solution for any energy planner.

3.9 Summary and Conclusions
This section in the thesis describes a methodology for developing a parametric solar PV planning
tool for grid-tied building applications that takes an adaptive modeling approach towards capitalizing on
the complex geometry of a rooftop. The robustness of the tool was demonstrated and confirmed against
existing PV systems, such that it may be utilized towards making the installation of a PV system more
cost-effective.

54

Chapter 4 FLUORESCENT TO LED LIGHTING RETROFIT
Businesses, institutions, hospitals, etc. are always under constant pressure to reduce operational
costs. Many estimates regarding commercial buildings, state that the lighting system accounts for roughly
one-third of its total energy consumption, while more than half of the total energy consumption in retail
and lodging buildings is attributed to the lighting system. However, it is all well documented that the
lighting system has the largest potential for reducing a building’s energy consumption [62].
A lighting system retrofit is defined as the practice of modifying or replacing an existing lighting
fixtures, or luminaries with equivalent, more energy efficient accessories. This could be as simple as
purchasing screw-in compact fluorescent lamps from a local hardware store to replace incandescent
lamps, or a retrofit could be as in-depth as a detailed energy audit, engineering, feasibility study, and
implementation. There are many benefits for retrofitting the lighting system, such as reduced labor and

maintenance, reduced greenhouse gas emissions and non-monetizable benefits, which include
productivity increase, and lower stress and anxiety. However, the largest benefit for retrofitting
the lighting system is the energy savings over time [63, 64, 65]. Depending on the project size and
complexity, there are roughly six phases in the retrofit process, which are highlighted in Table 4.1.

55

Table 4.1. Overview of the six steps for a successful lighting system retrofit
Phase
Rationale
Endeavors
 Perform an initial walk-through of the
Make a quick assessment of the
Assessment
building to assess the quality of lighting and
building lighting system
potential for improvements
 Survey and document the existing lighting
Collect pertinent information
 Interview building occupants
Data Procurement
needed for engineering
 Audit existing electric bills
 Monitor existing lighting system
 Evaluate current lighting quality
 Evaluate current energy usage
Evaluate the current lighting
 Determine retrofit approach
situation, identify and suggest
o Relamp vs. redesign
Engineering
lighting retrofit opportunities,
o Occupancy sensors
and evaluate cost effectiveness  Estimate retrofit costs based on approach
 Estimate energy savings based on approach
 Economic analysis
 Bid construction company
Procurement
Obtain cost information
 Bid lighting retrofit products
Construction and
 Lamp and ballast disposal
Implementation
Commissioning
 Install retrofit equipment
Assessing the retrofit
 Evaluate energy usage
Verification
implementation to ensure
 Verify economics
assumptions

In this chapter, an overview of each retrofit phase is provided, and is demonstrated with a
case study on three building spaces around the campus of Michigan State University. The
preliminary phase outlines key indicators for deeming if a lighting system retrofit is warranted.
Next is a procedure for obtaining the pertinent energy usage and financial history is described.
Then, an explanation about the main characteristics of light, along with a financial strategy for
developing a series of retrofit options is explained. Subsequently, this leads into a brief overview
about the procurement, construction, and design verification phase. Finally, this is concluded
with a case study.

4.1 Phase I: Preliminary Assessment
The preliminary assessment phase is an evaluation of the current lighting technology in order to
judge if a detailed lighting energy audit, engineering, and financial feasibility study is warranted. This
step may be as simple as a phone call to the building manager, and reviewing the lighting blueprints and
56

utility bill history. On the other hand, it may consist of a building walk-though to survey the primary
lighting technologies and lighting control devices. There are many indications to determine if a building
will have a financially successful retrofit, some of which are listed in Table 4.2 [63, 64, 65].
Table 4.2. Indicators for determining the viability of an economically feasible retrofit
Economically Viable
Economically Unviable
 Contains energy intensive lighting
 Contains lighting that was installed prior to
 Recently undergone a successful retrofit
1980
 Short hours of operation
 Long hours of operation
 Complies or exceeds ASHRAE/IESNA/ANSI
 Eligible for incentives and rebates
90.1, DOE, or LEED standards
 High cost of energy
 Ineligible for incentives and rebates
 Current lighting technology is being
 Low cost of energy
discontinued
 Lights remain on when a room is unoccupied

4.2 Phase II: Data Procurement
If the project is considered financially viable for a lighting system retrofit, the subsequent
procedure entails collecting information about the building. It is ideal to collect the data in an order
mannered, such that the analysis is performed in an efficient fashion with minimal errors. The information
being assembled is 1) the building schedule, which determines when the building is occupied; 2)
interviews with building managers such that they can detail problem areas; 3) detailed inventory of the
lighting equipment and size of each space; and 4) determining the lighting hours [64, 65].

4.2.1 Building Schedule & Utility Billing History
A building schedule is used to denote when the building is in operation.

This is very

advantageous when determining the energy usage that is spent when the building is occupied and energy
that is wasted when the building is unoccupied. The total energy is used for determining if a lighting
technology retrofit will be beneficial, or if occupancy sensors (O.S.) should be investigated. The energy
used during operation is intended for assessing a lighting technology retrofit, and the energy spent when
the building is unoccupied is used for appraising the feasibility of retrofitting a space for O.S [63, 64,

65].

57

The previous utility bills are used for determining when the peak power demand occurs, and
correlate the peak demand with the commodity rate of electricity during peak usage. This is useful when
calibrating the financial models for determining if a lighting system retrofit is financially viable [63, 64,

65].
4.2.2 Building Spaces & Inventory Survey
Well documented information about each space in the building, as well as a detailed inventory of
the lighting system in each space is typically the most time consuming portion of the data procurement
phase, but is essential for a successful retrofit analysis. In reference to Figure 4.1, the intent of this
document is for detailing each building space. The floor level is meant for organization, whereas the
space ID is unique for each space. The ASHRAE/IESNA/ANSI 90.1 room type is utilized for obtaining
the proper lumens per watt based on the function of the room. The space dimensions are for calculating
the appropriate lumens per watt, and the column for occupancy sensors is intended for detailing the
energy usage [63, 64, 65].

Floor Level

Space ID

ASHRAE/
IESNA/ ANSI
90.1 Room Type

Space Dimensions
Ceiling
Height

Length

Width

Currently
Occupancy
Sensors

Figure 4.1. Data input form for each space
In reference to Figure 4.2, this document is intended for detailing each type of fixture for each
space. The type of fixture, housing, ballast, bulbs per fixture, and lumens per watt are intended for
determining the lighting levels, and the most advantageous retrofit kit. As will be stated in Appendix III,
different lighting technologies produce various light levels and energy consumption, which have a large
effect on the engineering phase. Along with detailing each fixture, the conditions of the fixture were
documented, for instance, the cleanliness and age of the lighting technology. In some cases, a light meter
should be used for examining the current lighting quality and intensity [63, 64, 65].

58

Fixture
Number of
Fixtures

Type of
Fixture

Type of
Housing

Bi-Level or
Bulbs Per Lumens Per Watts per
Single-Level
Fixture
Fixture
Bulb
Lighting

Type of
Ballast

Figure 4.2. Data input form for each fixture for each room

4.2.3 Energy Audit
The economic viability is main driver for determining the success of a lighting system retrofit,
where this success depends on the accuracy of energy audit. Due to not using submetering equipment for
obtaining the energy usage, a procedure was developed, depicted in Figure 4.3, for obtaining the daily
average energy usage. The procedure begins by using the raw data from the walk-throughs, and
organizing it according to each perspective building space. Following, each building space was separated
by the height of the ceiling. This was because at MSU has a different installation cost metric for fixtures
above and below 15 feet had different installation costs. Then, each space was separated by whether or
not an O.S. was present. Next, each type of fixture and the lighting level (bi-level, single-level) was
separated. At this point, the procedure is split into sections. The first section determines the power
consumption of each type of fixture based on the bulbs, lighting level, and driver/ballast power factor.
The second section determines the average daily hours the lights are on in a space. This is established by
separating the walk-though times by morning, afternoon, and night, where the times of day are dictated in
Table 4.3.
Table 4.3. Walk-through times used to quantify energy usage
Time Period
Time of Day
Morning
8:00am-5:00pm
Afternoon
5:00pm-9:00pm
Night
9:00pm-8:00am
This is further separated by whether or not a room was occupied when the lights were on, which is used to
determine the economics behind a feasible lighting control retrofit. From this organization process, the
daily average energy consumption was determined.

59

Figure 4.3. Flowchart for determining the average daily energy usage of each fixture

4.3 Phase III: Engineering
The engineering phase of the lighting retrofit involves evaluating the current lighting quality, and
quantifying the current energy usage from the walk-throughs and inventory. From there, determining a
series of options for approaching the retrofit, such as relamping with a retrofit kit or a complete redesign
of the lighting system, as well as assessing the needs for occupancy sensors. Following, the economics of
each retrofit option are analyzed, which is a culmination of the cost of the retrofit and the plausible energy
savings.

60

4.3.1 Characteristics of Light
Light is defined as the portion of the electromagnetic spectrum that is visible to the human eye,
where the radiant power emitted at each wavelength over this portion is the spectrum or the spectral
power distribution. The spectra of a light source provides a complete description of a lamp’s
characteristics [63].
A light source’s chromaticity is its quality of color, independent of illuminance, and is derived
from the light source’s spectra. The chromaticity chart, depicted in Figure 4.4, was created in 1931 by the
Commission Internationale de l’Eclairage (CIE), where the x and y coordinates range from 0.0 to about
0.8, while the wavelengths corresponding to visible light are around the curved edge. For any given
spectral power distribution, only one set of chromaticity coordinates exists. However, for any set of
chromaticity coordinates, there is a variety of spectral power distributions. Often times, the chromaticity
coordinates pertaining to the diagram are used to compare two or more colors for their respective
appearance [63, 66].

Figure 4.4. 1931 CIE chromaticity chart [63]
The correlated color temperature (CCT) is the apparent color of a light source in relation to a
theoretical black body radiator. It is measured in Kelvins (K) and is derived from the spectra. Light
sources with temperatures ranging from 3000 K or less, and have an orangish-yellow appearance are
61

considered to be “warm”. Whereas light sources with temperatures 4100 K or greater, and have a bluishwhite appearance, are considered to be “cool” [63, 66]. The CCT chart is shown in Figure 4.5.

Figure 4.5. Color temperature chart courtesy of Sylvania Lighting
The CCT in a room has a dramatic affect on the occupants. For instance, a photographer usually prefers a
very cool light that ranges from 4100-6500 K, whereas an office building prefers a neutral to cool light
that ranges from 3500-4100. The IESNA preferred color temperatures are shown in Table 4.4.

62

Lamp CCT (Kelvin)
<2500
2500-3000
“Warm”
2950-3500
“Neutral”
3500-4100
“Cool”
4100-5000
“Very Cool”
5000-7500
“Cold”

Table 4.4. IESNA preferred color temperatures
Application
Bulk industrial and security lighting
Used for low lighting areas (<10 footcandles) such as residential lighting, fine
dining and family restaurants, and theme parks
General feature lighting such as retail stores and galleries
General display lighting such as offices, schools, stores, industry, medicine, and
sports
General work lighting such as offices, schools, stores, industry, medicine, and
sports, as well as special applications where color is important
Special purpose lighting where color discrimination is very important

The color rendering index (CRI) is an additional property derived from the spectrum, and it is a
general indicator of a light source’s ability to reveal colors of objects in comparison to an ideal or natural
light source. The CRI is a mathematical derivation that quantifies the shift in chromaticity when the light
source is compared to eight specified pastel colors, depicted in Figure 4.6, where the amount of shift is
ranked from 0-100, 100 being the best. For almost every application, a CRI of 80 should be the minimum
[63, 66].

Figure 4.6. The IES eight pastel reference colors used for quantifying the chromaticity shift [64]
ASHRAE/ IESNA/ ANSI defines lighting power density LPD as the maximum watts of lighting
per square foot of room floor area and is determined by [63, 67, 68]
(32)
where fc is the illuminance (footcandles or lumens per square foot) that is assigned for the luminaries in
the space. The light output effectiveness (LOE) is a function of the luminous efficacy LE (lumen output
63

per watt), a coefficient of utilization CU, and depreciation in the form of lamp lumen depreciation LLD
and luminaire dirt depreciation LDD, and room surface dirt depreciation RSDD. The CU is a ratio of the
lumen output to the room cavity ratio RCR, where the RCR is a relation between the fixtures height above
a work plane (typically desk height), the room dimensions, and the type of space. [63, 67, 68] There is an
assortment of CU’s, which can be found at [69]. The LOE is determined by
(33)
In the ASHRAE/ IESNA/ ANSI 90.1 energy standards manual, two scenarios, whole building and
space-by-space, are provided that characterizing maximum LPD [63, 67, 68]. Given that the approach we
employed examined each individual space, we performed the analysis based on the space-by-space
standard. A summary of the LPDs for common places around a university is provided in the Appendix II.

4.3.2 Retrofit Options
Typically, light sources are divided into two categories, natural and artificial. A natural light
source is light from the sun and an artificial light source is from technologies such as incandescent lamps.
In 2010, the U.S. commercial building sector estimated that 2.1 billion light fixtures were installed, which
is approximately one-fourth of all installed fixtures in the U.S. The other light installations included the
residential and industrial building sector. Depicted in Figure 4.7 illustrates the 2010 percentage of
installed light fixtures in the U.S. commercial building sector [70].

High Intensity
Discharge
2%

LED Incandescent
2%
Halogen
4%
2%
CFL
10%

Incandescent
Halogen
CFL
Fluorescent Tube

High Intensity Discharge

Fluorescent Tube
80%

LED

Figure 4.7. 2010 commercial building sector estimate of installed lamps technology
64

There is a copious amount of information available about the different types of lighting, which is
highlighted in Appendix III.
When developing the retrofit options, assuming the initial system meets the LPD criteria, the first
consideration taken into account was that Michigan State University wants a breakeven time-period of 3-5
years [71]. Therefore, the net present value NPV that incurs over the 3-5 year period is determined on a
monthly scale, where the cash flows CF include the energy costs, replacement costs, O&M and each
respective discount rate. The replacement time period T is determined by
(34)
The basic NPV formula is
(35)
where t is the discrete time step. It should be noted the if the existing LPD was not up to current
standards, this could result in either a complete redesign of the lighting system, or just a relamp.
The next step is evaluating the various types of lighting technology for their CCT, CRI, and LPD,
and ensuring that each criteria is met. As stated earlier, the CCT and CRI can easily be obtained by using
a spectrometer. Following each technology, if each characteristic has been met, will be evaluated using
Eqs. (30), (34) and (36). Then, the technologies for each room that are deemed economically viable go
through a bidding process.

4.4 Phases IV-VI
Phase IV, bidding the construction and material costs of the retrofit process, is dependent upon
the size of the project. Typically these bids are then brought to the building owner or the board of
directors, and they decide if the project should go into phase V, implementation. During implementation,
the new retrofit equipment is installed and the former technology is disposed.

Then phase VI is

evaluating the retrofit to verify the energy usage assessment and economics [63, 64, 65, 67].

65

4.5 Case Studies
The retrofit design methodology is demonstrated on a classroom, a hallway, and an office. These
building spaces experience long hours of operation, and they have the same F32T8 lighting technology
that is soon to be discontinued [72]. Therefore, these areas are an ideal candidate for a retrofit.
The case studies are separated in two sections. The first section is a comparison of the current
light quality characteristics to an LED retrofit option.

When the various retrofit options were being

determined, we were strongly encouraged by Michigan State University to examine LEDs. Consequently,
a handful of LEDs were examined, but the LED that proved to have the least power consumption while
still maintaining the desired light qualities is presented. The second section demonstrates the design
methodology for obtaining the energy usage, and presents an economic evaluation. Additionally, the
hallway walk-throughs were compared to metered data for the current lighting technology energy
consumption and hours of usage.
Throughout Michigan State University, the most common lighting fixture is the 2x4 foot F32T8
3-lamp bi-level lighting troffer with a 0.87 ballast factor and a 0.98 power factor, which is also the only
light fixture present in these building spaces. The lighting technology that is housed in the fixture has a
4100 K CCT and a 78 CRI.

4.5.1 Lighting Quality
For the three building spaces, the IESNA desired CCT is 3500-4200 K, and the preferred CCT by
Michigan State University is 4000-4100 K. In addition, the CRI should be 80 at a minimum. When
measuring the CCT and CRI based on the light sources spectra, there are two separate calculations. The
first calculation utilizes the spectra to determine the CCT, whereas the second calculation, the CCT is
declared in order to calculate a precise CRI. As mentioned previously, the CRI determines how close the
light source is to its specified CCT.

66

1

0.9

0.9

0.8

0.8
Normalized Spectra Intensity

Normalized Spectra Intensity

1

0.7
0.6
0.5
0.4
0.3
0.2

0.6
0.5
0.4
0.3
0.2

CCT Ref Black Body
Test Spectrum

0.1
0
350

0.7

400

450

500

550
600
650
Wavelength (nm)

700

750

800

CCT Ref Black Body
Test Spectrum

0.1

850

0
350

400

450

500

550
600
650
Wavelength (nm)

700

750

800

850

Calculate the CCT (measured)
Declare the CCT (manufacturer)
Figure 4.8. The light quality tests for the current F32T8 bulbs. The CCT was measured off the spectra in
a, and in b, the CCT was declared
The spectra along with the CCT in reference to a black body radiator for the current F32T8
lighting technology are depicted in Figure 4.8. When just the spectra was utilized, a CCT of 3700 K was
determined, and when the CCT was fixed at 4100 K, according to the spec sheet, a CRI of 65 was found.
These results are presented in Table 4.5.
Table 4.5. CCT and CRI parameter comparison of the current F32T8 lighting technology
Measure Spectra to
Light Quality
Manufacturer
Declare CCT to Calculate CRI
Calculate CCT and CRI
CCT
4100 K
3700 K
4100 K
CRI
78
82
65
The spectra along with the CCT in reference to a black body radiator for the perspective LED
lighting technology are depicted in Figure 4.9, When just the spectra was utilized, a CCT of 3800 K was
determined, and when the CCT was fixed at 4000 K according to the spec sheet, a CRI of 87 was found.
These results are presented in Table 4.6.
Table 4.6. CCT and CRI parameter comparison of the perspective LED technology
Measure Spectra to
Light Quality
Manufacturer
Declare CCT to Calculate CRI
Calculate CCT and CRI
CCT
4000 K
3800 K
4000 K
CRI
80
84
87

67

Calculate the CCT (measured)
Declare the CCT (manufacturer)
Figure 4.9. The light quality tests for the perspective LED retrofit kit. The CCT was measured off the
spectra in a, and the CCT was declared in b
While the lighting quality characteristics of the LED lighting technology are similar to the current
F32T8 lighting technology, a side-by-side comparison exhibits the validity of Michigan State University
utilizing LEDs throughout the campus.

Current F32T8 lighting technology
Perspective LED technology
Figure 4.10. Side-by-side color comparison, where the current F32T8 lighting technology is shown in a,
and the perspective LED lighting technology is shown in b

68

4.5.2 Financial Feasibility Study
The financial feasibility of the retrofits was analyzed by using the traditional NPV project
evaluation. The NPV was calculated by culminating the specifications and capital cost of the two lighting
technologies, along with the expected annual hourly usage that pertains to each building space. Table 4.7
exhibits a feature comparison of the current lighting technology to proposed LED retrofit technology, and
Table 4.8 portrays the annual usage hours for each building space. The annual usage hours were
calculated by following the procedure outlined in Figure 4.3, where if the lights were on at single-level,
they were normalized to bi-level for comparative purposes.
Table 4.7. A comparison of the two lighting technologies that are being assessed for the financial
feasibility of a retrofit
Current Fluorescent
Perspective LED
Light Characteristics
Lighting Technology
Lighting Technology
Anticipated Lifetime (Hours)
24000
50000
Peak Power Consumption (Watts)
97.5
45
Capital Cost
$6.51
$110
Efficacy (Lumens/Watt)
61.6
88.9
Table 4.8. Estimated hourly usage on an annual basis
Usage (Hours)
Amount
Space
of Fixtures
Walk-throughs
Metered Data
Classroom
16
3830
Hallway
12
8650
8425
Office
12
5840
Then by following the procedure illustrated in Figure 4.3, for sorting through the raw data from
walk-throughs, the annual energy use for the current lighting technology, including classroom, hallway,
and office is 4480 kWh, 10120 kWh, and 6830 kWh, respectively. The annual energy cost for the
classroom, hallway, and office is $585, $990, and $670, respectively, and the replacement period for
those same locations are 6.3 years, 2.8 years, and 4.1 years respectively. Whereas, for the LED Retrofit
Kit, the annual energy use for the classroom, hallway, and office is 2050 kWh, 4640 kWh, and 3130 kWh
respectively. The annual energy cost for the classroom is $265, hallway is $450, and office is $305. The
replacement period, in years, is 13, 5.8, and 8.5, respectively.

69

Table 4.9. Financial characteristics for the lighting retrofit of the three building spaces
Current Lighting Technology
LED Retrofit Kit
Financial
Characteristics
Classroom
Hallway
Office
Classroom
Hallway
Office
Annual Energy Use
4480
10120
6830
2050
4640
3130
(kWh)
Annual Energy Cost
$585
$990
$670
$265
$450
$305
Replacement Period
6.3
2.8
4.1
13
5.8
8.5
(years)
As depicted in Figure 4.11 and shown in Table 4.10, the classroom breakeven time period, in
years, is 6.0, while the office is 3.8, and the hallways is 2.5. The amount saved over the lifetime of an
LED Retrofit Kit is $2175 for the classroom, $1731 for the office, and $1806. The payoff system and
return on investment in the classroom is 111%, for the office is 119%, and for the hallway is 124%. The
return on the investment period in the classroom, office, and hallway is 13.1 years, 8.6 years, and 5.8
years respectively. The CO2 reduced was based on the 6.9x10-4 metric ton of CO2/kWh [73].

Figure 4.11. Economic evaluation of the three building spaces. The breakeven time period is where the
current lighting technology intersects with the proposed LED technology

70

Table 4.10. Economic feasibility comparison of the two lighting technologies
Financial Characteristics
Classroom
Hallway
Office
Break-Even Time Period (Years)
6.0
2.5
3.8
Amount saved over lifetime of LED
$2175
$1806
$1731
Retrofit Kit
Return on Investment
111%
124%
119%
Return on Investment Period (Years)
13.1
5.8
8.6
CO2 Reduced Over Lifetime of LED
17.2
12.9
12.9
Retrofit Kit (Metric Tons)

4.6 Discussion
The breakeven time period, amount saved, and return on investment were based on the energy
savings in comparison to the current lighting technology. This analysis included the capital cost of the
LED retrofit kit, installation and installation materials such as wire. Furthermore, the replacement costs of
the current technology, labor, and commodity rate of electricity were calculated at a fixed rate and
adjusted for their respective inflation. The labor rates were provided from Spencer Speerbreaker, a Master
Electrician at Michigan State University, where each building space scenario was categorized on a fixture
basis for the time required for installation, number of laborers (master electricians and apprentices) as
well as equipment need such a scissor-lift, ladder, or scaffolding. The degradation of the each lighting
technology in regards to energy consumption was not included. It should be noted this office under study
was a graduate student office, and typically these offices experience longer than normal periods of use.
The energy consumption of the current and perspective lighting technology was gathered using metered
data. However, the logged usage data for the hallway was not included in the financial analysis because of
large outlier values being present throughout the data.

4.7 Summary and Conclusions
The results of the case studies demonstrate that LED technology not only reduces energy costs,
but also has a noteworthy return on investment and a substantial reduction in greenhouse gas emissions.
While the classroom analysis may not fall in MSU’s desired range for a retrofit, it is clear that this office
and hallway are economically viable.

71

APPENDICES

72

Appendix I: Solar Photovoltaic Time Periods
The financial analysis for solar PV was performed on a monthly basis, where the warranty and
replacement time periods are based on industry norm [74]. Table A.1 depicts the initial capital costs,
which occur at time 0. Additionally, the inverter’s warranty expires after five years, and typically needs
replacement every ten years. The warranty on the modules and the disconnect switches are 25 years, but
are expected to be able to last up to the industry norm of 30 years, which is the length of the solar PV
project analysis. In our analysis, we assumed that it would take one month to install a rooftop solar PV
system.
Table A.1. Time periods for BoM and expected periods for replacement
0
5
10
15
20
25
Warranty
Modules
X
Expires
Mounting System
X
Storage
X
Warranty
Warranty
Warranty
Inverter
X
X
X
Expires
Expires
Expires
Tracker
X
Warranty
Combiner Boxes
X
Expires
Meter
X
System Monitor
X
Warranty
DC-AC Switch
X
Expires
Fuses and Holders
X
Wires, Conduit,
X
X
X
Nuts, Bolts, etc.
Time (Years)

30

Table A.2 depicts the construction labor time periods. The construction labor time periods follow
directly in line with the Table A.1 for initial capital costs and replacements of components.
Table A.2. Time Periods for Labor Costs and Expected Periods for Replacement
Time (Years)
0
5
10
15
20
25
Electrical
X
X
X
Hardware
X
X
X
Overhead
X
Profit
X

73

30

Table A.3 depicts the indirect capital costs. All of these costs occur during the construction stage
of the project, and sales tax is applied for when the inverter is being replaced. The land cost and site prep
is considered to be $0.00, because MSU owns its buildings and being that MSU has 110+ buildings, they
can choose the building that needs minimal preparation.

Time (Years)
Permitting &
Commissioning
Grid Interconnect
Land
Site Prep
Sales Tax

0

Table A.3. Indirect capital costs time period
5
10
15
20

25

30

X
X
X
X
X

X

X

Table A.4 depicts the discrete indirect costs. The maintenance and cleaning costs is largely due
to cleaning, which should be performed monthly. Additionally, the Chairs at MSU’s Infrastructure &
Planning Facilities requested that we account for insuring the system, and that they are billed monthly, as
well as a fund for inverter replacement such that they are able to prepare for these costs.

Time (Years)
Maintenance and
Cleaning
Insurance
Inverter
Replacement
Reserve

0

Table A.4. Discrete indirect costs
5
10
15

20

25

30

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

Table A.5 depicts the trend in a module. The module produces energy every day, but experiences
a degradation in the power generation capabilities.

Time (Years)
Panel Degradation
Energy Generation

0

Table A.5. Module time periods
5
10
15
X
X
X
X
X
X

74

20
X
X

25
X
X

30
X
X

Appendix II: Lighting Power Density for Common Spaces Around a University
Table A.6. The ASHRAE/ IESNA/ ANSI 90.1 LPD for the common space types around a university
Common Space Type
LPD (W/ft2)
RCR Threshold
Atrium
First 40 ft in height
0.03 per ft (height)
NA
Height above 40 ft
0.02 per ft (height)
NA
Audience Seating Area - Permanent
For Auditorium
0.79
6
For Performing Arts Theatre
2.43
8
For Motion Picture Theatre
1.14
4
Classroom/Lecture/Training
1.24
4
Conference/Meeting/Multipurpose
1.23
6
Corridor/Transition
0.66
Width<8ft
Dining Area
0.65
4
For Bar Lounge/Leisure Dining
1.31
4
For Family Dining
0.89
4
Dormitory
0.38
8
Dressing/Fitting Room for Performing Theatres
0.40
6
Electrical/Mechanical
0.95
6
Food Preparation
0.99
6
Laboratory
For Classrooms
1.28
6
For Medical/Industrial/Research
1.81
6
Lobby
0.90
4
For Elevator
0.64
6
For Performing Arts Theatre
2.00
6
For Motion Picture Theatre
0.52
4
Locker Room
0.75
6
Lounge/Recreation
0.73
4
Office
Enclosed
1.11
8
Open Plan
0.98
4
Parking Garage
0.19
4
Restrooms
0.98
8
Stairway
0.69
10
Storage
0.63
6
Workshop
1.59
6

75

Appendix III: Lighting Technologies
A.III.1 Daylight
Daylight is the combination of direct and indirect sunlight that is utilized to illuminate a room via
a window or skylight. When daylight is employed in a building, a delicate balance between the HVAC
system and the solar heat gain must be considered. Certainly, using daylight in comparison to artificial
lighting has significant energy savings in regards to the lighting system. However, a room that uses
daylight during the warmer months will inevitably experience solar heat gain, which in turn creates a
higher demand on the HVAC system to cool the room. Thus ensuing a delicate balance of daylight to the
HVAC energy demand.
Nonetheless, daylight is the most desirable light source in a building because of excellent
characteristic properties, as well as providing psychology benefits to occupants. Daylight has a CCT of
6500, a CRI of 100, and is a full spectrum light source, thus meaning it covers all wavelengths of the
visible light spectrum, or all wavelengths that are useful to plant and animal life. Thus, many artificial
light sources should roughly mimic the same properties of daylight [63, 67, 75].

A.III.2 Incandescent Light
Thomas Edison invented the incandescent light bulb in 1879. The incandescent bulb, shown in
Figure A.1, is comprised of a very fine tungsten filament wire, sealed by a glass bulb, and filled with
argon gas. When electrical current passes through the filament, the filament will heat up and emit light.
This technique converts five to ten percent of the energy into light, and the rest is emitted as heat [67, 76].

Figure A.1. Typical incandescent light bulb
76

In 2010, the incandescent lamp was the most common light bulb in all lighting applications. It
accounted for roughly 62% of installed lights in residential applications, but only 4% in non-residential
applications. The reason behind this dramatic installation difference in the various building sectors is due
to a recent DOE bill, which was created to phase out the use of these bulbs in non-residential applications
[70, 77].
There are many advantages to the incandescent lamp. The main advantage is the relatively low
capital cost, dimming capabilities and no need for extra equipment such as a ballast. Additionally, the
incandescent lamp has a CRI of 100 and a CCT that ranges from 2400-2900 K. However, there are many
disadvantages as well. Roughly 5-10% of the electrical energy input produces light and the rest is emitted
as heat. This heat gains leads to the building’s HVAC equipment working more often. Furthermore, in
comparison to other artificial lighting technology, the incandescent lamp is the least luminous, 12-18
lumens per watt, and has a lifetime of 750-1,500 hours [63, 78].

A.III.3 Halogen Lights
The halogen lamp, shown in Figure A.2, is much like the incandescent lamp. They both have a
tungsten filament, but the halogen lamp contains bromine and iodine gas encased in quartz at a pressure
of 7-8 atm, as compared to argon gas encased in glass. The halogen lamp produces light via the halogen
cycle. The cycle starts by passing an electrical current though the filament. This causes the filament to
heat up and release tungsten atoms. At moderate temperatures, the tungsten atoms will chemically unite
with the halogen gas, thus creating a halide. At higher temperatures, the halide will dissociate, which
releases tungsten atoms and frees the halogen atoms to repeat the process[66, 76].

Figure A.2. Typical halogen light bulb
In 2010, the halogen lamp accounted for 4% of installed lights in residential applications, and
2.3% of commercial applications [70]. One reason behind the small market percentage is the type of

77

applications used for this technology. Halogen lamps, known for high luminous output, are typically used
in stadium lights, work lights, and film lights [63, 76].
The main advantage is the relatively low capital cost, dimming capabilities no need for extra
equipment, such as a ballast, and no warm up time. Additionally, the halogen lamp has a CRI ranging
from 95-100, a CCT that ranges from 2800-3400 K, and an efficacy of 16-29 lumens per watt.
Furthermore, the lifetime of the lamp varies due to on-off cycles, but statistically has a lifetime of 2,0004,000 hours. However, there are many disadvantages as well. The halogen lamp becomes extremely hot
and easily capable of causing severe burns. In addition, this lamp is sensitive to oils left behind from
human skin. These oils when heated up can cause imbalance, resulting in possible rupture of the bulb [63,
76].

A.III.4 Fluorescent Light
The fluorescent lamp contains a mixture of argon gas and mercury vapor, sealed by a glass tube.
This glass tube has a phosphorous coating on the inside wall, and at each end of the glass tube, there are
two electrodes wired into an electrical circuit. When electrical current flows to the electrodes, it creates a
substantial voltage across at each electrode. This causes electrons to migrate through the argon gas from
one end of the tube to other. In doing so, some of the mercury in the tube changes from liquid to gas.
When electrons and charged atoms collide with the gaseous mercury atoms, the electrons of mercury
atoms become excited and jump to higher energy levels. When the electrons return to the original energy
level, they mainly release photons in the UV wavelength range. When these photons encounter the
phosphor atoms, it causes the phosphor atom to be bumped to a higher energy level. When the photon
returns to its original position, energy is released in the form of another photon, which also releases white
light. The color of the light is varied with different phosphor combinations. The typical fluorescent lamps
CCT range is 2,700-6,500K and CRI range is 80-90 [76].
For this thesis, the fluorescent lamp is divided into two parts. The first part describes the compact
fluorescent light (CFL), and the second part describes the fluorescent tubular light.

78

A.III.4.1 Compact Fluorescent Light
The compact fluorescent light (CFL), shown in Figure A.3, is one alternative to incandescent
lamps. In 2010, the CFL accounted for 23% of installed lights in residential applications, and 10% of
commercial applications [70]. The reason behind this dramatic installation difference in the various
building sectors is due to that most applications for CFLs are for specialty decorative applications [63,
76].

Figure A.3. Typical compact fluorescent light bulb
The greatest advancement in CFL technology is the built-in electronic ballast. Originally, the
CFL contained a magnetic ballast, which created an unpleasant buzzing sound, and a long startup time
[79].
The main advantage to using CFLs is that it can replace almost any incandescent lamp with little
to no change to the fixture, and it is relatively inexpensive. Additionally, in comparison to the
incandescent lamp, the CFL has an expected lifetime 13 times greater, and requires 80% less electrical
energy to produce the equivalent amount of light [76]. However, the main disadvantage of CFLs is that
they contain mercury, a highly toxic chemical. If any bulb is broken, one should take steps to expel the
gas in that room. However, some hardware stores offer CFL recycling at no charge to the customer. The
other disadvantage is that not all CFLs can be dimmed, or designed for rapid on/off switching [78].

A.III.4.2 Fluorescent Tubular Light
The fluorescent tubular lamp, sometimes referred to as linear fluorescent lamp is shown in Figure
A.4. In 2010, the fluorescent tubular lamp accounted for 29% of all artificial light applications, where
79

10% was for residential applications, and 80% was for commercial applications. The reason behind this
dramatic installation difference in the various building sectors is due to that most applications for
fluorescent tubular lamps are for lighting applications compared to aesthetics [63, 76].
The rectangular and strip fluorescent tube fixtures, containing one to four lamps, are the most
common in commercial applications. Until the 1980s, the standard lamp was the T12 lamp in 4 and 8-foot
lengths, whereas today the most common lamp is the T8. In the fluorescent lighting nomenclature, the
“T” designation stands for tubular, the shape of the lamp. The number following the T gives the lamps
diameter in eighths of an inch. Thus, the diameter of a T12 lamp is twelve-eighths of an inch, or one-andone-half inches. When comparing the light output of the T8 and the T12, the T8 has roughly 10% more
light output per watt [79].
The fluorescent lamp has the same design principles as the CFL, except the ballast is separate
from the actual lamp.

Therefore, this allows for large amount of flexibility of lamp and ballast

combinations for designing a room. Fluorescent lamps are very energy efficient in comparison to
incandescent lamps, where fluorescent lamps use 85% less energy, and offer a lifetime of about 20,000
hours. The main disadvantage for fluorescent lamps is the constant flickering. The flickering leads to
health effects of the occupants, such as headaches and eye strain [63, 79, 76].

Figure A.4. Typical fluorescent tube light

A.III.5 Light Emitting Diodes LED
As far as energy efficiency and LPW, the LED lighting technology is considered the best
alternative to incandescent lamps. However in 2010, less than 1% of residential applications and less than
2% of commercial applications used LEDs. These low numbers are largely due to the high capital cost of
the light during 2010, which was about $1.71/klm as compared to today, where the cost is about
$0.97/klm. Also from 2010 to 2012, the market has seen a 350-3,500% installation growth rate,
depending on the type of LED bulb [70, 80].

80

The LED lamp, unlike incandescent, halogen, and fluorescent lights, is based on solid-state
lighting (SSL) technology. The term SSL refers to light emitted by electron movement in a semiconductor
material. To further explain this, a diode consists of doped semiconducting chips to create a p-n junctions.
When current flows across the p-n junction, the n-side releases electrons and meet holes on the p-side. In
doing so, the electrons fall into a lower energy level and release energy in the form of a photon .
The main advantages of LEDs are they consume 10-30% less energy to produce the same amount
of light. They also have a lifetime of about 50,000 hours and a CCT ranging from 2,700-5,000K and a
CRI range of 70-95. The main disadvantage is the capital cost of LEDs. This is main deterrent for
switching over to LEDs. Also generally the CCT and CRI is much hotter than what most prefer. But, this
is also a hot-topic of R&D [76].

Figure A.5. Typical LED diode

81

Appendix IV: Economic Fundamentals
This section provides an overview of a few elementary financial terminologies such as cash
flows, interest rates, time periods, discount rates, expected present value, taxes, system depreciations, and
fixed charge rates. As well as, this section describes capital budgeting techniques for net present value,
whole-life cost, internal rate of return, and levelized cost of energy.

A.IV.1 Cash Flows
The cash flows are the most important economic analysis for almost any investment. The cash
flows for our project analysis can be broken up into 3 main components, which are: 1) overnight cost of
capital; 2) discrete indirect costs; and 3) accrued energy savings. The overnight cost of capital accounts
for all the components of a project, as well engineering, labor, land and project overhead. The monthly
discrete costs include O&M, unscheduled maintenance, component replacement reserve, insurance and
property taxes. The component replacement reserve for a solar PV project is the inverter, whereas for a
lighting system retrofit project, this could be a driver for a LED or a power pack for an occupancy sensor.
The accrued energy saving for a solar PV system is a function of the amount of energy generated from the
PV system, and the commodity rate of electricity from the power plant. The accrued energy savings from
lighting system retrofit is a function of the reduced amount of energy from lighting as compared to the
current technology, and the commodity rate of electricity from the power plant .

A.IV.2 Interest Rates
The cash flows can be expressed either in current dollars or in constant dollars. Current dollars
are defined as actual cash flows observed in the market, and change in value over time due to inflation
and or escalation, i.e. a dollar today is not worth the same as a dollar tomorrow. Constant dollars are
defined as the dollar worth at the base year. Constant dollars values do not take into account inflation and
or escalation, and therefore not considered in the financial analysis. The current dollar equation of a cash
flow is

82

(36)
Where CDCF is current dollar cash flow, CFt is cash flow at time t, e is the interest rate, and t is time
[81].
There is a multitude of inflation rate indices, such as Consumer Product Index (CPI), Gross
Domestic Product (GDP), and Producer Price Index (PPI). The CPI is the most widely followed inflation
measure, and measures the average change in price, in prices of a market basket of consumer goods and
services, and price trends at the retail level. The CPI for the United States, as defined by the Bureau of
Labor Statistics, is “A measure of the average change over time in the prices paid by urban consumers for
a market basket of consumer goods and services”. The national CPI is calculated monthly from the
average change in price of goods, which included in the basket, for 85 separate locations. The GDP is the
market value of the national output of goods and services, in a given period of time. The GDP per capita
is typically recognized as an indicator of a country’s standard of living. The movement of the GDP often
closely mimics the CPI [81].

A.IV.3 Time Periods
The time period of a cash flow is imperative for any financial analysis. The most important time
periods from a reporting standpoint is the base year, dollar year, and investment year. The cash flows
from each time period are converted to the base year, sometimes called year zero. The dollar year is at a
different period than the base year, which could be past or future. The dollar year is converted to the base
year by using its respective interest rate and the number of periods from the base year. The investment
year is the year when the investment occurs [81].
Other important time periods for a financial analysis include the lifetime of the investment L, the
analysis period N, the depreciation period M, the finance period Y, and the levelization period P. The
lifetime of the investment is an estimate of the useful lifetime of an investment. The typical lifetime for a
solar PV system’s is 30 years, and a LED bulb is 50,000 hours. The analysis period is a period of time
that an investment is being evaluated. The depreciation period, which is particularly important for tax

83

purposes, is the period of time an investment is amortized. The finance period, sometimes referred to as
the loan period, is the period of time for which finance is structured. The levelization period is the period
of time used for calculating the levelized cash flow stream. In a financial analysis, the time points selected
are generally equally, i.e. yearly or monthly, but could different [81].
To summarize time periods with an example, let’s say a solar PV system started construction in
2010, but the construction took two years to complete. Therefore, the base year is 2010, which is when
all the initial capital costs took place. The project goes into service in 2013, and any cash flows from
2013 on, need to be converted to the base year of 2010. Some of these cash flows occur monthly, such as
O&M, while others occur many years since operation, such as replacing the inverter. However, all of
these cash flows need to be converted to the base year of 2010.
Selecting a time and analysis time period is essential. If reporting cash flows, without reporting
the base year, the data is useless. Many times in practice, the investments lifespan is equal to the analysis
period.

A.IV.4 Discount Rate
The well known saying that, “a dollar today is worth more than a dollar tomorrow”, typically
referred to as the time value of money, is because a dollar received today can be invested to earn interest
immediately. The discount rate acts as a measure of the time value of money, and is pivotal to the present
value calculation. The discount rates may be viewed as inherent risk of an investment. In this analysis,
we consider the prime lending rate from a bank as the discount rate, because it may be viewed as the
opportunity or risk of investing in project vs. investing the money in a bank [81].

A.IV.5 Depreciation, Tax Rates, and Renewable Energy Tax Credits
Depreciation is the reduction in value of assets, and is a tool for recovering the cost of assets via
an income tax deduction. Depreciation is strictly an accounting detail and does not involve cash flows, or
the deterioration of the physical asset. The depreciable base is the capital sum that may be recovered by
depreciation, and the adjusted base is when the depreciable base is adjusted annually by the depreciable

84

amount [81]. However, Michigan State University is a public university that typically holds onto an
investment past its useful life, and is tax exempt. Thus, depreciation, renewable energy tax credits and all
tax rates, with the exception of sales tax, are not considered.

A.IV.6 Expected Present Value
Expected present value analysis is the anticipated current worth of a future cash flow stream,
converted to the base year, given a specified rate. In other words, expected present value is today’s value
of future cash flows. Expected present is one of the most fundamental and important analysis tools for any
investment. When calculating the expected present value, it is essential to keep track of constant or
current dollars. Real discount rates must be used with constant dollars, and nominal discount rates, which
include inflation, must be used with current dollars [81].
The basic expected present value equation pv, of a cash flow CFt, occurring at time t, with a
discount rate d is
(37)
An annuity immediate is a periodic cash flow of fixed size, for example, routine O&M. The
expected present value of an annuity immediate with interest rate i, and n cash flows is
(38)

A.IV.7 Net Present Value
The most widely accepted financial method to determine the profitability of an investment is the
net present value NPV. It examines the difference between the present value cash inflows, sometimes
called revenue, and the present value of cash outflows, sometime called costs. The NPV is sensitive to the
reliability of an investment, because it takes into account inflation and discount rates. A project that has a
positive NPV is a viable investment. The NPV formula, where CFt, is the cash flows at time t, and r is the
various rates, is
(39)

85

A.IV.8 Whole-Life Cost
The whole-life cost WLC, or life-cycle cost are the costs sustained during the ownership of the
project. This includes the initial capital costs, indirect costs, discrete costs, etc. The WLC analysis is a
well-defined procedure to estimate the overall present value costs of a project, and compare the results to
alternatives [81].

A.IV.9 Internal Rate of Return
The internal rate of return IRR is a capital budgeting calculation for comparing the profitability of
a project. The internal rate of return is the rate of return that makes the net present value of all cash flows
for a particular investment equal to zero. IRR is not recommended for rating mutually exclusive projects,
but should rather be used when deciding if a single project is worth investing into [81]. The IRR is found
by solving the polynomial developed in

40

A.IV.10 Levelized Cost of Energy
The levelized cost of energy LCOE is a calculation that compares the cost of energy from
different fuel sources. In our case, the LCOE is the cost of energy generated from a solar PV system. The
LCOE includes the entire present value system costs, such as initial capital, loan payments, O&M, etc.
This type of calculation creates a comparison, which helps guide decision makers [81].

86

BIBLIOGRAPHY

87

BIBLIOGRAPHY

[1] U.S. Energy Information Administration, "Monthly Energy Review April 2014," U.S. Department of
Energy, Washington D.C., 2014.
[2]
P. Thakur, S. Ballard and R. Nelson, "Radioactive Fallout in the United States Due to the Fukushima
Nuclear Plant Accident," Journal of Environmental Monitoring, no. 5, pp. 1317-1324, March 2012.
[3] G. Liu, B. Liu, W. Wang, J. Zhang and R. Athalye, Advanced Energy Retrofit Guide - Practical Ways
to Improve Energy Performance, Richland, Washington: Department of Energy, 2011.
[4] J. Kelso, "2007 Buildings Energy Data Book," Silver Spring, Maryland: D&R International, Ltd,
2007, pp. 1.1-1.17.
[5] The Franklin Institute, "Photovoltaics Student Guide," 2013.
[6] U.S. Department of ENergy, "The History of Solar," Energy Efficiency and Renewable Energy,
Washington D.C..
[7] J. Perlin, From Space to Earth: The Story of Solar Electricity, Ann Arbor: Harvard University Press,
2002.
[8] G. Masters, Renewable and Efficient Electric Power Systems, 2nd ed., John Wiley & Sons, 2013.
[9] A. Sproul, “Understanding the p-n Junction,” Resources for the Secondary Science, 2007.
[10] S. Wenham, M. Green, M. Watt and R. Corkish, Applied Photovoltaics, Sterline, Virginia: Routledge,
2011.
[11] A. Pradhan, D. S. Ali and C. Jena, “Analysis of Solar PV Cell Performance with Changing Irradiance
and Temperature,” International Journal Of Engineering And Computer Science, vol. 2, no. 1, pp.
214-220, January 2013.
[12] A. Parretta, M. Bombace, G. Graditi and R. Schioppo, "Optical Degradation of Field Aged C-Si
Photovoltaic Modules," Solar Energy Materials & Solar Cells, vol. 86, pp. 349-364, 2005.
[13] D. C. Jordan and S. R. Kurtz, "Photovoltaic Degradation Rates - An Analytical Review," Progress in
Photovoltaics: Research and Applications, June 2012.
[14] R. Ramaprabha and B. L. Mathur, "Impact of Partial Shading on Solar PV Module Containing Series
Connected Cells," International Journal of Recent Trends in Engineering, vol. 2, no. 7, pp. 56-60,
November 2009.
[15] B. Marion, J. Adelstein and K. Boyle, "Performance Parameters for Grid-Connected PV Systems,"
2005.
[16] Solar Energy Industries Association, Greentech Media Inc, “U.S. Solar Market Insight,” 2012.
[17] National Renewable Energy Lab, "Best Research-Cell Efficiencies," 2013.
88

[18] A. Luque and S. Hegedus, Handbook of Photovoltaic Science and Engineering, John Wiley & Sons,
LTD, 2011.
[19] International Renewable Energy Agency, "Solar Photovoltaics," Renewable Energy Technologies:
Cost Analysis Series, vol. 1, no. 4, June 2012.
[20] T. Saga, "Advances in Crystalline Silicon Solar Cell Technology for Industrial Mass Production,"
NPG Asia Materials, vol. 2, no. 3, pp. 96-102, July 2010.
[21] 6W Research, "Thin Film PV Market (2011-2016): Global Market Forecast," Business Wire, New
York, 2011.
[22] J. Poortmans and V. Arkhipov, Thin Film Solar Cells: Fabrication, Characterization and Applications,
Belgium: John Wiley & Sons, Ltd, 2006.
[23] H. Ullal, "Overview and Challenges of Thin Film Solar Electric Technologies," in World Renewable
Energy Congress X and Exhibition, Glasgow, 2013.
[24] T. Markvart and L. Castaner, Practical Handbook of Photovoltaics Fundamentals and Applications,
Elsevier, 2003.
[25] W. Bower, "International Guideline for the Certification of Photovoltaic System Components and
Grid-Connected Systems," Internation Energy Association, Paris, 2002.
[26] L. Arribas, "World-Wide Overview of Design and Simulation Tools for Hybrid PV Systems,"
International Energy Agency, 2011.
[27] G. Klise and J. Stein, "Models Used to Assess the Performance of Photovoltaic Systems," Sandia
National Laboratories, Albuquerque, 2009.
[28] “Modeling Tools,” National Renewable Energy Lab, [Online]. Available:
http://www.nrel.gov/international/tools/fate-2hpt.
[29] “RETscreen International,” [Online]. Available: http://www.retscreen.net/.
[30] “HOMER Energy,” [Online]. Available: http://www.homerenergy.com/.
[31] “Energy Research Centre,” University of Cape Town, [Online]. Available:
http://www.erc.uct.ac.za/hybrid-deisgner.
[32] “PVSYST Photovoltaic Software,” [Online]. Available: http://www.pvsyst.com/en/.
[33] Solar Energy Industries Association, Greentech Media Inc , “U.S. Solar Market Insight,” 2014.
[34] M. Platzer, "U.S. Solar Photovoltaic Manufacturing: Industry Trends, Global Competition, Federal
Support," Congressional Research Service, 2012.

89

[35] R. Ramesh, "Sunshot Vision Study," Washington D.C., 2012.
[36] E. Holt and L. Bird, "Emerging Markets for Renewable Energy Certificates: Opportunities and
Challenges," National Renwable Energy Lab, Golden, 2005.
[37] K. Cory, J. Coughlin and C. Coggeshall, "Solar Photovoltaic Financing: Deployment on Public
Property by State and Local Governments," National Renewable Energy Lab, Golden, 2008.
[38] U.S. Department of Energy, North Carolina Solar Center, "Database of State Incentives for
Renewables & Efficiency," U.S. Department of Energy, North Carolina Solar Center, Interstate
Renewable Energy Council, Energy Efficiency & Renewable Energy, 2014. [Online]. Available:
http://www.dsireusa.org/. [Accessed February 2014].
[39] P. Glasserman, Monte Carlo Methods in Financial Engineering, New York: Springer, 2003.
[40] A. Goodrich, T. James and M. Woodhouse, "Residential, Commercial, and Utility-Scale Photovoltaic
(PV) System Prices in the United States: Current Drivers and Cost-Reduction Opportunities,"
National Renewable Energy Lab, 2012.
[41] MathWorks, MATLAB.
[42] Image Magick, 1999.
[43] L. Shapiro and G. Stockman, Computer Vision, Upper Saddle River, New Jersey: Prentice Hall,
2001.
[44] R. Gonzalel, R. E. Woods and S. L. Eddins, Digital Image Processing Using MATLAB, 2nd ed.,
Gatesmark Publishing, 2009.
[45] R. Ramabadran and B. Mathur, "Effect of Shading on Series and Parallel Connected Solar PV
Modules," Modern Applied Science, 2009.
[46] A. J. Marsh, "Why Do Shading Calculations Take so Long?," Natural Frequency Journal, no. 4,
November 2007.
[47] D. J. Luchini Colbry, Human Face Verification by Robust 3D Surface, Michigan State University:
ProQuest, UMI Dissertations Publishing, 2006.
[48] N.-E. Yang, J. W. Lee and R.-H. Park, "Depth Map Generation Using Local Depth Hypothesis for
2D-To-3D Conversion," International Journal of Computer Graphics & Animation, vol. 3, no. 1,
January 2013.
[49] W. Huang and K. He, “A New Heuristic Algorithm for Cuboids Packing with No Orientation
Constraint,” Computers and Operations Research, no. 36, pp. 425-432, 2009.
[50] D. S. Liu, S. Y. Huang and S. K. Goh, “A Two-Dimensional Bin Packing Problem with Size
Changeable Items for the Production of Wind Turbine Flanges,” European Journal of Operational
90

Research, no. 190, pp. 357-382, 2008.
[51] A. Pasha, “Geometric Bin Packing Algorithm for Arbitrary Shapes,” Gainesville, 2003.
[52] Y.-P. Chang, “Optimal The Tilt Angle for Photovoltaic Modules Using PSO Method with Nonlinear
Time-Varying Evolution,” Energy, no. 35, pp. 1954-1963, 2010.
[53] J. D. Mondol, Y. Yohanis and B. Norton, “Optimal Sizing of Array and Inverter for Grid-Connect
Photovoltaic Systems,” Solar Energy, no. 80, pp. 1517-1539, 2006.
[54] K. Kalaitzakis and G. Stavrakakis, “Size Optimization of a PV System Installed Close to Sun
Obstacles,” Pergamon, vol. 57, no. 4, pp. 291-299, 1996.
[55] H. Tian and F. Mancilla-David, “Determination of Optimal Configuration for a Photovoltaic Array
Depending on the Shading Condition,” Solar Energy, no. 95, pp. 1-12, 2013.
[56] T. Khatib, “Optimization of a Grid-Connected Renewable Energy System for a Case Study in Nablus,
Palestine,” International Journal of Low-Carbon Technologies, March 2013.
[57] A. Dobos, "PVWatts Version 1 Technical Reference," National Renewable Energy Laboratory,
Golden, 2013.
[58] H. Haeberlin, "Optimum DC operating Voltage for Grid-Connected PV Plants: Chose of Vmpp for
Measurement of Efficiency and Vmpp-stc at PV Plants for Grid-Connected Inverters with a Wide DC
Input Voltage Range," in 20th European Photovoltaic Solar Energy Conference, Barcelona, 2005.
[59] National Electric Code, 52 ed., Quincy, Massachusetts: National Fire Protection Agency, 2011.
[60] D. King, W. Boyson and J. Kratochvil, "Photovoltaic Array Performance Model," Sandia National
Laboratories, Albuquerque, 2004.
[61] Black & Veatch, "Cost and Performance Data for Power Generation Technologies," Black & Veatch,
Overland Park, 2012.
[62] A. Williams, B. Atkinson and K. Garbesi, "Lighting Contols in Commercial Buildings," Leukos, vol.
8, no. 3, pp. 161-180, January 2012.
[63] D. DiLaura, K. Houser, R. Mistrick and G. Steffy, The Lighting Handbook, Illuminating Engineering
Society, 2011.
[64] Lawrence Berkeley National Laboratory, Lighting Retrofit Handbook, Seattle, Washington: The U.S.
Department of Energy: Federal Energy Management Program, 2001.
[65] J. Kesselring, "Lighting Retrofit Manual," Palo Alto, 1998.
[66] J. Snyder, "Light Sources and Color," Lighting Research Center, vol. 8, no. 1, October 2004.
[67] C. Eley, T. M. Tolen, J. R. Benya, F. Rubinstein and R. Verderber, "Advanced Lighting Guidelines,"
91

Eley Associates, Pleasant Hill, 1993.
[68] ASHRAE, ANSI and IESNA, Energy Standard for Buildings Expect Low-Rise Residential Buildings,
2012.
[69] Illuminating Engineering Society of North America, Lighting Power Density (LPD) Public User’s
Manual, 2005.
[70] M. Ashe, D. Chwastyk, C. d. Monasterio, M. Gupta and M. Pegors, "2010 U.S. Lighting Market
Characterization," U.S. Department of Energy, 2012.
[71] Michigan State University, "Energy Conservation and Optimization Measures at Michigan State
University," East Lansing, 2012.
[72] "Illuminating Engineering Society," 2012. [Online]. Available:
http://www.ies.org/pdf/publicpolicy/DOE-Office-of-Hearings-and-Appeals-Approves-T8Fluorescent-Lamp-Exception.pdf.
[73] United States Environmental Protection Agency, "Clean Energy," [Online]. Available:
http://www.epa.gov/cleanenergy/energy-resources/refs.html. [Accessed February 2013].
[74] K. Ardani and R. Margolis, "2010 Solar Technologies Market Report," U.S. Department of Energy,
Golden, 2011.
[75] P. Boyce, "The Impact of Light in Buildings on Human Health," Indoor and Built Environment, 2010.
[76] "Lighting Technologies: A Guide to Energy-Efficient Illumination," U.S. Environmental Protection
Agency, Washington D.C..
[77] "Illuminating Engineering Society," 2011. [Online]. Available: http://www.ies.org/LDA/Enewsletter/2012/April/newswire/2012_04-nema.cfm. [Accessed 2013].
[78] "Information on Compact Fluorescent Light Bulbs (CFLs) and Mercury," 2010.
[79] M. Krarti, Energy Audit of Building Systems, vol. II, F. Kreith, Ed., Boca Raton, Florida: Taylor &
Francis Group, 2011.
[80] M. Yamada and D. Chwastyk, "Adoption of Light-Emitting Diodes in Common Lighting
Applications," Department of Energy, 2013.
[81] S. Broverman, Mathematics of Investment and Credit, 5 ed., ACETEX Academic Series, 2010.

92