Design of Roof PV Installation in Oslo: Siv Helene Nordahl

Design of Roof PV Installation in Oslo
Siv Helene Nordahl
Master of Energy and Environmental Engineering

Submission date: June 2012
Supervisor: Lars Einar Norum, ELKRAFT
Norwegian University of Science and Technology

Department of Electric Power Engineering
Problem Description
The international photovoltaic industry has experienced a powerful growth, and

has had a yearly growth between 30 − 135 % the last three years. During this
growth the module and inverter prices have been reduced to a level where it is
competitive with conventional power production in several countries. With this
tendency in mind it is increasingly interesting to evaluate photovoltaic installa-
tions in Norway.
The objective of this thesis is to design a suiting photovoltaic system for a flat
roof in Oslo, Norway. The production of the photovoltaic system is found with
the use of the simulation software PVsyst. Input to the simulation software such
as meteorological data has been evaluated. The energy production of the final two
or three alternative systems is evaluated with respect to the energy consumption
within the building where the installation is to be located. The regulations when
connecting a photovoltaic system to the grid are also studied.
Assignment given: Trondheim, 23.01.2012

Supervisor: Lars Norum,
Department of Electrical Power Engineering
iii
iv
Summary
This thesis is centered around the design of a grid-connected photovoltaic (PV)

roof installation at a specific location in Oslo, Norway. The motivating factor in
this study has been the growth of the solar industry globally, while there has been
little to none larger PV investments in Norway. The objective is to investigate
how much renewable PV energy that can be produced from the designed system,
with an electrical focus.
Factors such as the suns position during each day of the year, the shadings on
modules, the electrical effect of shading and bypass diodes, and other factors
influence the production of a PV installation. Due to the complexity of power
production in a PV system, the simulation software PVsyst was used as support.
A 3D representation of the building and shading elements was constructed in
the simulation program for shading calculation purposes. Meteorological data
from local weather stations in Lier, Ås and Blindern was compared with me-
teorological data provided by interpolation and satellite images. The distance
between modular rows was dimensioned after a shading criterion so that there
would be no shading from other modular rows during spring equinox (March
21st). The modular tilt was adjusted (from the optimal tilt angle of 40◦ ) in order
to reduce shading loss and improve the performance ratio of the system. The
number of module and inverter types and manufacturers was limited to three
different module types, and four different inverter series. The simulated produc-
tion from the three best alternatives, based on performance ratio and production
were compared with the energy consumption in the building. Simple economical
evaluations of the three best alternatives have been performed using the simple
payback method and life cycle costing.
As a result of the limited area on the roof, the shading objects and the dimen-
sioning criteria (maximize performance ratio and production of the system) it
was found that the module tilt was 20◦ . The modules in the system are directed
towards geographical south, and there is a pitch distance of 2 meters between
the module rows. The resulting three final alternatives were two polycrystalline
alternatives and one monocrystalline alternative. The polycrystalline alterna-
tives used the same REC modules and different inverters, one from Eltek and the
other from SMA. The monocrystalline alternative was simulated with SunPower
v
modules and SMA inverters. The installations have a simulated energy produc-
tion of 22.4, 22.9 and 31.0 MWh/year, which would cover the average energy
consumption of a household in Norway (20.4 MWh/year). However, the instal-
lation will only contribute to reduce the energy consumption in the six storey
commercial building by approximately 1 % per year. Comparing the simulated
productions and the consumption in 2011, it is found that the installation will
not result in a surplus of energy which could have been injected into the grid.
The installation will, therefore, not change the buildings customer status to a
surplus customer (plusskunde). With the simplified economical evaluation it is
found that the energy from the PV installation will cost more than the energy
agreement of today and it is triple the yearly average market price of electricity
the last three years. The polycrystalline alternative with SMA inverters was the
least expensive alternative of the three and the polycrystalline alternative with
highest production. The monocrystalline alternative gave best simulated produc-
tion and performance ratio of the three alternatives, but was the most expensive
alternative.
vi
Sammendrag
Denne oppgavens hovedfokus er å designe et solcelleanlegg tilknyttet nettet på et

spesifikt tak i Oslo. Motivasjonen i denne studien har vært den kraftige veksten
av solindustrien i verden, mens det har vært lite og nesten ingen større utbygging
av solcelleanlegg i Norge. Hensikten har vært å undersøke hvor mye fornybar
solenergi som kan produseres ut fra det designede anlegget med hovedfokus på
det elektriske aspektet.
Solens varierende posisjon i løpet av hver time, hver dag i løpet av et år, skyggene
på modulene, den elektriske effekten av skygge på modulene og bypassdioder
er blant flere faktorer som påvirker produksjonen i et solcelleanlegg. Simuler-
ingsprogrammet PVsyst ble brukt som en støtte grunnet kompleksiteten i et
solcelleanlegg og for å beregne produsert energi. I simuleringsprogrammet ble
det laget en 3D-representasjon av bygningen og skyggeelementene for at skygge-
beregninger skulle utføres i programmet. Meteorologiske data til programmet
ble samlet fra de lokale værstasjonene på Blindern, Lier og Ås, samt databaser
som benytter interpolering blant værstasjoner i sin database og databaser som
baserer seg på satellittbilder. Dataen ble deretter analysert for å lage et nytt me-
teorologisk datasett til simulerings programmet. Dimensjoneringskriteriet for å
finne avstanden mellom modulradene var at det ikke skulle oppstå skygge fra an-
dre modulrader ved vårjevndøgn (21. mars). Helningen til modulene ble justert
fra den optimale 40◦ helningen, for å redusere skyggetapene og oppnå en høyere
effektivitet i anlegget. Den simulerte produksjonen fra de tre beste alternativene
ble sammenlignet med energiforbruket til bygget anlegget er tenkt plassert på.
Til slutt ble det gjort en forenklet økonomisk vurdering av de tre beste alterna-
tivene.
Som et resultat av begrenset areal, skyggeelementene på taket, samt dimen-
sjoneringskriteriet (maksimering av effektiviteten og produksjonen til anlegget)
ble helningsvinkelen på modulene 20◦ . Modulene i systemet er vendt mot ge-
ografisk sør og avstanden mellom radene er modellert til 2 meter. To av de tre
beste alternativene var polykrystallinske REC moduler med ulike vekselrettere.
Det ene alternativet var simulert med vekselrettere fra Eltek mens det andre var
simulert med vekselrettere fra SMA. Det tredje alternativet var simulert med
monokrystallinske SunPower moduler og en type vekselretter fra SMA. De tre
vii
alternativene resulterte i en årlig energiproduksjon på henholdsvis 22,4, 22,9 og
31,0 MWh. Denne produksjonen ville ha dekt forbruket til en gjennomsnittlig
bolig i Norge (som er på 20,4 MWh/år). Sammenlignet med energiforbruket i
bygget solcelleanlegget er tenkt å stå på, dekker det bare ca. 1 % av forbruket
til bygget i 2011. Dette er på grunn av størrelsen til bygget, som er syv etasjer.
Den simulerte produksjonen sammenlignet med forbruket i 2011 i bygget viste
at det ikke ville forekomme tilfeller hvor produksjonen oversteg forbruket med
tanke på effekt. Bygget vil derfor ikke endre kundestatus til plusskunde med et
slikt solcelleanlegg plassert på taket. Ut i fra de forenklede økonomiske bereg-
ningene er det vist at energien fra solcelleanlegget er dyrere enn den nåværende
energiavtalen og dyrere enn den gjennomsnittlige årlige markedsprisen på energi
levert til Oslo over de tre siste årene. Av de tre alternativene er alternativet med
polycrystallinske moduler og vekselrettere fra SMA billigst og det som produserer
mest av de to polykrystallinske alternativene. Det monokrystallinske alternativet
er dyrest, men er alternativet med høyest effektivitet og produksjon.
viii
Preface
This thesis is the result of the work done this final semester and marks the end of
a five year masters programme (MSc) in Energy and Environmental Engineering
at the Norwegian University of Science and Technology. I have learned much this
past semester and I would like to thank those who have helped me during this
semester with my thesis.
First of all, I would like to thank my supervisor Dr.Ing Lars Norum. He has asked
important questions which I did not know that I also had. He has discussed with
me and helped me progress. Secondly, I would like to thank my co-supervisor
Dr.Ing Bjørn Thorud at Multiconsult who suggested the problem which was the
start of this thesis. His enthusiasm has been an inspiration during this semester.
I would also like to thank him and Multiconsult who gave me a flying start with
this thesis with introductory presentations to the PV system and a trip to Glava
Energy Center to see a PV plant in Sweden. I would also express my gratitude
to PhD student Georgi Yordanow at the University of Agder, who has informed
me of his work, answered my questions and helped analysing strange irradiation
measurements from Lier and Ås’ weather stations. In addition I would like to
thank Linn Dalaker at Multiconsult and Elvedin Grudic at Hafslund, who have
provided me with information concerning the building and answered my questions
regarding connection to the grid.
In addition, I want to thank my fellow students, friends and family for the sup-
port they have given me through these five years. Thanks to my mother for all
the love and support she has always provided. I would especially like to thank
those who have helped me with the finishing touches in this thesis, Camilla Sun
Kilnes, Morten Smedsrud, and Adam Gray. Finally I would like to thank Ingve
Løkken for his support, endurance and love during these final semesters of my
masters degree.
Siv Helene Nordahl

Trondheim, 20.06.2012
ix
x
Contents
Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Sammendrag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives and Limitations . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Photovoltaic systems 5
2.1 PV Systems in General . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 The Electrical Characteristics of a PV Cell, PV Module, PV String
and a PV Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 The Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3 PVsyst 11
3.1 Project: Geographical Location and
Meteorology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.1.1 Solar Radiation and Meteorology . . . . . . . . . . . . . . . 12
3.1.2 PVsysts Geographical and Meteorological
Considerations . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 Optimum Tilt Angle . . . . . . . . . . . . . . . . . . . . . . 29
3.2.2 PVsyst and Orientation . . . . . . . . . . . . . . . . . . . . 30
3.3 Shade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3.1 Electrical Effect of Shading . . . . . . . . . . . . . . . . . . 31
3.3.2 Site Assessment . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3.3 PVsyst: Horizon and Near Shadings . . . . . . . . . . . . . 40
3.4 System - Matching Array and Inverter . . . . . . . . . . . . . . . . 44
3.4.1 Choosing Module . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4.2 Choosing Inverter . . . . . . . . . . . . . . . . . . . . . . . 45
3.4.3 Matching Module and Inverter . . . . . . . . . . . . . . . . 46
3.4.4 Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4.5 Cable sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
xi
3.5 Module Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.6 Connection to the Grid and Metering . . . . . . . . . . . . . . . . 57
3.6.1 Connection to the Grid and Customer Status . . . . . . . . 57
3.6.2 Metering . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.7 Economics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4 Simulations 63
4.1 Fraction of Electrical Effect . . . . . . . . . . . . . . . . . . . . . . 66
4.1.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.2 Module Tilt Angles . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.2.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.3 Meteorological Data Set . . . . . . . . . . . . . . . . . . . . . . . . 69
4.3.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.4 Inverters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.4.1 Homogeneous System . . . . . . . . . . . . . . . . . . . . . 71
4.4.2 Heterogeneous System . . . . . . . . . . . . . . . . . . . . . 72
4.4.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.5 Shade Analysis and Module Layout . . . . . . . . . . . . . . . . . . 77
4.5.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.6 Cable Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.6.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5 Customer Status and Economics 85

5.1 Production and Consumption . . . . . . . . . . . . . . . . . . . . . 85
5.1.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.2 Economics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.2.1 Simple Payback . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.2.2 Life Cycle Costing . . . . . . . . . . . . . . . . . . . . . . . 92
5.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6 Further Work 95
7 Conclusion 97
References 99
Appendix A Data Collection Process 105
Appendix B Meteorological Data 107
Appendix C Shade 115
Appendix D Simulation Settings 123
Appendix E Cable sizing 125
Appendix F Technical Data 135
xii
Appendix G Simulation Results 149
Appendix H Production and Demand 153
Appendix I Economics 157
xiii
xiv
List of Figures
2.1.1 Schematic of a stand-alone and a grid-connected PV system . . . 5

2.2.1 PV cell, PV module, PV string, PV array . . . . . . . . . . . . . 6
2.2.2 I-V characteristics of a diode . . . . . . . . . . . . . . . . . . . . 7
2.2.3 The single diode model . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.4 I-V and power characteristics of a PV cell . . . . . . . . . . . . . 8
2.2.5 Interconnection of PV modules in series and parallel . . . . . . . 8
2.3.1 A PV module characteristic curve and the operating and MMP
range of an inverter . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1.1 Solar radiation in the atmosphere . . . . . . . . . . . . . . . . . . 13

3.1.2 AM 1 solar spectrum after atmospheric absorption effects . . . . 14
3.1.3 Air mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.4 Solar irradiation map of Norway in January and July . . . . . . . 15
3.1.5 Yearly global irradiation on inclined modules, Europe . . . . . . 16
3.1.6 Altitude and azimuth angle . . . . . . . . . . . . . . . . . . . . . 17
3.1.7 Cell/module I-V curves with varying irradiance and constant
temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.1.8 Cell/module I-V curves with different temperatures and constant
irradiance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.1.9 Cell/module power with different temperature and constant ir-
radiance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.1.10 Solar paths at Oslo, (Lat 59.5◦ N, long. 10.4◦ E, alt. 5m) . . . . . 20
3.1.11 Yearly global horizontal irradiation (kWh/m2 /year) comparison
between Meteonorm, NASA, PVGIS, RETScreen and two Bio-
forsk weather stations in Lier and Ås. . . . . . . . . . . . . . . . 26
3.1.12 Global horizontal irradition, Meteonorm, Lier and Ås . . . . . . . 26
3.1.13 Comparison of temperature MET and Meteonorm default mete-
orological set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2.1 Tilt angle β . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3.1 Connection of bypass diodes . . . . . . . . . . . . . . . . . . . . . 32
3.3.2 Module I-V curves with and without bypass diodes . . . . . . . . 32
3.3.3 Shading situation where two to eight modules in a string are shaded 33
3.3.4 Shading situation of an array where two of the strings are shaded
from two to eight modules . . . . . . . . . . . . . . . . . . . . . . 34
xv
3.3.5 Shading situation of an array where two modules on one to four
strings are shaded . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.6 Location of the commercial building . . . . . . . . . . . . . . . . 36
3.3.7 Horizon north west . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.8 Horizon south west . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3.9 Two different mounting systems one in landscape, the other in
portrait . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3.10 Shading of modules in rows . . . . . . . . . . . . . . . . . . . . . 39
3.3.11 Horizon line drawing in PVsyst . . . . . . . . . . . . . . . . . . . 41
3.3.12 The 3D representation of the building with a PV installation on
the roof in PVsyst . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3.13 Modules drawn as rectangles . . . . . . . . . . . . . . . . . . . . . 43
3.3.14 Shading factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.1 PV module characteristic curves and operating range of an inverter 48
3.4.2 Array or inverter sizing in PVsyst . . . . . . . . . . . . . . . . . . 50
3.4.3 Wiring connections for parallel strings . . . . . . . . . . . . . . . 53
3.4.4 Wiring connections for groups of parallel strings . . . . . . . . . . 53
3.4.5 Incidence angle modifier . . . . . . . . . . . . . . . . . . . . . . . 55
3.6.1 Net and gross metering systems in the same diagram . . . . . . . 59
4.0.1 Main results of the base case system . . . . . . . . . . . . . . . . 64

4.0.2 Normalized energy production and the performance ratio of each
month, base case system . . . . . . . . . . . . . . . . . . . . . . . 64
4.0.3 Loss diagram for the base case, from PVsyst . . . . . . . . . . . . 65
4.2.1 Overview of 2 meter and 2.5 meter pitch distance . . . . . . . . . 68
4.3.1 The normalized productions for the two meteorological data sets 70
4.4.1 The REC module I-V curve with different irradiation, from PVsyst 75
4.4.2 The ITS module I-V curve with different irradiation, from PVsyst 76
4.4.3 Overview of the roof divided into three fields . . . . . . . . . . . 76
4.5.1 Module layout of the REC alternative with Eltek inverters, 100
modules in total . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.5.2 Module layout of the REC alternative with SMA inverters, 102
modules in total . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.5.3 Module layout of the SunPower alternative with SMA inverters,
104 modules in total . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.1.1 Simulated energy production per month of the three alternatives 86
xvi
List of Tables
3.1.1 Overview of usual albedo values given in PVsyst . . . . . . . . . . 21

3.1.2 Module temperatures during summer and winter . . . . . . . . . . 23
3.1.3 The site-dependent design parameters which will be used in the
simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.4 A summary of the meteorological databases with importing tool
in PVsyst . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.1.5 The second meteorological data set which will be used in the sim-
ulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3.1 Spring equinox pitch distances in meters, NASA . . . . . . . . . . 37
3.3.2 Different mounting types and some specifications . . . . . . . . . . 40
3.7.1 The tariff in 2012 for low voltage connections to Hafslunds grid . . 61
4.1.1 The main results when varying the fraction of electrical effect . . . 66
4.2.1 Losses in % with variable module tilt angles . . . . . . . . . . . . . 67
4.2.2 Main results with variable module tilt angles . . . . . . . . . . . . 67
4.2.3 Increasing pitch distance between the module rows to 2.5 meters . 68
4.3.1 Main results with the default and the new meteorological data set 70
4.3.2 The losses in the system with the default and the new meteoro-
logical data set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.4.1 The main results from simulation with inverters from Eltek, SMA
and Danfoss with the polycrystalline REC module: REC250PE . . 72
4.4.2 The main results from simulation with inverters from Eltek, SMA
and Danfoss with the polycrystalline ITS module: EcoPlus PolyUp,
250W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.4.3 The main results from simulations with inverters from Eltek, SMA
and Danfoss with the monocrystalline SunPower module: SPR-
327NE-WHT-D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.4.4 Main results and main differences from base case simulations with
the polycrystalline REC and ITS modules . . . . . . . . . . . . . . 74
4.4.5 Main results when simulating a heterogeneous system with differ-
ent Eltek inverters with the polycrystalline REC module . . . . . . 77
4.5.1 The main result for the shade analysis . . . . . . . . . . . . . . . . 77
4.6.1 Minimum CSA and loss, polycrystalline and Eltek alternative . . . 82
4.6.2 Minimum CSA and loss, polycrystalline and SMA alternative . . . 82
4.6.3 Minimum CSA and loss, monocrystalline alternative, 1 . . . . . . . 82
xvii
4.6.4 Minimum CSA and loss, monocrystalline alternative, 2 . . . . . . . 83
5.1.1 The energy production as percentage of the total energy use in the
building, 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.1.2 The energy production as percentage of the energy used for cooling
in the building, 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.1.3 The average and maximum power produced during the simulated
year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.1.4 The minimum and maximum power consumed for each month in
2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.2.1 Investment cost for the three alternatives . . . . . . . . . . . . . . 90
5.2.2 Calculated income for a worst and a best case scenario when con-
sidering the power production . . . . . . . . . . . . . . . . . . . . . 91
5.2.3 The theoretical income if the total amount of energy was sold at
spot price on Nord Pool . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.4 Main results of the present value calculations . . . . . . . . . . . . 91
5.2.5 Payback time in years when using the simple payback method . . . 92
5.2.6 The cost per kWh of the three alternatives . . . . . . . . . . . . . 92
xviii
Acronyms and
Abbreviations
AC Alternating current
AM Air mass
BIPV Building Integrated Photovoltaic
BoS Balance of System
CCC Current Carrying Capacity
CSA Cross Sectional Area
DC Direct current
EF Electrical Effect
GW Gigawatt
I-V Current and Voltage
IAM Incidence Angle Modifier
ITS Innotech Solar
MET The Norwegian Institute of Meteorology
MPP Maximum Power Point
MPPT Maximum Power Point Tracker
NASA National Aeronautics and Space Administration
NOCT Nominal Operating Cell Temperature
NVE Norwegian Water Resources and Energy Directorate
PV Photovoltaic
PR Performance Ratio
xix
REC Renewable Energy Corporation
SSE Surface meteorology and Solar Energy
STC Standard Test Conditions
TF Transposition Factor
TSO Transmission System Operator
USNO U.S. Naval Observatory
Vp Voltage peak
Wp Watt peak
xx
Nomenclature
α Solar altitude (◦ )
βopt Optimum tilt angle (◦ )
∆T Temperature difference between the STC temperature (25◦ C) and the
highest/lowest cell temperature
δ Declination (◦ )
ηm Module efficiency (-)
γISC Short circuit current temperature coefficient of the module (A/◦ C)
γV O C Open circuit voltage temperature coefficient (V/◦ C)
γVM P P Voltage temperature coefficient (V/◦ C)
ω Hour angle (◦ )
φ Latitude (◦ )
ψ Solar azimuth (◦ )
ρ Resistivity (Ω/m/mm2 )
θz Zenith angle (◦ )
A Absorption coefficient (-)
C Initial investment cost
G Irradiation (W/m2 )
Gef f Effective irradiance (W/m2 )
kt,a Annual average clearness index (-)
L Length (m)
LDC cable Route length of DC cable (m)
S Annual cost savings
xxi
T The payback period (years)
Ta Ambient temperature (◦ C)
Tm Module temperature (◦ C)
Uc Constant component of the thermal loss factor (W/m2 ·k)
Uv Factor proportional of the wind velocity (W· s/m3 ·k)
v Wind velocity (m/s)
w latitude minus optimum tilt angle (◦ )
Yf Final PV system yield (kWh/kWp/day)
Yr Reference yield (kWh/m2 /day)
xxii
Chapter 1
Introduction
1.1 Motivation
The energy demand in the world is increasing. From 1976 to 2009 there has been
a 40 % increase in the final energy consumption in Norway [1]. Generally, there is
found a link between increasing economic growth and an increased consumption
of energy. In the analysis and reference case of the International Energy Outlook
2011, an energy consumption in countries outside the Organization for Economic
Cooperation and Development (non-OECD) increased by 85 %, compared to an
increase of 18 % for the OECD economies [2]. The reference case considered does
not incorporate prospective legislation or policies that might affect energy mar-
kets. Two major global energy problems that have become more accepted during
the last 12 years are that the oil reserves is a limited resource and that climate
change needs urgent international action in order to mitigate its effects resulting
from the worlds consumption of fossil fuels. The Intergovernmental Panel on
Climate Change (IPCC) provides with regular interval assessment reports of the
state of knowledge on climate change, the last report being produced in 2007.
Renewable power production is playing, and will continue to play, a large role in
solving these issues.
Over the past seven years, the Photovoltaic (PV) sector has transformed from a
small industry centered in Germany to become an industry with global reach. PV
prices have fallen due to growth on account of government subsidies, capacity ad-
ditions from both existing and new entrants and continual innovation. The global
installed capacity exceeded 65 gigawatt (GW) in 2011 [3]. A new world record
was set Friday, May 25th, this year (2012) when the German PV installations
produced an accumulated power of 22 GW. During that Friday and Saturday the
PV installations covered respectively one third and one half of Germany’s power
demand [4]. One other target from the German government which most likely
1
2 CHAPTER 1. INTRODUCTION
includes the PV industry is that all its new buildings should be climate neutral
from 2012, which means they will run entirely on renewable energy [5].
A common assumption is that PV systems are not suited for latitudes as far
north as Norway, since the amount of sunny days are not as many as for countries
located further south. However, small scale PV installations have been used for
some time in Norway as stand-alone PV systems installed at cottages around the
country. Currently there is planned a commercial building in Trondheim with a
PV façade, a sea water heat pump and a heat exchanger. This building should
produce more than it consumes during its life time [6]. Construction start is
planned next year, 2013. Becoming a producer of energy can be a comprehensive
process. In order to simplify the process of becoming a surplus customer there
has been given a dispensation from entering into a balancing agreement with
the national Transmission System Operator (TSO) from the Norwegian Water
Resources and Energy Directorate (NVE) in 2010. The surplus customers are
customers which do not normally exceed their own consumption during a year,
but has some surplus energy which could be sold to the grid. In counties Oslo,
Akershus and Østfold there is at this time only one surplus customer.
1.2 Objectives and Limitations
The objective of this thesis is to design a grid-connected PV system for a specific

roof on a commercial building in Oslo, Norway. The system should be optimized
with respect to the performance ratio of the system and the production. When
a few final system alternatives (with different components) are finalized, a com-
parison of the production of the PV systems should be done with respect to the
energy demand in the building where the system is to be placed. An investigation
should be performed concerning the regulations and compliances which should
be met when connecting to the grid.
A PV system as a whole is complicated as there are many variables and several

different disciplines that have to work together in order to achieve a good PV
system. However, this thesis has been limited to consider the electrical behaviour
of a PV system and the elements related to the electrical behaviour of the system.
PVsyst, the simulation program that was used in this thesis, was used as a
support tool. It is considered a comprehensive PV simulation program. However,
all simulation programs have their limitations, whether considering the accuracy
of the methods used or whether or not elements are considered at all. The choice
of module and inverter manufacturers was limited to three each. Furthermore,
the number of possible inverters and modules were limited to one series each.
1.3. OUTLINE OF THE THESIS 3
1.3 Outline of the Thesis

Chapter 2 gives a basic introduction to a grid-connected PV system and the
electrical characteristics of the solar cell and the inverter. This chapter could
be disregarded by readers with this knowledge. Chapter 3 is first structured in
chronological order of the steps that are necessary in order to perform a simula-
tion of a grid-connected PV system in the simulation software PVsyst:
1. Project
2. Orientation
3. Shade
4. System
5. Module layout
In order to perform the simulation, all of these steps has to have defined param-
eters. In order to define the parameters some background information will be
provided. The final two sections of Chapter 3 are sections regarding the connec-
tion to the grid and economic models used.
Chapter 4 presents the results from simulations where some of the parameters
are varied. The simulations are compared with a base case of a PV system in
order to evaluate the change in the results when varying parameters. Chapter
4 is, therefore, documenting the process in reaching the final three alternative
PV systems which then will be examined further in Chapter 5. The production
of the final three alternatives is more thoroughly examined in Chapter 5 and is
compared with the energy consumption within the building where the PV system
is placed. Simple economical calculations are also performed in chapter 5. Finally
there are some recommendations for further work in Chapter 6, and conclusions
are drawn in Chapter 7.
4 CHAPTER 1. INTRODUCTION
Chapter 2
Photovoltaic systems
2.1 PV Systems in General

PV systems can in general be divided into two different systems, stand-alone sys-
tems and grid-connected systems (see Figure 2.1.1). Stand-alone systems would
have PV modules, a energy storage medium and a load. The storage and a
backup generator is drawn into Figure 2.1.1 with dotted lines, since this is op-
tional. Stand-alone systems can for example be found in cottages in Norway. In
this study a grid-connected PV system is to be dimensioned. As the dimension-
ing methods for these two systems differ, only the grid-connected PV system will
be discussed from this point on.
Figure 2.1.1: Schematic of a stand-alone PV system to the left and a grid-

connected PV system to the right
A grid-connected PV system comprises of PV modules and Balance of Sys-

tem (BoS) components. The BoS components include the inverters, mounting
systems, array cabling to a possible junction box, possible cables from the junc-
tion box to the inverter, protection and disconnection switches, lightning pro-
5
6 CHAPTER 2. PHOTOVOLTAIC SYSTEMS
tection, the Alternating current (AC) cabling from the inverter, metering and
system monitoring [7, p.131]. The main components of a grid-connected PV sys-
tem which will be explained in more detail are the modules and the inverters.
The modules produce electricity from solar irradiation and the inverters convert
the Direct current (DC) current produced by the modules into AC current which
can be injected into the electricity grid.
2.2 The Electrical Characteristics of a PV Cell,

PV Module, PV String and a PV Array
Figure 2.2.1 illustrates the configuration in the generating part of the PV system,
from the cell to an array. Each cell produces energy, and each module consists of
several cells in series. It is a 36 cell module illustrated in the figure. If each cell in
the module produces 0.5 V peak (Vp), the module would be rated approximately
with 18 Vp, 12 V nominal [7, p.50] Modules coupled in series are called strings,
while strings coupled in parallel are called an array.
Figure 2.2.1: PV cell, PV module, PV string, PV array
Returning to the cell, its electrical properties can be described as a large diode
(see Figure 2.2.2). When the current flows through the diode as shown in the
figure, the characteristic curve in the first quadrant applies. At a particular
voltage the currents starts to flow, in the figure at a threshold voltage of 0.7 V.
If the current flows in the opposite direction (the reverse direction) the current
flow is prevented until the breakdown voltage of the diode is exceeded (in this
case 150 V). Reaching the breakdown voltage could destroy the diode.
An illuminated solar cell can be explained with a diode in parallel with a current
source. As the cell is illuminated, the current source will produce a photoelectric
current (Iph ). The diode characteristic curve is then shifted by the magnitude
of the photoelectric current downwards in quadrant three and four, seen as the
dotted line in Figure 2.2.2 [8]. The standard model used to represent a PV
2.2. THE ELECTRICAL CHARACTERISTICS OF A PV CELL, PV
MODULE, PV STRING AND A PV ARRAY 7
Figure 2.2.2: I-V characteristics of a diode. F = forward, R = reversed
cell is an extended equivalent circuit called the single-diode model (illustrated

in Figure 2.2.3). A parallel resistance (RP ) is added to represent the voltage
drop which occurs as the charge carriers migrate from the semiconductor to the
electrical contacts. In addition a series (RS ) resistance is added in order to
represent leakage currents.
Figure 2.2.3: The single diode model
Figure 2.2.4 illustrates the current and voltage (I-V) characteristics and the power
characteristics of a PV cell in the same figure. The Maximum Power Point (MPP)
marked in this figure illustrates the operating point of the cell, with regards to
current and voltage, in order to maximize production.
Figure 2.2.5 illustrates the I-V characteristic of an array when it consists of both
series and parallel connected modules. When the modules are connected in series
the voltage increases along the abscissa axis. When the modules are connected
Figure 2.2.4: I-V and power characteristics of a PV cell
Figure 2.2.5: Interconnection of PV modules in series and parallel

2.3. THE INVERTER 9
in parallel the current increases as seen in the figure.

There are several cell technologies available on the market. Mocrystalline sil-
icon, polycrystalline silicon, ribbon silicon, crystalline thin-film silicon, amor-
phous silicon, micromorphous silicon, hybrid HIT solar cell, cadmium telluride,
high-efficiency III-V multijunction cells and dye-sensitised cells are some of the
technologies that can be named [9]. The mono- and polycrystalline silicon cells
have a commercially high efficiency which is of importance when there is limited
amount of space for a PV field. The single-diode model of a cell si well suited
for the silicon crystalline cells, however, other adaptions are required when re-
producing the thin film technology [10]. Based on this, it is these two cell types
which will be considered in this study.
2.3 The Inverter

As mentioned, the inverter transforms DC current from the PV installation to
AC current. The transformed AC current produced by the inverters matches
the voltage and phase of the electricity grid. In order to produce AC current,
semiconductor switches are used to produce pulses of voltage where the width of
these pulses are modified in order to achieve an accurate approximation of a sine
wave [7, p.103]. These are so called switch-mode inverters, and there are several
inverter switching schemes which can be used.
Figure 2.3.1: A PV module characteristic curve and the operating and MMP
range of an inverter
Figure 2.3.1 illustrates a PV module I-V characteristic with the operating and
MPP range of an inverter. Within the MPP range of the inverter, it has a
tracker, a Maximum Power Point Tracker (MPPT). The purpose of the MPPT
is to locate the MPP, marked with a circle in the figure, in order to obtain as
much power as possible from the system. There are several MPPT schemes which
each have their advantages and limitations with regards to efficiency [11]. The
subject of MPPT schemes will, not be discussed in this study. The PV array
voltage has to match the operating range of the inverter in order for the inverter
to function. Inverters can or can not be equipped with a transformer. Invert-
ers without transformers offer improved efficiency, while inverters with provide
galvanic (electric) isolation between the DC input side and the AC output side.
Without the galvanic isolation between the DC and AC side, DC currents might
be injected into the grid. The improved efficiency without a transformer is due
to the losses which occur in a transformer. Transformer-less inverters are often
lighter and could cost less than inverters with transformers, due to a smaller
amount of metal [7].
Chapter 3
PVsyst
There are several PV system simulation programs. Some are:

• SOLDIM
• PVS
• PV*SOL
• DASTPVPS (Design and Simulation Tool for PV Pumping Systems)
• Greenius
• PV Design Pro (Solar Studio Suite)
• PVsyst
These are known as so called time-step simulation programs. All of the mentioned
simulation programs except DASTPVPS apply for grid-connected systems. SOL-
DIM, PVS, PV*SOL, Greenius, PV Design Pro and PVsyst all apply for stand-
alone systems, hybrid systems, PV pumping systems and some apply for other
technologies such as wind, fuel cells, solar thermal plants etc. All provide an
economical overview of viability. PVS, PV*SOL, PVsyst, PV Design Pro and
SOLDIM provide shading calculations down to intervals of one hour. In addition
PVS, PVsyst, PV Design Pro and SOLDIM provide a solar radiation generator
[8, p.199].
PVsyst is considered one of the most comprehensive programs of the ones listed,
and also one of the most complicated to use. PVsyst has a higher accuracy of
shading calculations than PVS and PV*SOL, and is the only program which
provides a 3D representation of the PV field. Greenius provides an extensive
economical calculation, while the other programs are fairly similar with varying
degrees of functionality. Tests of various programs in Europe showed that shad-
ing came in top for causing reduced yields and was responsible for considerable
11
12 CHAPTER 3. PVSYST
reductions of as much as 30 % and more. The yield reductions generally turn out
to be greater than one would suspect based on the shaded surface area. Thus,
the shading analysis is known to be a sensitive point in PV system simulations
[8, p.193]. Examples of other PV simulation systems are PVSIM, PVFORM,
PVNet and SimPhoSys [12].
The simulation software PVsyst was used in this study in order to calculate
the production of the PV installation. PVsyst is developed at the University of
Geneva by André Mermoud [13]. PVsyst has the option of simulating a prelim-
inary design or a project design. In addition PVsyst has a tool option, where
background data management and didactic tools are located. The preliminary
design option of PVsyst has not been used in this study, since this is a rougher
simulation which is assumed to not contribute with significant information com-
pared to the project design simulation option. The five first sections of this
chapter (Section 3.1 to 3.5) follows the structure of a grid-connected project sim-
ulation in PVsyst. Section 3.6 describes the process of obtaining a grid-connected
system in Norway. The last section, section 3.7 describes the economic methods
used in this study.
3.1 Project: Geographical Location and

Meteorology
The geographical location of the project and the local weather conditions influ-
ence the optimal tilt of the PV modules and is, therefore, of great importance.
The building in this study is located in Skøyen, Oslo, Norway at a latitude of
approximately 60◦ N.
3.1.1 Solar Radiation and Meteorology
The amount of effect produced by a PV module relies on how much radiation it

receives at its precise location. The energy produced by the sun at a temperature
close to 5800 K (5526.85◦ C) is emitted primarily as radiation. When the radia-
tion enters the atmosphere it can collide with clouds and air molecules and the
radiation can then scatter or be absorbed (illustrated in Figure 3.1.1). The beam
or direct radiation is the radiation which is not reflected or absorbed and reaches
the surface of e.g. a PV module in a direct line from the sun. The radiation after
a scattering can either be re-emitted into the atmosphere or reach the surface
of the module, this is called diffuse radiation. Albedo radiation is the radiation
reaching the module surface after being reflected by the ground. Global radiation
consists of all three components: beam, diffuse and albedo radiation [14].
On account of the absorption and scattering, the amount of solar radiation that
reaches the Earth’s surface decreases. The radiation passes through gases when
3.1. PROJECT: GEOGRAPHICAL LOCATION AND
METEOROLOGY 13
Figure 3.1.1: Solar radiation in the atmosphere
entering the atmosphere, as a result the radiation at Earth’s surface has a dif-
ferent spectral composition than the radiation which has not passed through the
atmosphere. Air mass (AM) characterises the relative length of the direct beam
path through the atmosphere, and is defined as:
1
AM = (3.1.1)
cos θz
Where θz is the zenith angle, between the sun and the line to a point directly
overhead, see Figure 3.1.3.
The PV modules are rated at Standard Test Conditions (STC), which are:
• AM = 1.5
• Solar radiation = 1000 W/m2
Figure 3.1.2: AM 1 solar spectrum after atmospheric absorption effects [15]
Figure 3.1.3: Air mass
• Cell temperature = 25◦ C
The characterisation of the PV module is further done by measuring the Nominal

Operating Cell Temperature (NOCT), defined under the following conditions at
open circuit [7]:
• AM = 1.5
METEOROLOGY 15
• Solar radiation = 800 W/m2

• Ambient temperature = 20◦ C
• Wind speed = > 1 m/s
An AM of 1.5 is a typical solar spectrum on the Earth’s surface on a clear day.
Although the global irradiance can be as high as 1000 W/m2 and even somewhat
higher, the available irradiance is usually considerably less because of the rotation
of the Earth and adverse weather conditions [14]. There has, however, been
measured over-irradiance or cloud enhancement events even in southern Norway
peaking above 1400 W/m2 for several minutes [16]. Naturally much stronger
over-irradiances may be expected for the lower latitudes near the Equator, with
magnitudes exceeding 1800 W/m2 . However, the contribution of over-irradiance
events to the annual irradiation is very small. Figure 3.1.4 illustrates the solar
irradiation in Norway in January and July while Figure 3.1.5 shows the yearly
global irradiation of optimally inclined PV modules in Europe.
Figure 3.1.4: Solar irradiation map of Norway on a horizontal in January and

July, measured in Wh/m2 day [17]
As seen in these two figures the solar irradiation increases when the latitude de-
creases. The mean annual irradiance is highest near the latitudes of the tropics
of Cancer (23.5◦ N) and Capricorn (23.5◦ S). The mean annual irradiance is, how-
ever, lower in equatorial regions as a result of the cloud cover. At the tropics
of Cancer and Capricorn, approximately at June 21st and December 21st, the
sun is located with a zenith angle of 0◦ , which would be directly situated over a
Figure 3.1.5: Yearly global irradiation on optimally inclined south oriented PV

modules, Europe [18]
possible PV module. It is known as solstice when the sun is over either of the
tropics. The reason why this does not occur at the equator is due to the axial tilt
of the Earth. Approximately at March 21st and September 23rd the sun crosses
the equator, at the equinoxes (vår- og høstjevndøgn). The equinoxes mark the
day when the day and night is equally long.
Position of the sun
The location of the sun is defined by two angles:

• Altitude (α)
• Azimuth (ψ)
METEOROLOGY 17
Figure 3.1.6: Altitude (blue angle) and azimuth angle (green angle)
As seen in Figure 3.1.6, the altitude of the sun is the angle between the sun and
the ground. This angle is always between 0◦ and 90◦ . The sun rises in the east
and sets in the west. Azimuth is the angle between north and the point where
the sun is positioned. Altitude and azimuth can be determined from [15, p.30]
the following equations:
sin α = sin δ sin φ + cos δ cos φ cos ω (3.1.2)
sin α sin φ − sin δ

cos ψ = (3.1.3)
cos α cos φ
where δ is the declination and ω is the hour angle. The declination is the angle
of deviation of the sun from directly above the equator. The hour angle is the
difference between noon and the desired time of day in terms of a 360◦ rotation
in 24 hours. The equation describing the declination is often given as an ap-
proximation since a year is not exactly 365 days. The hour angle can also be
described using an equation. However, there are different equations describing
the hour angle depending on how many variables that are taken into account.
For example, one could take only the hour of the day into account [15, p.29] or
take into account the local official time, the local longitude, a reference longitude
and time difference with respect to time zones [9, p.989].
National Aeronautics and Space Administration (NASA) and the astronomical
applications department under the U.S. Naval Observatory (USNO) both pro-
vide information concerning the azimuth and altitude angles of a given location
described with coordinates. The USNO site takes as an additional input the time
zone of the specific site. NASA gives monthly averaged hourly altitude angles
and azimuth angles. The azimuth and altitude numbers provided by USNO and
NASA are, therefore, not exactly the same.
Temperature and irradiance influence on the PV cell
The nominal operation of PV systems is at STC conditions, however, the PV

systems rarely operate in nominal conditions. The modules usually operate in
partial load due to its high dependence on temperature and irradiance.
Figure 3.1.7: Cell/module I-V curves with varying irradiance and constant tem-
perature
As explained in Section 2.2 the cell can be illustrated with an diode, and the
photoelectric current produced by the cell is a result of illumination on the cell.
The short circuit current of the cell is roughly directly proportional to the ir-
radiance as seen in Figure 3.1.7. The figure illustrates how the current in the
cell or the module decreases with decreasing irradiance. As seen it is the current
that is mostly affected, however, the voltage is slightly affected as well. The cell
has a negative temperature coefficient. A temperature coefficient of 2.3 mV/C◦
METEOROLOGY 19
Figure 3.1.8: Cell/module I-V curves with different temperatures and constant
irradiance
Figure 3.1.9: Cell/module power with different temperature and constant irradi-
ance
describes the decrease of the open circuit voltage with increasing temperature.
The short circuit current on the other hand remains nearly constant. The cell
power, therefore, decreases as well. Figure 3.1.8 illustrates how the voltage and
current change under different temperatures under constant irradiance. As seen
the voltage is mostly affected, however, there are small changes in the current
as well. Figure 3.1.9 illustrates how the cell or module power decreases as the
temperature decreases.
3.1.2 PVsysts Geographical and Meteorological

Considerations
The project location is the first which is defined when performing a project design
simulation in PVsyst. Country and site are chosen, and if there are multiple
meteorological files, one can be chosen. When opening the meteorological file the
latitude and longitude of the location is defined in decimals as well as in degrees
and minutes. The altitude above sea level and time zone are also displayed.
The monthly meteorological values are displayed along with the data source.
The monthly meteorological values which are required are the horizontal global
irradiation and the ambient temperature. Extra data which could be provided
are data for the horizontal diffuse irradiation and the wind velocity. The sun
path for the location is then constructed, as illustrated in Figure 3.1.10.
Figure 3.1.10: Solar paths at Oslo, (Lat 59.5◦ N, long. 10.4◦ E, alt. 5m)
The solar path seen in Figure 3.1.10 is constructed from the default meteorological
file for Norway. The location for this meteorological file is approximately 44 km
METEOROLOGY 21
from the location of the building. The meteorological file is constructed for a
smaller location in Norway called Hurum. For the site it is further possible
to define monthly albedo values. The common value equal for all months is
default 0.2. Within PVsyst an overview of which albedo values represent certain
surroundings is represented in Table 3.1.1. As seen, snow and aluminium has
high albedo values due to a larger part of the light is reflected compared to e.g.
asphalt. As the building is situated in an urban situation it has been assumed an
albedo value equal to the default value 0.2 all year. It was considered to increase
the albedo value from December through February, due to snow. In 2011 there
was recorded a snow coating from January 1st until April 4th, with some snowfall
during a few days in December [19]. However, the default value of 0.2 is chosen
for all months since it is considered giving a more conservative result.
Table 3.1.1: Overview of usual albedo values given in PVsyst
Surroundings Usual albedo values

Urban situation 0.14 - 0.22
Grass 0.15 - 0.25
Fresh grass 0.26
Fresh snow 0.82
Wet snow 0.55 - 0.75
Dry asphalt 0.09 - 0.15
Wet asphalt 0.18
Concrete 0.25 - 0.35
Red tiles 0.33
Aluminium 0.85
New galvanised steel 0.35
Very dirty galvanised steel 0.08
Site-dependent design parameters which are connected to the sizing of the array
design with respect to the inverter input voltages, are defined in PVsyst under
site and meteorological. The lower temperature for VmaxAbs limit has -10◦ C as
default value. This represents the absolute lower cell temperature. However, for
this limit the cell temperature is considered equal to the ambient temperature
[10]. The reason for this is that during morning at first light the sun has not had
time to heat the modules. The temperature of -10◦ C is the default value since it
applies to most European countries. The minimum ambient temperature in Oslo
from 2000 to 2011 was, however, found to be a few decimals lower than -20◦ C
(for details see Appendix B). This value was, therefore, altered from -10◦ C to
-20◦ C. The lower temperature for VmaxAbs limit parameter is used when de-
termining the maximum possible voltage of the array. The absolute maximum
voltage, the open circuit voltage at the lowest temperature, has to stay below the
absolute maximum inverter input voltage. Furthermore it should not overcome

the maximum system voltage specified for the PV module. The usual operating
temperature under 1000 W/m has 20◦ C as default value. This parameter is not
used for sizing constraints in PVsyst and is, therefore, not changed. The winter
operating temperature for VmppMax design parameter, describes the minimum
cell temperature during operating conditions in winter. The maximum array op-
erating voltage has to stay below the maximum inverter voltage (Vmax of the
MPPT range of the inverter explained in Section 2.3). This is set with a default
value of 20◦ C. The summer operating temperature for VmppMin design param-
eter describes the maximum cell temperature in operating conditions and is used
to determine the array minimum operating voltage. The minimum array oper-
ating voltage should be above the minimum inverters operating range (Vmin of
MPPT range). The summer operating temperature for VmppMin design parame-
ter has a default value of 60◦ C. As explained in Section 3.1.1 higher temperature
results in lower cell and module voltage.
The module and cell temperature is affected by several factors, such as irradiance
and the mounting system. If there is free air convection on both sides of the
module it could reduce the module temperature, which would not be the case with
Building Integrated Photovoltaic (BIPV). An additional way in order to reduce
the temperature of the modules even further, is by e.g. water cooling [20]. This
will, however, not be taken into consideration in this study. The temperature of
the module is roughly proportional to the incident irradiance and can be described
with the following equation [14, p.88]
Tm = Ta + Ct · Gef f (3.1.4)
where Ct is described as follows
N OCT (◦ C) − 20
Ct = (3.1.5)
800 W/m2
where Tm and Ta is the module and the ambient temperatures respectively, Gef f
is the effective solar irradiance and where N OCT (◦ C) is the module type depen-
dent test temperature at an irradiance of 800 W/m2 .
The NOCT is normally in the range between 42 − 46◦ C [9][14]. Checking eight
module series (four from Innotech Solar, three from REC and one from SunPower)
four had a NOCT of 45◦ C, while the other four had NOCT of 47.9◦ C with an
uncertainty of ± 2◦ C. When determining the module temperature in summer and
winter the highest and lowest ambient temperature is applied (33◦ C and -20◦ C).
As previously mentioned in Section 3.1.1 there occurs over-irradiance events in
the southern parts of Norway which has been measured up till 1400 W/m2 [16].
However, due to the short time periods when this occurs this is considered not
applicable as Gef f . By assuming a Gef f of 1000 W/m2 , disregarding a difference
METEOROLOGY 23
between summer and winter, 42◦ C, 45◦ C and 49.7◦ C (47.9+2) as NOCT
coefficients, results in a range of module temperatures as shown in Table 3.1.2.
Table 3.1.2: Module temperatures during summer and winter
NOCT Summer Winter

◦ ◦
( C) Tm ( C) Tm (◦ C)
42.0 60.500 7.500
45.0 64.250 11.250
49.9 70.375 17.375
As seen in Table 3.1.2 the default value of the summer operating temperature
is in the lowest calculated temperature range with 60◦ C. The default value
for the winter operating temperature of 20◦ C is above the range of calculated
module temperatures. Modules on a roof could heat up to 70◦ C while with
well ventilated systems a maximum temperature of 60◦ C could be assumed
[8, p.142]. The results in Table 3.1.2 is assumed to be consistent with a well
ventilated system due to the fact that the NOCT coefficient is the temperature
attained by the PV modules with free air circulation all around the module. The
summer operating temperature is chosen to be fixed at a temperature of 70◦ C
and the winter operating temperature at 17◦ C in accordance to Table 3.1.2 with
an NOCT of 49.9◦ C. It was chosen as a precaution if a module with a NOCT
of 47.9◦ C is chosen, and if a module with a lower NOCT is chosen the system
would not be operating outside the minimum operation range of the inverter.
Table 3.1.3: The site-dependent design parameters which will be used in the
simulations
Parameters Values
Lower temperature for VmaxAbs limit -20◦ C
Winter operating temperature for VmppMax design 17◦ C
Usual operating temperature under 1000 W/m 50◦ C
Summer operating temperature for VmppMin design 70◦ C
Meteorological data sources
Within PVsyst there are possibilities to define new monthly meteorological values
and redefine the location of the project as well as import both monthly and
hourly meteorological data from a number of other databases, seen in Table
3.1.4. A comparison of the free web based databases was done alongside with
local meteorological data where possible. A custom-made second meteorological
set was assembled, with a focus on monthly irradiance and temperature values,
since these are the ones compulsory in order to run a PVsyst simulation.
Table 3.1.4: A summary of the meteorological databases with importing tool in

PVsyst. Gh = global horizontal irradiation, Dh = diffuse horizontal irradiation,
Ta = ambient temperature, WindVel = wind velocity [21].
Database Region Values Variables Availability

Meteonorm Worldwide Monthly Gh, Ta, Software
WindVel
Meteonorm Worldwide Hourly Gh, Dh, Software
Ta, WindVel
Satellight Europe Hourly Gh, NO Ta Web free
US TMY2 USA Hourly Gh, Dh, Web free
Ta, WindVel
ISM-EMPA Switzerland Hourly Gh, Dh, Included
Ta, WindVel in PVsyst
Helioclim Europe Hourly Gh, NO Ta Web restricted
(SoDa) Africa Hourly
NASA-SSE Worldwide Monthly Gh, Ta Web free
WRDC Worldwide Hourly Gh, NO Ta Web free
Daily
Monthly
PVGIS-ESRA Europe Monthly Gh, Ta, Web free
Light turbidity
Africa
Helioclim -1 Europe Monthly Gh, NO Ta Web restricted
(SoDa) Africa
RETScreen Worldwide Monthly Gh, Ta, WindVel Software, free
SolarGIS Europe Hourly Gh, Dh, Ta Web,
Africa paid access
Asia, Brazil,
West Australia
As seen from Table 3.1.4 there are four databases which do not provide ambi-
ent temperatures: Satellight, Helioclim, Helioclim -1 and WRDC. The ambient
temperature would have to be supplemented in case these databases were to be
utilized. Further it is seen from the table that two of the databases are valid only
for USA and Switzerland. The default meteorological data embedded in PVsyst
METEOROLOGY 25
for Oslo is a monthly data set from Meteonorm. Meteonorm is a database contain-
ing climatological data for solar engineering and could also be used to calculate
solar radiation on arbitrarily oriented surfaces [22]. PVsyst uses the Meteonorm
software in order to import a synthetic hourly meteorological file [10]. A com-
parison between Meteonorm values and values from the sources in Table 3.1.4
has been performed by the PVsyst team in [23]. The comparison was done of
twelve locations in Europe, and there was made no comparison with respect to
real measured values. It was, therefore, not concluded which source that could
be most representative of real weather conditions. What was observed was that
Meteonorm often gave lower values than the average. Meteonorm is, therefore,
considered conservative and would give prudent results for the final yield of the
simulated systems.
Figure 3.1.11 shows a comparison between four different meteorological databases

which were made for Skøyen (the location of the PV installation) alongside two
local weather stations in Lier and Ås. Further information and the tabular values
for the different meteorological databases are found in Appendix B. There are two
values for Meteonorm, one is the default value for Oslo, 44 km south of Skøyen
(marked with a D), while the other is the value extracted from the program with
the Skøyen coordinates. Meteonorm is, therefore, represented with two values.
The weather stations at Lier and Ås are located approximately 27 km south west
and south east of Skøyen. These local weather stations have since 2007 measured
one hour averages of global horizontal irradiance. The irradiation form the local
weather stations in Figure 3.1.11, is the average measurements of each month
from 2007 − 2011. The reliability of the measured data could be questioned since
it is only a five years data series. Each year there are natural fluctuations which
could differ as much as 10 − 15 % to the average over a 20 year period. This data
is considered to be important and is, therefore, included.
As seen in Figure 3.1.11 the default meteorological file used in PVsyst has a
higher annual global horizontal irradiance than the two local weather stations
with approximately 30 kWh/m2 . The NASA numbers, which are the highest,
are based on satellite measurements from 1983 to 1993 and are valid within the
boundaries of 60 and 59◦ N, and between 10 and 11◦ E. The values from Me-
teonorm are based on interpolation between the nearest weather stations between
1981 − 2000. For Skøyen there were three stations available: the first in Karlstad
(169 km away), the second in Borlaenge (273 km away) and the third in Bergen
(301 km away).
Figure 3.1.12 illustrates the two different datasets from Meteonorm which are
the closest in annual irradiation to the two local weather stations. It is seen that
both in May and July the irradiance in the default set is about 13 % and 7 %
higher than the local measurements. However, the other set from Meteonorm is
between 8 and 15 % lower than the local weather sets. Since there is no reason
to exclude any of the three sources of information (four data sets) an average of
them are made. With the local stations being further south than Skøyen it could
Figure 3.1.11: Yearly global horizontal irradiation (kWh/m2 /year) comparison

between Meteonorm, NASA, PVGIS, RETScreen and two Bioforsk weather sta-
tions in Lier and Ås.
Figure 3.1.12: Global horizontal irradition, Meteonorm, Lier and Ås
be assumed that the irradiation at Skøyen would be somewhat lower than the
local weather station measurements in Lier and Ås. This is also the result of the
average of the four different data sets.
METEOROLOGY 27
Figure 3.1.13: Comparison of temperature MET and Meteonorm default meteo-

rological set
Figure 3.1.13 illustrates the difference between the Meteonorm default meteoro-
logical set for Oslo and the normalised temperature values from The Norwegian
Institute of Meteorology (MET). The normalised temperatures from MET are
based on daily normalised temperature values from 1961−1990, measured at Blin-
dern in Oslo. Blindern is approximately 3 km from Skøyen. As seen in Figure
3.1.13 there are some deviations of two to three degrees between the Meteonorm
default values and the values from MET, especially in the months November to
February. Due to the length of time the series of temperature values are mea-
sured, and the short distance between Blindern and Skøyen, the temperature
values from MET are used in the second meteorological data set (shown in Table
3.1.5).
The two last components of the meteorological data set, diffuse irradiance and
wind, were also extracted from the databases available. Since these data are
not necessary in order to perform a simulation, they were not compared with
measurements from Lier, Ås or MET. Neither weather stations in Lier, Ås or
Blindern provide diffuse irradiance data, however, they do provide wind data.
There were two databases which supplied diffuse irradiation data, Meteonorm
and NASA. There were two data series from Meteonorm, one from the default
location Hurum, and the other from Skøyen. Since the two Meteonorm sets were
closest to the local sets considering global irradiation, these two data series where
assumed closest also in this regard. The diffuse data series used in the second
meteorological data set are, therefore, the average of these two Meteonorm data
series.
There were three databases which supplied wind data: Meteonorm, NASA and
RETScreen. RETScreen is a database with values from 1961 to 1990. When there
are no ground measurements either because they are not reliable or available,
the data is supplemented from the NASA database. From RETScreen there were
available ground measurements from Blindern. When comparing the temperature
data, RETScreens data was as good as a complete match to the measurements
from MET (see Appendix B). No average was, therefore, made between the three
databases and the data from RETScreen was assumed to be the most correct
data.
Table 3.1.5: The second meteorological data set which will be used in the simu-
lations
Global Diffuse Temperature Wind

Month Met, L, A Met MET RETScreen
January 11.09 7.5 -4.30 2.5
February 27.38 17.0 -3.99 2.6
March 67.47 36.0 -0.20 2.6
April 111.80 53.5 4.50 3.0
May 157.95 76.5 10.81 2.8
June 166.57 85.0 15.19 2.8
July 149.39 79.5 16.41 2.5
August 119.70 63.0 15.20 2.6
September 77.56 42.5 10.80 2.6
October 39.05 24.5 6.31 2.6
November 14.85 9.5 0.70 2.6
December 7.15 5.0 -3.10 2.6
Yearly 949.97 499.5 5.70 2.65
3.2 Orientation
For this project it was chosen to use fixed tilted PV modules. Other options
are seasonal tilted modules, tracking systems, one or two axis tracking systems
or BIPV such as roof tiles or shades. Systems with one or two axis tracking
3.2. ORIENTATION 29
are mostly used for ground mounting. In addition, such systems are assumed
to require more maintenance since there are more mechanical parts, and thus
not very practical for roof installations. Furthermore, such mounting systems
are more expensive. The modules should be directed towards south in order to
obtain as much irradiation as possible. This corresponds to an azimuth angle
of 0◦ . Magnetic and geographical south should in theory be the same. Due to
the magnetic flux lines on Earth they deviate with varying degree depending on
location. The modules should be directed towards geographic south, and not
magnetic south, since this would affect the system yield [7, p.41].
3.2.1 Optimum Tilt Angle
The amount of radiation collected on the solar modules should be as large as pos-
sible. The tilt angle of fixed modules can be maximized with regards to seasonal
performance or annual performance. The optimum tilt angle βopt , illustrated in
Figure 3.2.1 is defined as the tilt angle of highest annual irradiation and depends
on both latitude (φ) and local climate. The rule of thumb with regards to the
highest annual performance is a tilted angle approximately equal to the latitude
of the site, βopt ≈ φ [7, p.39] [9, p.862]. The larger the latitude, the larger
the difference between summer daytime and winter daytime and, therefore, the
larger the difference between the summer and winter irradiation. As a result it
can be anticipated that as the latitude increases, the optimal tilt angle should
give priority to the collection of summer over the collection of winter irradiance.
The system would be facing the equator and tilted at approximately 10◦ − 15◦
less than the local latitude by using Equation 3.2.1 [9, p.1014].
Figure 3.2.1: Tilt angle β
βopt = 3.7 + 0.69 | φ | (3.2.1)

It is argued in [15, p.38] that for optimum spring, fall or optimum annual perfor-
mance, the collector should be mounted at about 0.9φ. It is also argued in [24]
that the optimum tilt angle is approximately 0.9φ, however, only for latitudes
less than 65◦ . In [25] it is confirmed that the rule of thumb applies for nearly
all regions in the world, i.e. optimum tilt angle should equal the latitude angle.
However, larger deviations are given for regions of latitude higher than 45◦ N or
lower than 45◦ S. This is a consequence of significantly more clouds, hence more
diffuse irradiation which is best captured by flat tilted modules. A difference
between latitude and optimum tilt angle can be defined as [26]
w = φ − βopt (3.2.2)
In a simulation of 217 typical meteorological year data sites in the United States
[26] it was found higher w values associated with higher latitudes and lower
annual average clearness index kt,a . The values for w for fixed south-facing
panels, ranged between 0◦ and 14◦ . The results showed that the greatest values
of w occurred at locations in the coastal Northwest and the Great Lakes regions
with low kt,a values. These results were from the United States with highest
latitude reaching 50◦ north. Equation 3.2.3 is based on the results done in [26]
for fixed south-faced panels:
w = 0.0083φ2 − 0.1928φ (3.2.3)
When using equation 3.2.3 with the latitude for Oslo, it yields a w of 17.5171◦ .
However, this would be a result of extrapolation since the latitude in Oslo is
above 50◦ , and the reliability of this result is therefore uncertain. By using
equation 3.2.1 w becomes 14.59◦ . When using PVsyst the optimal tilt angle is
suggested to be 40◦ , facing geographical south. This corresponds to w = 19◦ .
This is with regards to the highest annual performance of the PV modules.
3.2.2 PVsyst and Orientation
In PVsyst an information panel indicates the corresponding Transposition Fac-

tor (TF), the difference with respect to the optimum orientation, and the avail-
able irradiation on this tilted plane. When the TF is zero, the optimal tilt
angle is found. PVsyst uses monthly meteorological calculations, which perform
quick transpositions. Solar data is often only available for the horizontal plane,
transposition from the horizontal irradiance to the tilted and oriented plane is,
therefore, preformed [27]. The calculation for the curves is performed for several
situations, in both directions from the point chosen (tilts at fixed azimuth, and
azimuths at fixed tilt). The Hay model [28] is used as the transposition model
in PVsyst, while the Erbs model (or the EKD model, Erbs, Klucher and Duffie
model) is the diffuse fraction model used [27]. The diffuse fraction model is used
to split the global irradiance into diffuse and beam components.
3.3. SHADE 31
As a result the optimal tilt of the PV modules is not equal to the latitude,
βopt 6= 59◦ . The optimal tilt is instead in the range of 40◦ − 44.41◦ , due to local
weather of more clouds, more diffuse radiation and a larger difference between
summer and winter irradiation. The lowest optimal tilt angle is chosen as the
starting point to use in the simulations, since it is a rooftop where the PV modules
are to be placed and, therefore, it is a limited area. Increasing the tilt angle of the
modules results in a larger pitch distance between the rows in order to maintain
the same length of shading from other rows. Decreasing the tilt angle further
from 40◦ should be tested in order to observe the effect this has on the energy
production and the Performance Ratio (PR) of the system. This will be further
discussed in Section 3.4.4.
3.3 Shade
Shade on the modules result in a reduction of irradiation on the cells, which
has a substantial influence on the current produced in the cell and module. The
irradiance deficit is one of the losses when considering shading. The irradiance
loss induces a non-linear electrical loss.
3.3.1 Electrical Effect of Shading
In every module there are placed bypass, or anti-parallel, diodes in order to

prevent ”hot spots” on the cells. A ”hot spot” can occur under certain operating
conditions where for example one cell is shaded and heated to the extent that
the cell material is damaged [8]. When shade occurs on one cell this does not
generate any currents. Instead the characteristics in this cell reverses, and the cell
becomes a load which absorbs energy from the circuit. The amount of currents
driven through the cell is converted into heat which could lead to ”hot spots”.
When a module is then integrated into an array without bypass diodes, the array
may force an even larger current through the cell. The bypass diodes prevent large
voltages from building up across the solar cells in the reverse-biased direction (see
Figure 3.3.1). The modules would obtain the largest shading tolerance if each cell
had a bypass diode connected across it. However, due to manufacturing reasons
bypass diodes are usually connected across 18−20 solar cells. A standard module
with 36 − 40 cells would have two bypass diodes [8, p.73]. A module with one
bypass diode could be explained as a module consisting of two sub-modules.
Shading which gives a lower amount of irradiance on the cell, results in a decreas-
ing current. With the bypass diode, the current does not decrease in the first
sub-module (the first 18 − 20 cells) which are not shaded. However, it decreases
when reaching the second sum-module with the shaded 18 − 20 cells, as seen in
Figure 3.3.2. The figure illustrates a standard module with 36 cells under STC
conditions in which one cell is shaded by 75 %.
Figure 3.3.1: Connection of bypass diodes
Figure 3.3.2: Module I-V curves with and without bypass diodes
Figure 3.3.3, 3.3.4 and 3.3.5 show how the I-V characteristic and the power
characteristic curve is affected as multiple modules or sub-modules are shaded
either in a series or a parallel connection. As seen in Figure 3.3.3, shades in series
connection gives two possible power point maxima. Both are possible operating
points if they lie within the inverter MPPT range. Which power maxima becomes
the operating point depends on the course of shading over time and the behaviour
of the MPPT [8, p.123]. The left operating point would with increasing number
of shaded modules not be within the inverters MPPT range.
The characteristic curves for arrays with parallel connections vary depending on
whether the shades cover several different strings or cover more modules within
few stings (illustrated in Figure 3.3.4 and 3.3.5). The power loss remains almost
constant for an array with parallel connections when shaded modules (or sub-
3.3. SHADE 33
Figure 3.3.3: Shading situation where two to eight modules in a string is shaded,
where the black curves = no shading, blue curves = 2 modules shaded, green
curves = 4 modules shaded, orange curves = 6 modules shaded and the red
curves = 8 modules shaded.
modules) on the same string increases. The power loss for parallel connections
increases when the numbers of shaded strings increase. Comparing Figure 3.3.4
and 3.3.5 it is seen that a shading situation covering fewer modules in several
strings is preferable to several modules in few strings. Considering the power loss
in a string connection, it is greater than the loss for a parallel connection.
Considering the case where the size of the array is not set, another argumentation
is called for. Considering only one string with only one module shaded, the loss
is stable independent of how many modules are connected in that string [29].
However, with other strings in parallel, the shaded cell distorts the array I-V
characteristics and displaces the MPP. As a consequence there occurs additional
loss in the other strings (not more than the production of one module). With
several shaded sub-modules within an array, consisting of series and parallel
connections, the loss increases rapidly. It was shown that the energy loss (in
percent of the normal production of the shaded string) increased significantly in
the array when it had several parallel strings, and one of the strings was shaded
[29]. The shaded string displaces the MPP operating point of the entire array,
so that the other parallel coupled strings do not work at their own MPP any
longer. This result is in accordance with a single cell shaded in one string and
Figure 3.3.4: Shading situation of an array where two of the strings are shaded
from two to eight modules, where the black curves = no shading, blue curves =
2 modules shaded, green curves = 4 modules shaded, orange curves = 6 modules
shaded and the red curves = 8 modules shaded
then connecting it to other non-shaded strings.

As a design criteria it would, therefore, be most beneficial to have the modules
with equal shading in the same string with a separate MPPT from the other
strings in order to get the maximum amount of energy produced from the entire
system. Fewer modules per string would in addition make sure that more module
strings would operate at a MPP closer to their own.
3.3.2 Site Assessment
A site assessment of the roof where the PV installation is proposed in Oslo was
performed January 27th, 2012. The commercial building is located at Skøyen in
Oslo, 59◦ 550 23.4900 N and 10◦ 400 49.5400 E which can be seen in Figure 3.3.6.
On the roof the shading conditions were evaluated. The ideal installation of a
PV system is where there are no shadows at all. When considering shadows on
the site the following was evaluated [7, p.161]:
• Natural landscapes, e.g. surrounding mountains
3.3. SHADE 35
Figure 3.3.5: Shading situation of an array where two modules on one to four
strings are shaded, where the black curves = no shading, blue curves = 2 modules
shaded, green curves = 4 modules shaded, orange curves = 6 modules shaded
and the red curves = 8 modules shaded
• Trees or other vegetation

• Other buildings
• Parts of the actual building where the system is to be located, chimneys
etc.
The roof is approximately 1200 m2 . Figures 3.3.7 and 3.3.8 show the space where
the PV modules should be placed and the present elements located on the roof.
These elements will cast shadows and reduce the area where the modules could
be placed. In Figure 3.3.7 there is seen a twelve storey building which also could
cast a shadow on the PV installation. It is considered that neither the vegetation
nor the natural environment could be considered a great contributor of shadow.
Mounting system
The type of mounting system chosen for the installation and the distance be-
tween these greatly affects the amount of shadow on the modules. Figure 3.3.9
shows two different mounting systems where Figure 3.3.9a shows aerodynamic
Figure 3.3.6: Location of the commercial building in Oslo marked with a red
circle [30]
mounting system, and Figure 3.3.9b shows a ballasted mounting system. With
these two mounting systems there is no need to penetrate the roof which could
cause problems with regards to isolation etc. Roof penetrating mounting system
exists, but is not taken into consideration in this study. The roof has been di-
mensioned for heavy snow loads and it is, therefore, assumed that the weight of
the installation is not a problem. The issue of weight requirements on the roof
has not been considered further in this study.
The amount of shading on the modules depend on the spacing between them,
the module height, row length, tilt angle and latitude location. Figure 3.3.10 and
Equation 3.3.1 describes two modules and the shadow cast by the first onto the
second when oriented towards the equator [33].

D + A cos β
HS = A 1 − (3.3.1)
A cos β + A sin β cos ψ/tan α
In equation 3.3.1; HS describes the length onto the second module cast by the
first, with a distance D between them, a tilt β, and a width of the modules A on a
given time marked by azimuth ψ and solar altitude α. Equation 3.3.1 was solved
with respect to the distance D between the modules and HS = 0, and tested in
Matlab with different types of modules at different days, time of day and with
different tilt (see Appendix C). As a dimensioning criterion the modules should
3.3. SHADE 37
Figure 3.3.7: Horizon north west
not shade each other at the equinoxes. Since this marks when day and night is
equally long, either of the equinoxes could be used. In the Matlab calculation
spring equinox was used, March 21st.
Table 3.3.1: Spring equinox pitch distances in meters, NASA
March 21st 20◦ 30◦ 40◦ 53◦ 59◦

Mono
10:00 1.661 1.897 2.076 2.213 2.239
11:00 1.648 1.878 2.051 2.182 2.205
12:00 1.649 1.878 2.053 2.185 2.209
Poly
10:00 1.574 1.780 1.967 2.097 2.121
11:00 1.561 1.779 1.943 2.067 2.089
12:00 1.562 1.781 1.946 2.070 2.092
Table 3.3.1 shows the pitch distance between the rows in order not to shadow
Figure 3.3.8: Horizon south west
(a) Knubix [31] (b) DPW [32]
Figure 3.3.9: Two different mounting systems one in landscape, the other in
portrait
each other on spring equinox. The NASA numbers for azimuth and solar altitude
has been used in this table. As seen, a higher tilt angle results in a larger pitch
distance. The pitch distance is illustrated as D + A cos β in Figure 3.3.10. The
required distance D to avoid shading during March 21st for a polycrystalline
module with 40◦ tilt is between 1.08 − 1.2 meters (pitch of 1.95 − 1.97 meters).
The distance close to December 21st is in a larger range of 5.1−10.4 meters (pitch
3.3. SHADE 39
Figure 3.3.10: Shading of modules in rows [33]
of 5.8 to 11.2 meters) depending on which time period the modules are supposed
to be shade free (see Appendix C). The distance for a monocrystalline module
would be somewhat extended due to a larger width. When having a limited space
to place the modules it, therefore, becomes critical that the modules are placed
as the Knubix modules in landscape, not as the DPW modules in portrait, in
order to decrease the distance between the module rows. Since the difference
between the mono- and the polycrystalline modules is so small when tilted 40
degrees on the March 21st a pitch distance of 2 meters will be used in the 3D
representation of the PV field.
The choice of mounting system would have an effect on the edge zones sur-
rounding the modules and the module tilt. The edge zone is the distance from
the installation to the edge of the building. Table 3.3.2 shows the mounting
systems manufacturers that have been looked into in this study and their spec-
ifications. The different mounting types have not been considered as a limiting
factor of which modular tilt the installation should have. These are simply used
as an illustration of some of the manufacturers within ballasted and aerodynamic
mounting systems. The edge zone specifications from Solar Dock arise from the
State of California’s building and fire safety codes (6 feet). DPWs edge zone
requirements are based upon the American Society of Civil Engineers (ASCE)
Code (require at least 3 feet). The edge zone requirements from Knubix are based
on a calculation considering the height of the building. The edge zone for the
SolarSTEP mounting system is stated to be valid for buildings with maximum
18 meters, however, could be valid for buildings up to 40 meters. The building
in question is approximately 22.5 meters high.
The edge zone for the Skøyen installation is assumed to be approximately 2

meters, since this is the middle value of the five options in Table 3.3.2.
Table 3.3.2: Different mounting types and some specifications
Type Edge zone minimum Modular tilt

Knubix 4.48 10◦ - 20◦ Landscape Aerodynamic
DPW 0.91 20◦ - 45◦ Portrait Ballasted
◦ ◦
SolarSTEP 2.00 5 - 20 Landscape Aerodynamic
◦ ◦
SunPower 5 - 10 Landscape Aerodynamic
Solar Dock 1.83 15◦ - 35◦ Landscape Ballasted
3.3.3 PVsyst: Horizon and Near Shadings
In PVsyst there has been made a distinction between far and near shadings. The
far shadings are defined as shades cast by the horizon while the near shadings
are defined as shades cast by near objects which change during the day.
Horizon
The far shadings are supposed to act globally on the PV plane. Acting globally
would mean that it would not give any partial shading on the installation. The
sun would either be or not be visible on the field. Horizon defining shadows
would naturally consist of surrounding environment such as mountains and are
defined with a horizon line in PVsyst (illustrated in Figure 3.3.11). The obstacles
should be limited to approximately twenty times the PV-array size [10]. The
length of the PV installation is approximately 40 meters, which gives a radius
of approximately 800 meters before obstacles could be defined as horizon. The
reason for this is that the horizon line is supposed to be viewed in the same way
from any point of the field.
When a horizon line is accounted for in PVsyst the beam component of the ir-
radiance is or is not visible on the field. PVsyst determines the exact time when
the sun crosses the horizon line and weights the beam hourly value before per-
forming the transposition. The diffuse part of the irradiance is assumed isotropic
in the program. The diffuse part does, therefore, not depend on the position
of the sun, it is the same irradiance which is coming from any direction of the
sky and is therefore a constant factor during the year [29]. The albedo contribu-
tion in PVsyst is considered to be linearly decreasing according to the horizon
height. When using meteorological data from ground stations the horizon effect
is already taken into account for that station. A comparison of the measuring
station horizon with respect to the field horizon could be accounted for.
A horizon line for the Skøyen field is not drawn, because of low mountains sur-
rounding the site. In addition some horizon could already have been taken into
account in the two measured data sets from Lier and Ås.
3.3. SHADE 41
Figure 3.3.11: Horizon line drawing in PVsyst
Near Shadings
Shadings which change during the day and the season and only partly affect the
PV field are the near shadings. The fraction of how the PV field is effectively
shaded is defined in PVsyst by shading factors. In order to calculate the shading
factor at any time as a function of the position of the sun requires a full 3D
representation of the field and its surroundings.
The 3D representation of the building (Figure 3.3.12) was constructed mainly by
the help of drawings of the building and Google Earth. Some assumptions were
in addition made when considering the height of roof top shading elements with
the assistance of photos taken during the site assessment. As mentioned there is a
800 meter radius where near shading objects should be taken into account. Inside
this radius the twelve storey building illustrated in Figure 3.3.7 is located. This
building was, therefore, represented in 3D, with measurements from drawings
provided by plan-og bygningsetaten in Oslo. Moreover, Google Earth was used
in order to place the building in approximate accordance with the PV field.
Within PVsyst the user has three options regarding shade:
1. No shadings
2. Linear shadings
3. Shading according to module strings
The first option would not give any shading and would, therefore, not require
a 3D representation of the PV field and the surroundings. The second option,
linear shadings, uses the table of shading factors as a function of the altitude
Figure 3.3.12: The 3D representation of the building with a PV installation on

the roof in PVsyst
of the sun and azimuth. The table is used since calculation of a shading factor
for each hour would take a considerable amount of time. However, the hourly
shading factor can be calculated fast by interpolation during the simulation [10].
The shading factors are applied to the beam component of the irradiance and
allow for an evaluation of the irradiance deficit on the PV plane were the effect
is proportional to the shaded area. Claiming that the shaded area on the field is
proportional to the irradiance deficit is, however, not representative for the real
energy losses. The electrical behaviour also represents a loss. In order to take
into account the electrical behaviour a third option is given to the user: shading
according to module strings. The PV field is split into rectangular areas (as seen
in Figure 3.3.13), each representing a whole string of modules. When calculating
the shading factors the corresponding string becomes unproductive if a rectangle
is shaded. It is stated that the real shading loss should lie between linear loss
and the loss caused by shading according to module strings.
In order to calculate the shading factor table the program first carries out a
transformation of the coordinates of the whole system in order to point the z-
axis in the direction of the sun. For each sensitive element of the PV field, the
program projects each elementary surface of the system on the plane of the field
being considered. The intersections between the field elements and the positive
projections are calculated for each element. The global shading on the field
3.3. SHADE 43
Figure 3.3.13: The modules on the roof drawn with rectangles
element consists of these elementary shadows which forms a polygon. The ratio
of the area of the shadow polygon, to that of the sensitive element is the shading
loss factor. For each sensitive field element (PV elements) this process is repeated
[10].
Figure 3.3.14: The shading factors of a PV system with a 20 ◦ modular tilt and
a 2 meter pitch and a 2 meter edge zone, calculated in PVsyst
The diffuse irradiance component is assumed isotropic as in horizon, and is there-

fore constant during the year. The value is, however, not necessarily the same
value as in horizon, since it is affected by near shading objects. In near shadings
the diffuse irradiation could be explained as the integral of the shading factor of
the visible part of the vault of heaven seen from the module [10]. This integral is
the same during the year, but depends on the 3D representation of the field. The
albedo component is also considered as a constant during the year in PVsyst.
The diffuse and albedo shading factors are the same for linear shading and for
shading according to module strings. Reflections are not taken into account in
PVsyst due to their very special hourly presence and since the reflection usually
only hits a few modules within a string.
Along with having the choice of which type of shading is to be considered, a frac-
tion for electrical effect in percentage is also possible to adjust in PVsyst. The
difference between the value given by the linear loss (deficit of irradiance) and
the loss according to module strings (taking into account the electrical effects) is
due to electrical mismatch. The fraction of electrical effect acts on this difference
and is adjusted by a percentage of intensity of the real effect. A percentage of
0 would equal pure linear shading, while 100 percent would equal full shading
according to strings. The fraction for electrical effect is dependent on the shade
distribution on the field and the electrical array configuration [29]. For shade
arrangements where the shades are regular (mutual shadings of sheds) a suffi-
cient set of sub-modules are shaded at the same time, completely suppressing the
production of the full string, which would be close to 100 percent of shading ac-
cording to strings. In cases where there are more distributed shades (for example
from a chimney) where only some sub-modules are affected, some of the energy
could be recovered due to the bypass diodes, and the fraction of electrical effect
could be in the order of 60−80 % [10]. A fraction of electrical effect of 60 % is as-
sumed to be a realistic value with as many shading objects as there is on the roof.
Shade analysis
With the help of the shading animation during a day which is a possibility in
PVsyst, a shade analysis was performed. The shading simulation in PVsyst gives
a graphical view of the shading situation for every half hour of a chosen day. For
each half hour during the day the modules that where shaded was recorded as
either partially shaded or fully shaded. A shading analysis was performed for
March 21st since this is the day when there is not supposed to occur module
shading owning to other modular rows. The shading analysis would, therefore,
only illustrate the shading caused by the shading elements on the roof. There
has been made a distinction between partially shaded modules and fully shaded
modules. However, the elimination of redundant modules is to be done on account
of the longest duration of shade, not depending on if it is partially or fully shaded.
A shade analysis was done for the final alternatives, since a change in modular
tilt and pitch distance would change the shading lengths.
3.4 System - Matching Array and Inverter

In PVsyst the main components in the PV system are chosen: the module and
inverter type and manufacturer. Type of cable and mounting system is indirectly
chosen when determining or calculating the ohmic loss, and when drawing the
3D representation of the PV field.
3.4. SYSTEM - MATCHING ARRAY AND INVERTER 45
3.4.1 Choosing Module
Hopesolar, BestSolar, King-Pv, Upsolar, Schuco, Kyocera, REC, Innotech Solar,

LG, Samsung, SolarWorld, Boch and SunPower are some of the many manufac-
turers of poly- and monocrystalline modules today. In order to narrow down the
selection of modules, three different module manufacturers where chosen:
1. Renewable Energy Corporation (REC)
2. Innotech Solar (ITS)
3. SunPower
REC was chosen since it is originally a Norwegian based company. However, it
is now a worldwide organisation with several material factories in the USA and
wafer and module production in Singapore. From each of the three manufacturers
one module series was chosen to be evaluated further in this study. From REC
the Peak Energy EU Series was chosen. From ITS the EcoPlus series was chosen
since the module is manufactured in Sweden. SunPower was chosen since they
manufacture high efficiency monocrystalline modules, and the series chosen was
the E20 series. One module type was chosen from each of the three series. In
order to produce as much energy as possible the module type from each series
with the highest watt peak (Wp) was chosen. From REC REC250PE was chosen
(250 Wp, polycrystalline, 15.1 % module efficiency), from ITS EcoPlus-PolyUp
(250 Wp, polycrystalline, 15.2 − 15.8 % module efficiency) and from SunPower
SPR-327NE-WHT-D (327 Wp, monocrystalline, 20.1 % module efficiency).
When choosing a module within a kilometre from the coastal line a module
certified with IEC61701 should be used [7, p.166]. IEC61701, ”salt mist corrosion
testing of PV modules”, ensures that the modules are salt tested. The building
in question at Skøyen is located 500 − 600 meters from the coastal line. The
certification IEC61215 could be extended to a higher load capacity rating of
5400Pa which could be relevant in regions that are subjected to heavy snow
loads. When investigating the module certifications these two certifications were
checked for explicitly. However, going into depths of finding out whether the
manufacturer has chosen to increase the load capacity has not been investigated
further.
Both series chosen from REC and ITS has the IEC61701 and IEC61215 certifica-
tions. The SunPower series chosen is certified with UL1703 which is an American
certification standard. It is assumed that the SunPower certification corresponds
to the IEC61701 and IEC61215.
3.4.2 Choosing Inverter
There are different choices between inverters, such as central inverters, multi-
string inverters, string inverters, modular inverters, three and one phase inverters
and inverters with or without galvanic isolation in form of transformers. When

the inverters are selected the following has to be taken into account [7, p.168]:
• The peak rating of the PV array
• Whether the solar modules are all in the same plane that is the same angle
and direction.
• The type of shading that occurs on the array
• The average annual yield
The peak rating of the PV array vary depending on how many modules that
are placed on the roof and whether it is a polycrystalline or a monocrystalline
module. With the modules placed as illustrated in Figure 3.3.12, the array has
a peak rating of 26 kWp with poly crystalline modules, and a peak rating of
34 kWp. Since the array has a peak rating above 20 kWp a central inverter
could be chosen. However, due to the amount of shading elements on the roof
string inverters were chosen. If the modules were not all in the same plane, this
would also have been a reason to use string inverters instead of central inverters.
As with the modules the list of inverter manufacturers is long, and has been
narrowed down to the three following:
• Eltek
• SMA Solar Technology AG
• Danfoss
The string inverters THEIA TL String, and THEIA HE-t by Eltek were chosen
since they have high efficiency and one MPPT. The multi-string inverters Sunny
Tripower of SMA was chosen since it has two MPPT inputs and high efficiency.
Danfoss TLX series was chosen since they manufacture multi-string inverters
with as much as three MPPT inputs per inverter and has high efficiency. For
further information see the technical information in Appendix F.
For each of the three modules chosen, inverters from one of the three inverter
series were chosen. In order to choose inverters all possible inverters (which are
not over- or undersized) were simulated in the system as close up to 104 modules
which is the limit when maintaining the edge zone of 2 meters. The system had
otherwise 20◦ module tilt angle, 60 % electrical effect and was simulated with
the new meteorological data set.
3.4.3 Matching Module and Inverter
The array and the inverter have to match in three areas: voltage, current and
power. First, by taking voltage into account the array should be within the
MPPT range of the inverter. Second, the number of strings in parallel cannot
exceed the maximum input current of the inverter. Third, the output power of
the inverter limits how many modules in total that can be connected to it. In the
simulations, the single-diode model explained in Section 2.2 is the model used
to describe a PV module in PVsyst. By using this model to describe a module,
it implies that all cells in the module are identical. More complicated models
for the PV cell exists, such as the two-diode model. The single-diode model is
considered sufficiently accurate in PVsyst, since there is a moderate accuracy of
the input parameters (supplied by the manufacturers) and the fact that mismatch
losses are taken into account separately [10]. Mismatch losses will be explained
in Section 3.4.4.
Matching the PV array to the voltage specifications of an inverter
As explained in Section 3.1.1 the module voltage is dependent on the temperature.

When arrays are matched with inverters the extreme temperatures of winter and
summer are used in order to establish the size of the strings in terms of minimum
and maximum number of modules.
Determining the minimum amount of modules in a string, the maximum cell
temperature is used. The minimum inverter voltage is divided by the module
maximum temperature voltage in order to find the minimum amount of modules
in a string. The reason for this is that the MPP voltage of the array should
not fall below the minimum operating voltage of the inverter, since the inverter
could switch off or not operate at the arrays maximum possible output (see the
red characteristics in Figure 3.4.1). The minimum amount of modules in a string
is defined by
VM P P inverter min
nmin = (3.4.1)
VM P P module max temp
where
γV
VM P P module max temp = 1 + ∆T · M P P · VM P P ST C (3.4.2)
100
and where ∆T is the temperature difference between the highest module or cell
temperature and the STC temperature (for example 70◦ C - 25◦ C) and where
γV is the voltage temperature coefficient of the module [8].
The inverter may not always be operating at the ideal MPP, because of shading,
less irradiance etc. A safety margin of 10 % could, therefore, be allowed. In
addition a factor for the voltage drop expected across the DC cables could be
taken into consideration in this calculation [7, p.187]. Equation 3.4.1 is used
in PVsyst without a safety margin of 10 %. The voltage drop across the DC
cables is taken into consideration in another calculation in PVsyst as explained
in Section 3.4.4 and it should, therefore, not be taken into consideration twice.
If taken into account, the safety margin of 10 % would increase the minimum
amount of modules. This consideration has not been taken into account in this
study.
Figure 3.4.1: PV module characteristic curves and operating range of an inverter.

The red module temperature 70◦ C, green module characteristic = 20◦ C, blue
characteristic -20◦ C
The maximum amount of modules in series is determined by the minimum tem-

perature of the modules since it is at this temperature the voltage of the array
is largest. At first light in the morning the cell temperature will be close to the
ambient temperature since the sun has not had time to heat the modules. The
open circuit voltage has to be lower than the maximum voltage at the inverter
in order to stay within the MPP voltage of the inverter (see the blue character-
istic in Figure 3.4.1). The maximum amount of modules in series is, therefore,
computed as
Vmax inverter
nmax = (3.4.3)
VOC module min temp
where
γ
VOC
VOC module min temp = VOC ST C − · VOC · ∆T (3.4.4)
100
and where γVOC is the open circuit temperature coefficient of the module (in
V/◦ C) and ∆T is the temperature difference between the STC of 25◦ C and the
lowest cell (or module) temperature (for example -20◦ C, -25◦ C) [8]. Equation
3.4.3 is used in PVsyst. It is recommended in [7, p.190] that a safety margin of
5 % below the specified maximum inverter voltage should be allowed to account
for manufacturer tolerances and in case of extremely cold weather. This safety
margin is not taken into account in PVsyst. The two sizing criteria are shown in
Figure 3.4.2 as the two vertical lines marking the VM P P range of the inverter.
Matching the PV array to the current specifications of an inverter
The current of the modules is only vaguely affected by temperature changes

as explained in Section 3.1.1. The largest current of the modules has to be
smaller than the maximum inverter current in order to function. By evaluating
the current of the modules the number of strings in parallel can be assessed in
accordance with
Imax inverter input

nstrings in parallel = (3.4.5)
ISC ST C + (γISC · ∆T )
where γISC is the short circuit current temperature coefficient of the module in
A/◦ C and ∆T is the temperature difference between the STC and the highest
cell (or module) temperature[8]. The current criteria is shown in Figure 3.4.2 as
the horizontal line of 210 A.
Matching the PV array to the power specifications of an inverter
In the inverters specification it is common to include maximum PV array rated

power and maximum DC input power. The dotted line in Figure 3.4.2 show
the inverter maximum DC input power, which is the maximum amount of DC
power that the inverter can convert to AC. This number is usually lower than the
maximum PV array power due to the losses in the system before it reaches the
inverter. Finding how many modules that can be connected to an inverter the
maximum PV array rated power or maximum DC input power is divided by the
peak effect of the module in question. In PVsyst it is assumed that the division
is done with respect to the maximum DC input power on account of Figure 3.4.2.
3.4.4 Losses
The losses that occur in a PV system could be explained as the cause of a

reduction of the available array output energy with respect to the nominal power
of the PV module stated for STC conditions [10]. The dimensioning criteria
Figure 3.4.2: Array or inverter sizing in PVsyst, 16 REC245PE modules in series

and 18 strings, using a Eltek inverter, Theia 7200TL
of the performance ratio is discussed first followed by the losses which are taken
into account in PVsyst (irradiance loss, thermal loss, ohmic loss, mismatch losses,
quality loss, soiling loss and incidence angle modifier loss) A summary of the final
parameters used during the simulations are found in Appendix D.
Performance ratio
PR is defined in [34] as
Yf
PR = (3.4.6)
Yr
where Yf is the final system yield, and Yr is the reference yield. The reference
yield which is the system yield without losses, while the final system yield is
taking into account the losses in the system. The PR as a result indicates the
overall effect of losses. The losses are affected by array temperature, incomplete
utilisation of the irradiation and system component inefficiencies or failures [34].
Since the PR is normalized it can be used in order to compare PV different system
types for the same location, as well as identical systems installed in different
locations [7, p.247].
It was presented operational performance results of 368 grid-connected PV in-
stallations in [35]. PV systems installed during 1991 − 1994 and 1996 − 2002
were compared, and an increase of the annual PR values where observed from
an average mean PR of 0.65 to 0.74. The maximum PR range of the systems
installed in 1991 − 1994 was 0.65 to 0.7. For the newer installations (1996 − 2002)
the maximum PR was in the range of 0.75 to 0.8. It was stated that by taking
into account the efficiency values of inverters of today (article published in 2004),
optimum values of annual PR between 0.81 and 0.84 could be achieved [35]. An
annual PR of 0.6 to 0.8 for a grid-connected system was the result of the perfor-
mance analysis of 260 PV plants in the IEA-PVPS Task 2 database [12], which
is in the same range as indicated range for a grid-connected system in [9] 0.65 to
0.75. However, it is also stated in [9] that well designed PV installations, with
high performance have a PR in the range of 0.7 to 0.8.
Taking the arguments above into account the PV installation in Oslo should have
a PR optimally above 0.8 and if lower, not below 0.78. This should be used as a
dimensioning criteria in the simulations.
Irradiance Loss
Irradiance loss describes the loss which is caused when the irradiance differs
from the STC conditions. The nominal efficiency for the modules are specified
for STC conditions. This efficiency decreases when the irradiance reaching the
module decreases, that is going below 1000 W/m2 .
Thermal losses
The thermal losses describe the loss as a result of the temperature difference
between the modules in the array and the 25◦ C which is the temperature at STC
conditions. The operating temperature of modules are normally much higher
than 25◦ C. They can reach as much as 60 − 70◦ C as calculated in Section
3.1.1. In PVsyst the thermal loss is calculated following the single-diode model
(as explained in Section 2.2).
In PVsyst there are two ways to influence the calculations of the thermal loss,
either by defining the field thermal loss factor, or the standard NOCT coefficient.
The program gives the equivalence between the two. The energy balance between
ambient temperature and the cell temperature is due to irradiance as seen in
Equation 3.4.7
U · (Tm − Ta ) = A · G · (1 − ηm ) (3.4.7)
Equation 3.4.7 describes the thermal behaviour, where A is the absorption coef-
ficient of solar irradiation, and ηm is the module efficiency. The default value for
the absorption coefficient is 0.9. However, this value can be changed in PVsyst.
PVsyst calculates the PV efficiency according to the operational conditions of
the module when possible. PVsyst uses 10 % as the PV efficiency if calculation
is not possible.
The thermal loss factor is described by
U = Uc + Uv · v (3.4.8)
where Uc is the constant component of the thermal loss factor, Uv is a factor

proportional to the wind velocity v. These factors depend on the mounting of
the modules, sheds, roofing, façade, etc. The mounting gives an idea of how well
the system is ventilated, e.g. sheds would give free circulation of air on both sides
of the module. For the field thermal loss factor there are three default options
which could be chosen:
1. ”Free” mounted modules with air circulation
2. Semi-integrated with air duct behind
3. Integration with fully insulated back
The standard NOCT coefficient can be defined instead of the field thermal loss
factor. In such a case PVsyst uses Equation 3.4.7 with the definitions of the
NOCT conditions (described in Section 3.1.1).
The NOCT coefficient is usually specified by the manufacturers. The NOCT
coefficient is the temperature attained by the PV modules with free air circulation
all-around the module under the conditions described above. There are two ways
of defining NOCT, as an open circuit (at Voc ) or loaded (at Pmpp ) [10]. In
the definition of NOCT the operating state is not clear. As a result PVsyst
assumes that the definition concerning open-circuit modules has a ηm = 0 and
ηm = 10 % when the system is loaded. PVsyst lets the user choose how the
thermal loss should be defined, either by the NOCT coefficient of the module or
by the thermal loss factor. In this study it has been chosen to use the NOCT
coefficients defined as open-circuit in order to define the thermal losses.
Ohmic losses
Ohmic losses (R · I 2 ) taken into account in PVsyst are the ones caused by the
wiring resistance. In PVsyst the user has the option of including transformer
losses if an external transformer exists. When determining the global wiring
resistance of the DC circuit it can be done by either setting the ohmic loss ratio
or by explicitly defining the loss in mOhm. There is also the option of performing
a computation in PVsyst in order to calculate the global wiring resistance.
The two wiring layouts which are taken into account in PVsyst are shown in
Figure 3.4.3, a layout with parallel strings and Figure 3.4.4, a layout with groups
of parallel strings. The blue lines illustrate the string module connections. The
pink lines show the connections to the main box or the connection box, while the
green lines illustrate the wiring between the connection boxes of the strings to
the inverter. The wiring connection most similar to the one in Skøyen would be
the one illustrated in Figure 3.4.3. In the Skøyen case there would not be any
need for a connection box, the string module connections would go straight to
the inverters.
Figure 3.4.3: Wiring connections for parallel strings
Figure 3.4.4: Wiring connections for groups of parallel strings
In addition to taking DC cable lengths into account the user has the option of
including the AC circuit which is the wiring from the inverter to the injection
point. If it is of significant length this should be taken into account. If there is an
external transformer present in the system then, iron loss and resistive/inductive
losses can be taken into account as well. The ohmic loss ratio is the ratio between
2
the wiring ohmic loss Pwir = Rwir · Isc and the nominal power Pnom (array) =
2 Vmp
Rarray · Isc . Where Rarray = at STC and Rwir is equal to the global wiring
Imp
resistance of the system. The global wiring resistance is obtained by placing
the sub-array wiring resistances in parallel. The amount of wiring loss could
be adjusted for by choosing an alternative cross section. In this case a resistive
wiring loss of 1.5 % is chosen as an upper limit which can be altered with other
cross sections were the prices of different wires could be taken into account.
Module quality - mismatch
Module efficiency loss describes the deviation of the average effective module effi-
ciency with respect to the manufacturers’ nominal specification. This used to be
a great uncertainty when evaluating PV system performance. Now modules are
sold with guaranteed power assertions, a given tolerance and flash-test assertions.
The module efficiency loss default is 0.1 %. However, it is stated that this value
should express the users own confidence in the modules performance [10]. With
respect to the manufacturers’ specifications this value is, therefore, set to 0 % in
this study.
Mismatch losses take into account that real modules in an array does not neces-
sarily present the same I-V characteristics. PVsyst has made a graphical tool for
determining the effect of parameter mismatch, for either the cells in a module
or the modules in an array. In this tool there is used a random dispersion of
the characteristics of the short-circuit current for each module by Gaussian dis-
tribution or square distribution. The tool gives the power loss at MPP and the
current loss at 90 % of VM P P . The simulation asks for a mismatch loss factor
which is different for MPP or fixed-voltage operation. This is taken as a constant
during the simulation. The type of module used would give different mismatch
losses. Performing the simulation the default values for the mismatch losses are
considered sufficient and are, therefore, proceeded with in the simulation.
Soiling loss
In PVsyst there is the possibility of defining either a yearly soiling loss factor or
monthly values all expressed in percent. Soiling could behave as mismatch losses
since some cells are shaded and, therefore, give a different I-V characteristic
than normally. The soiling loss is defined as a percentage of the STC power.
It corresponds to a global diminution of the array power and is equivalent to
irradiance diminution. It is not assumed soiling on the modules. Even if Oslo
is the largest city in Norway, the air is considered clean. What is known, is
the possibility of snowfall (se Figure 3.3.7 and 3.3.8). January is assumed to
be the month with most snow in Oslo, still it is possible to observe snow in
December, February and March. Snow around the modules would improve the
efficiency due to a reduction in temperature, and the reflections from the snow
cover could affect the modules in a positive manner. With increasing tilt angle of
the modules it could be argued that the snow would simply slide off the modules
or melt. However, if the snowfall is heavy during night and does not slide off the
modules the snow would act as soil and cause a diminution of irradiance. As a
result monthly soiling losses has been assumed to 35 % in December and February
and 50 % in January. It could be argued that soiling losses should also be added
in March. A discussion or an evaluation whether snow should be considered as
soiling loss and not taken into account in another manner in PVsyst could be
done in a later stage. The snow is taken into account as soiling loss in this study
with the assumption that it would give a more conservative perspective on the
energy production during winter.
Incidence Angle Modifier (IAM) losses
IAM could be called reflection loss. The reflection loss is an optical effect which
occurs when the irradiation reaches the glass which is protecting the PV cells
which lie underneath. The loss follows Fresnel’s laws with regards to transmission
and reflection on the protective layer (the glass), and on the cell’s surface. In
PVsyst this loss is taken into account by Equation 3.4.9
1
FIAM = 1 − b0 · ( − 1) (3.4.9)
cos i
Figure 3.4.5: Incidence angle modifier
where i equals the incidence angle on the plane. Equation 3.4.9 is the Fresnel
laws by using a parameterisation called ASHRAE. The default value for b0 is
0.05. This value is set for crystalline PV modules. For thermal solar modules
b0 = 0.1 [10].
3.4.5 Cable sizing
Cable sizing is indirectly considered in PVsyst when defining the ohmic losses in
the system. In this study where only the maximum ohmic loss has been defined
in the program, approximate cable sizing was done after the simulation. Cable
sizing was only done for the final alternatives since it has to be done after all the
strings have been assigned their string. In order to size the wirings to have a loss
below 1.5 %, all lengths of the string wirings have to be measured. Choosing the
wiring path several factors could be taken into account, such as:
• The Current Carrying Capacity (CCC)
• Practical considerations
• Aesthetic consideration
• Economic considerations
• Voltage loss
The CCC refers to the maximum amount of current a conductor can have flowing
through it without causing damage. The CCC is affected by the cross sectional
area of the conductor, type of insulation around it and the environment in which
the cable is installed. A reduction factor of the CCC could be added if several
conductors were placed together causing heat between the conductors [36]. Here
only practical considerations were taken into account due to the vast amounts of
assumptions which would have to be made in order to evaluate both the CCC
and the economic considerations. The upper limit of the voltage loss is taken
into account in PVsyst which leaves the practical and aesthetic considerations.
The aesthetic considerations are evaluated as not important due to the fact that
the installation is on a roof where only a few people have access.
The minimum cross section size was calculated with Equation 3.4.10:
2 · LDC cable · IDC · ρ

ADC cable = (3.4.10)
Loss · VM P P string
where LDC cable is the route length of the DC cable, IDC is the string current, ρ
is the resistivity of the wire, Loss is the percentage of maximum voltage loss in
the conductor and VM P P string is the string voltage. The route length of the DC
cable is adjusted for with the factor of 2, since this is closer to the total circuit
wire length.
3.5 Module Layout

Module layout is a descriptive tool in the version of PVsyst which has been tested
during this study. In a future version this will be related to the 3D representation
3.6. CONNECTION TO THE GRID AND METERING 57
of the PV field. The module layout is independent of the electrical design and
the simulation of the PV system. In module layout the user has the possibility
to recreate the PV field where each module is represented. Once all the modules
in the system have been placed, defining the electrical configuration of the field
can be performed. Each module is assigned to a string which then is assigned
automatically to a MPPT input. The number of bypass diodes per module is
defined by the user as well as the coupling of them resulting in sub-modules in
length or width. When all modules have been assigned the electrical shading
effects of a string can be observed. Shades are drawn manually over the modules,
and the global and diffuse irradiance on the plane can be set manually as well
as the temperature. The effect of the shading drawn is given in percentage for
the linear shading factor on the beam component, on global irradiance and the
additional electrical non-linear loss. Results of the shading test are given for the
whole system as well as for each inverter. In addition I-V and power characteristic
curves are shown for each inverter.
Since the shadings are drawn manually in module layout the results would be for
that specific shading situation in that exact moment in time. It would require
several shading situations during different times of the day and different days
during the year in order to get an idea of the shading effect on each inverter and
on the whole field. The module layout has, therefore, not been used in this study.
However, when module layout could be used with the 3D representation of the
field and its surroundings it would give a more realistic result and more details
concerning the effect of shading on each string and the whole system.
A layout of the module string configuration has been made for the final alterna-
tives. The module constellation in near shadings have been the basis for the set
up with its initial modules. In cases where the 3D representation had to many
modules, evaluating which modules which would not be part of the final layout
was done with the shading analysis which was made in near shadings.
3.6 Connection to the Grid and Metering
3.6.1 Connection to the Grid and Customer Status
End users producing energy is relatively new in Norway. Nonetheless, in March

2010 there was given a general dispensation to simplify the processes related to
customers with an energy surplus. Such customers, so called surplus customers,
are defined by NVE as:
End users of electricity that have an annual generation that normally does not
exceed their consumption, but have during certain hours a surplus of electricity
that can be fed back to the grid. Generation which requires licensing or producers
that supply electricity to other end-users are not covered by the dispensation.[37,
p.32]
The dispensation from NVE entails that the local network company buys the
surplus energy from the customer. The surplus customer does then not have to
enter into a balance agreement with the TSO of the Norwegian electric power
system. The dispensation also entails that the surplus customers could be ex-
cluded from trade concession (omsetningskonsesjon). Per now it is voluntary
for the local network company to offer the suggested arrangement for surplus
customers. If connection of new surplus customers demands a reinforcement of
the network, then the local network companies may require a connection charge
from the surplus customer [37, p.33]. All new surplus customers would have to
contact their local network owner in order to investigate whether they would like
a surplus customer in their grid. The regional network owner and operator for
Skøyen is Hafslund Nett AS.
Hafslund Nett does not demand a specific size of energy surplus from the surplus
customer other than that it is small enough that it does not fall under NVEs
trade concession. Technical demands from Hafslund involve for example that the
surplus customers plant cannot contribute to a reduction in voltage quality of
other customers in the network. Other demands which are set by Hafslund are
due to security considerations. In addition, Hafslund requires a four quadrant
meter which can measure the exchange of energy in both directions. Other re-
quirements when connecting to the Hafslund 230 − 400 V network are found in
[38] which is in accordance with Forskrift om Leveringskalitet (FoL).
One important question which would arise when considering connection to the
grid is: does the PV installation result in a surplus situation of energy during
the year? If it does, it could be considered a surplus customer if Hafslund agrees
to the connection. If the PV installation does under no circumstances during the
year produce more than the building consumes, then the installation would be
considered a private installation.
3.6.2 Metering
In case the production exceeds the consumption the location of the meter must
be considered. There are many different types of meters, depending on what the
consumer needs to measure with respect to the purchasing agreement with the
electrical distributor, in this case Hafslund. Figure 3.6.1 illustrates two different
metering, net metering and gross metering. With a gross metering system the
main switch solar supply would be shifted to the point in front of the meter
measuring the solar supply production, marked with a dotted line.
A net-metering agreement would allow the customer to use all the electricity
produced by their own PV system and only be charged for any excess electricity
which is consumed. This would mean that the electricity distributor purchases
the electricity at the same rate as they are selling to the consumer [7, p.144].
The net metering schematic could or could not measure the solar production,
3.7. ECONOMICS 59
Figure 3.6.1: Net and gross metering systems in the same diagram
due to its placement after the domestic circuits. A net import meter would bill
the consumer only for the units that are imported. Using an import meter and
an export meter or a corresponding dual-element electronic meter, provides the
opportunity for the electrical distributor to pay a different tariff for the electricity
exported than the one imported. A gross metering would on the other hand
provide information concerning the solar production, this could also be supplied
by the inverter chosen. Gross metering could either consist of two meters, a
generation meter measuring the solar supply and a consumption meter measuring
the consumption or it could be one dual-element electronic meter functioning in
the same manner. Which meter set up and which metering agreement that is
to be used, should be discussed with Hafslund. This would only be necessary in
case the PV installation changes the customer status to a surplus customer.
3.7 Economics
A simple overview of the economic aspects of the final alternatives should be done
in order to get an idea of the costs of such an installation. All manufacturers of
modules and inverters should receive a request concerning the selling price of the
module or inverter used in order to give the most realistic picture of the price
situation per now. However, such prices could change in a possible next phase of
implementation and would only be valid for the time when or if such a request
was to be answered. The income aspect of the installation would depend on

whether the installation would cause the Hafslund customer to become a surplus
customer in accordance with the definition given by NVE.
Two methods which take into consideration the economic aspects of the instal-
lation are:
1. Simple payback
2. Life cycle costing
Simple payback is calculated by the following formula
C
T = (3.7.1)
S
where T is the payback period in years, C is the initial investment cost and S
is the annual cost savings of electricity that does not need to be purchased [7,
p.307].
In order to perform life cycle costing, a life cycle cost analysis is used in order
to determine the cost per kilowatt-hour for the PV system. The life cycle cost
analysis has been made in order to determine the investment cost of the equip-
ment, the operation and maintenance cost. In addition component life time and
replacement costs should be taken into consideration. To determine the cost per
kWh the following equation is used [7]:
P resent value of the system over X years

cost/kW h = (3.7.2)
Y ield (kW h) generated over X years
In this study a analysis period (X years) of 20 years has been used, since it has
been assumed a lifetime of approximately 20 years of the modules. Determining
the present value of the system over 20 years requires knowledge concerning
maintenance, repair and replacement costs during the period of analysis, which
is here set to 20 years. The PV modules and the BoS components have been
taken into account. There are small maintenance costs associated with a PV
system as a whole. The maintenance cost of the PV modules is set to be 1 %
yearly of the investment cost. With the assumed lifetime of the modules to be
approximately 20 years there would be no replacement costs. Inverters can last
10−20 years [7, p.314]. They would, however, need a repair during this period. A
repair cost of the inverters as well as the rest of the BoS components is included
every five years. The repair cost is assumed to be 1 % of the investment costs of
the BoS investment costs. In these calculations it is assumed that the inverters
are repaired every 5 years and would, therefore, not need replacing until after
the analysis period of 20 years. In a more thorough calculation in a possible next
phase of such a project, the manufacturer of the inverters and modules should
3.7. ECONOMICS 61
be consulted on how long expected life time their product has and take possible
replacement costs into account.
Two different situations that would generate an income has been considered in
these calculations:
1. The production is lower than the buildings consumption, which leads to no
energy surplus situations. The income can be considered as the savings for
the energy which is not supplied by Hafslund.
2. If the installation produces more than the building consumes, Hafslund will
then pay for the surplus energy which is produced. The income would be
the market price and a small addition.
The market price for electricity changes hourly on Nord Pool. Nord Pool or-
ganize and operate the main market place for trading spot power and is owned
by the Nordic TSOs. For simplicity it is in this study assumed that the market
price in the future will correspond to the average spot price for the last three
years with delivery in Oslo, 0.34 NOK/kWh [39]. The small addition is a re-
imbursement which at least private customers in the Haflund grid gets for the
network tariff. This addition would change, however, could be approximately
0.0425 NOK/kWh [40]. For the situation where the production is lower than the
building consumption the prices are set each year with a tariff. The tariff consists
of a fixed price and two variables depending on the power consumption peak and
the amount of energy used. Table 3.7.1 illustrates the tariff which the building
pays in 2012. This is valid for buildings with overload protection over 125 A with
a 230 V supply, and 80 A with a 400 V supply. It also includes buildings with
an expected yearly consumption above 100 000 kWh. The power variable in the
tariff is charged with regards to the highest power peak within a month.
Table 3.7.1: The tariff in 2012 for low voltage connections (230 V and 400 V) to
Hafslunds grid [41]. Summer = April - October, Winter = November - March.
Fixed price Power variable Energy variable

NOK/month NOK/kW/month øre/kWh
Summer Winter Summer Winter
415 25 74 5.25 7.40
In the calculations of life cycle costing, the inflation rate and the discount rate
is set to is 2.5 % and 6 % respectively. The inflation rate is set based on the
long term target in the monetary policy set by the Government in Norway [42].
The project discount rate should reflect the risk in the project and the expected
return of investment. The discount rate is individual and would, therefore, vary
depending on how the project is regarded with respect to these two factors. In
the handbook of socio-economic analysis of energy projects by NVE it is proposed
a discount rate of 6 − 8 % for end user initiatives [43, p.25]. 6 % is to be used
if the initiative is considered to have a clear environmental advantage, while 8 %

should be used for other initiatives. Based on this, a discount rate of 6 % has
been used in the economic calculations in this study.
Chapter 4
Simulations
This chapter documents the process of finding the final three systems providing
the best PR and energy production. Parameters which have been altered are
the modular tilt angles, pitch distance, fraction of electrical effect and multiple
inverters. A base case was used in order to compare the effect of varying these
parameters. The base case consisted of REC modules (type REC250PE) and
inverters from SMA (Sunny Boy SB 5000 TL-20). The base case system consisted
of 100 modules, 10 modules in each string connected to five inverters where each
inverter has a two MPPT inputs. The NOCT coefficient specified in accordance
with that of the REC module (29◦ C), wiring loss at 1.5 % at STC and the
soiling loss specified 50 % for January and 35 % for February and December.
The modules were placed with an edge zone of two meters and all modules had
an azimuth of 0◦ . With a module tilt of 40◦ , the default meteorological data
set and linear shading gave the following results illustrated in Figure 4.0.1 and
Figure 4.0.2. These figures also illustrates how a general simulation report is
presented in PVsyst.
As seen the base case system is simulated to produce 23.17 MWh/year with a
specific production of 927 kWh/kWp/year and a performance ratio of 77.0 %.
The losses within the system are illustrated in the loss diagram in Figure 4.0.3. It
is seen in Figure 4.0.3 that there is a gain of 22 % due to the tilt of the modules,
a loss of 8.1 % due to the near shadings calculated as linear shades, a 2.9 % loss
due to the IAM factor. The the efficiency of the PV modules are then taken into
account, following with the losses due to irradiance level, temperature, soiling,
mismatch and wiring loss. Finally the inverter losses are taken into consideration.
63
64 CHAPTER 4. SIMULATIONS
Figure 4.0.1: Main results of the base case system
Figure 4.0.2: Normalized energy production and the performance ratio of each
month, base case system
65
Figure 4.0.3: Loss diagram for the base case, from PVsyst
4.1 Fraction of Electrical Effect

Table 4.1.1 shows the main results when varying the fraction of electrical effect
from 0 % electrical effect (linear shadings) to 100 % electrical effect. The system
is otherwise unchanged from the base case, which means that the loss diagram
in Figure 4.0.3, is unchanged except from the near shading loss which is shown
in Table 4.1.1.
Table 4.1.1: The main results when varying the fraction of electrical effect. Tilt
angle: 40◦ , azimuth: 0◦ , EF: electrical effect
Produced Specific Performance Near shading

energy production ratio (PR) loss
MWh/year kWh/kWp/year % %
Linear shadings 23.174 927 77.0 -8.1
20 % EF 22.884 915 76.0 -9.3
40 % EF 22.592 904 75.0 -10.6
60 % EF 22.296 892 74.0 -11.8
80 % EF 21.999 880 73.1 -13.1
100 % EF 21.700 868 72.1 -14.3
4.1.1 Discussion
As seen in Table 4.1.1, when the fraction of electrical effect is taken into account
the near shading loss increases, decreasing the performance ratio, specific pro-
duction and produced energy. This is an expected result since a higher fraction of
electrical effect influences the production of every module in the affected string.
A 100 % electrical effect fraction represents the whole string being equally af-
fected as the shaded module in that string, as mentioned in Section 3.3.3. A
100 % electrical effect would be an incorrect representation of partial shadings
on the modules due to the bypass diodes in the modules and the module layout.
The fraction of electrical effect is set to 60 % in the next simulations since this is
considered a more realistic fraction when taking into consideration the shades on
the roof. Varying the electrical effect does, however, illustrate the range where
the shading loss lies.
Limitations to the electrical effect calculation
In order to calculate a realistic electrical effect, all modules would have to be

exactly positioned in the 3D representation with a clear notion of which modules
4.2. MODULE TILT ANGLES 67
belong to which strings, how many bypass diodes that are used in each module
etc. The creators of PVsyst have started this process with ”module layout”. In
the article for the 25th European Photovoltaic Solar Energy Conference in Spain,
2010 [29] the creator of PVsyst stated that the implementation of joining module
layout with the near shadings would be realized by the end of 2010. Nevertheless,
it is not implemented in the version used during this study (V5.54).
4.2 Module Tilt Angles

Table 4.2.1 illustrates the losses in the systems where the modular tilt angle
has been changed. Table 4.2.2 shows the main results with different module tilt
angles. Global addition in the table describes the global irradiance addition on
the plane due to the module tilt.
Table 4.2.1: Losses in % with variable module tilt angles
Module tilt 40◦ 30◦ 25◦ 20◦

Global addition 22.0 20.2 18.3 15.8
Shading loss -11.8 -8.5 -6.5 -4.6
IAM -3.0 -3.1 -3.3 -3.5
Irradiance level -3.7 -3.6 -3.6 -3.7
Temperature -1.4 -1.4 -1.3 -1.2
Soiling loss -2.4 -2.3 -2.3 -2.3
Mismatch loss -2.1 -2.1 -2.1 -2.1
Wiring loss -0.7 -0.7 -0.7 -0.7
Inverter loss -3.9 -3.9 -3.8 -3.8
Table 4.2.2: Main results with variable module tilt angles, done with the base
case and a 60 % fraction of electrical effect
Tilt Produced Specific Performance Global Shading

angle energy production ratio (PR) addition loss
◦
MWh/year kWh/kWp/year % % %
40 22.30 892 74.0 22.0 -11.8
30 22.82 913 77.0 20.2 -8.5
25 22.94 918 78.6 18.3 -6.5
20 22.87 915 80.0 15.8 -4.6
Table 4.2.3 presents the main results when the pitch distance has been increased
from 2 meters to 2.5 meters (illustrated in Figure 4.2.1). In order to reduce the
pitch distance the number of modules were reduced from 100 to 72 modules,
requiring another type of inverter in order to match the new number of modules.
In this case an Eltek inverter, type Theia 2.9 HE-t was chosen. Both the Eltek
inverter and the SMA base inverter have a maximum efficiency of 97 %.
Table 4.2.3: Increasing pitch distance between the module rows to 2.5 meters
Tilt Produced Specific Performance Global Shading

angle energy production ratio (PR) addition loss
◦
MWh/year kWh/kWp/year % % %
40 16.58 921 77.2 22 -8.7
(a) 2 meter pitch (b) 2.5 meter pitch
Figure 4.2.1: Overview of 2 meter and 2.5 meter pitch distance
4.2.1 Discussion
Investigating Table 4.2.1 reveals that there is little change of the following losses;
IAM, loss due to irradiance level, loss due to temperature, soiling loss, mismatch
loss, wiring loss and inverter loss. More considerable changes are seen in the
losses due to shading and the gain due to module tilt. Considering the difference
in shading loss of a system with a modular tilt of 40◦ and 20◦ there is a change
of 7.2 %. The gain of global irradiance on the tilted planes varies in a range of
7.6 %. Since these are the two loss components that change the most, they are
displayed with the main results in Table 4.2.2.
As seen in Table 4.2.2 the shading loss decreases with decreasing tilt angle of the
modules. This is a result of the pitch distance maintained at 2 meters while the
4.3. METEOROLOGICAL DATA SET 69
module tilt is decreased. The global irradiance addition decreases with decreasing
modular tilt, due to a larger difference from the optimal tilt of 40◦ . The specific
production and the produced energy increases with decreasing tilt angle from
a tilt angle of 40◦ to 25◦ , and decreases somewhat from 25◦ to 20◦ tilt angle.
The reason for this could be explained by adding up the gain from the global
irradiance with the shading loss. This results in a gain which increases from
10.2 %, 11.7 % and to 11.8 % at 40◦ , 30◦ and 25◦ modular tilt angles receptively.
It then decreases to 11.2 % gain at a 20◦ modular tilt. How this gain and loss is
weighted with respect to importance in the system is, however, not necessarily
the same. A difference could be assumed, since the resulting gain between the
gain and shading loss at 20◦ decreases to a gain value lower than the one at 30◦
modular tilt. This is not reflected in the specific production or the produced
energy. The possible weight difference between the two factors could also be the
reason for the increasing performance ratio with decreasing tilt angle.
Comparing the same tilt angle of 40◦ with s pitch distance of 2.5 meter (seen in
Table 4.2.3) it is seen that the global addition due to the tilt of the modules are
the same as for the 2 meter pitch distance. The shading loss and the performance
ratio for the system with 40◦ tilt angle and 2.5 meter pitch distance is approx-
imately the same as for a tilt angle of 30◦ with a 2 meter pitch distance. The
specific production for the system with 2.5 meter pitch distance has increased
as a result of both high gain of global irradiance due to the tilt of the modules
and the relatively low shading loss. However, since this system had to remove
modules in order to maintain an edge zone of 2 meter the kWp of the system
has decreased from 25 kWp to 18 kWp and the produced energy has decreased
from 22.30 MWh/year to 16.58 MWh/year. When dimensioning the PV system
for a performance ratio above 80 % and having limited amount of space on the
roof it is, therefore, considered more functional to reduce the tilt angle of the
modules and not operate with an optimal tilt angle than increase the pitch and
reduce the number of modules further. The following simulations will, therefore,
be performed with a system of a modular tilt angle of 20◦ and a pitch distance
of 2 meter (and an electrical effect of 60 %).
4.3 Meteorological Data Set

Table 4.3.1 shows the main results of a year of production with the two different
meteorological data sets. Figure 4.3.1 shows the normalized productions per
month for both meteorological data sets. The purple in Figure 4.3.1 illustrates
the collection loss (the PV array losses), the green part illustrates the system loss
(mainly the inverter loss) while the red illustrate the produced useful energy (the
inverter output). The collection loss for the default case is 0.53 kWh/kWp/day
compared to 0.51 kWh/kWp/day for the new case. The system loss is equal in
the two cases with 0.1 kWh/kWp/day. The inverter output for the default case
is 2.51 kWh/kWp/day compared to 2.44 kWh/kWp/day for the new case. Table
4.3.2 shows the losses in the systems with the two different meteorological data
sets.
Table 4.3.1: Main results with the default and the new meteorological data set
(Meteo2), 20◦ tilt angle, 60 % electrical effect
Meteo Produced Specific Performance

set energy production ratio (PR)
MWh/year kWh/kWp/year %
Default 22.87 915 80.0
Meteo2 22.28 891 80.1
(a) Default meteorological data set (b) New meteorological data set
Figure 4.3.1: The normalized productions (per installed kWp): nominal power
25.00 kWp for 20◦ tilt angle, 60 % electrical effect
4.3.1 Discussion
As seen in Table 4.3.1 both the produced energy and the specific production
per year are higher for the default meteorological data set. It is also possible to
see the same tendency in Figure 4.3.1, keeping in mind that the ordinate axis is
larger for Figure 4.3.1a than for Figure 4.3.1b.
The performance ratio for the simulation with the new meteorological data set is
0.1 % higher than the simulation with the default data set. In order to explain
this, the losses in Table 4.3.2 has to be taken into consideration. The inverter
loss, wiring loss, mismatch loss, soiling loss, irradiance level loss and loss due
to IAM has only changed by 0.1 % or none at all and will, therefore, not be
discussed further. The losses due to temperature, shading and the gain due
to increasing global irradiance due to the tilt of the modules display a more
considerable change. The temperature loss is decreased to half from 1.2 % to
4.4. INVERTERS 71
Table 4.3.2: The losses in the system with the default and the new meteorological
data set, 20◦ tilt angle, 60 % electrical effect
Loss, % Default Meteo2

Global addition 15.8 17
Shading loss -4.6 -4.9
IAM -3.5 -3.6
Irradiance level -3.7 -3.8
Temperature -1.2 -0.6
Soiling loss -2.3 -2.3
Mismatch loss -2.1 -2.1
Wiring loss -0.7 -0.7
Inverter loss -3.8 -3.9
0.6 %. This can be explained with the yearly average of temperature in the two
meteorological sets which are 7.04◦ C and 5.7◦ C for the default meteorological
data set and the new data set respectably. As explained in Section 3.1.1, the
modules function with a higher voltage and efficiency when the temperature is
lower. Therefore, the loss is decreased with the new data set. The shading loss
increases with 0.3 % from the default to the new meteorological data set. The
global irradiance in the new meteorological data set is 37 kWh/m2 less than the
global irradiance in the default set. However, the gain due to the tilt of the
modules has increased with 1.2 %. The two meteorological datasets are very
similar comparing the global irradiance except for May, July and August where
the default dataset has a higher global irradiance. Comparing the gain of more
irradiance by tilting the modules results in a higher gain for the new dataset in
all the months except for May, July and August. The reason for this is that the
new meteorological data set has overall a lower global irradiance and still obtain
higher gain in 9 out of 12 months (for further information see Appendix G).
4.4 Inverters
4.4.1 Homogeneous System
A homogeneous PV system is where all the module and inverters types, module
tilt and azimuth direction are the same for the whole system. Table 4.4.1, Table
4.4.2 and Table 4.4.3 show the main results from simulation of a homogeneous
PV system with different types of inverters with the two types of polycrystalline
modules and the one type of monocrystalline module. The tables only display
the inverters which matched the PV system such that a minimum of 90 modules
were included. Table 4.4.4 shows the main results and the main difference from
the base case with the difference of using REC modules in one simulation and
ITS modules in the other simulation. Figure 4.4.1 and Figure 4.4.2 illustrate the
corresponding I-V curve for different amounts of irradiance for the REC module
and the ITS module respectively.
Table 4.4.1: The main results from simulation with inverters from Eltek, SMA
and Danfoss with the polycrystalline REC module: REC250PE
Inverter Produced Performance Number Series/

type energy ratio (PR) of Strings/
MWh/year % modules Inverters
SMA SunnyBoy 22.28 80.1 100 10/10/5
Eltek Theia 4300 TL 22.81 80.4 102 17/6/6

Theia 4800 TL 22.83 80.5 102 17/6/6
Theia 5300 TL 22.40 80.6 100 20/5/5
Theia 2.0 HE-t 22.15 80.5 99 11/9/9
Theia 2.9 HE-t 21.46 80.4 96 12/8/8
SMA Tripower 8000TL 22.92 80.8 102 17/6/3
Danfoss TLX Pro 6k 23.15 80.1 104 13/8/4
TLX Pro 8k 22.80 80.4 102 17/6/3
TLX Pro 12.5k 22.75 80.2 102 17/6/2
4.4.2 Heterogeneous System
A heterogeneous system of different Eltek inverters was simulated and resulted

in the main results displayed in Table 4.4.5. The modules have the same 20◦
modular tilt and azimuth direction the only difference is, therefore, that more
than one type of inverters are used in the PV system. The roof was divided into
three fields where each field have their inverter type. Fields A, B and C are as
displayed in Figure 4.4.3.
4.4.3 Discussion
Considering the homogeneous simulations done for the polycrystalline alterna-

tives it is seen that changing the base inverter (Sunny Boy) with one of the other
high efficiency inverters gave a higher performance ratio in most cases. The base
4.4. INVERTERS 73
Table 4.4.2: The main results from simulation with inverters from Eltek, SMA
and Danfoss with the polycrystalline ITS module: EcoPlus PolyUp, 250W

SMA SunnyBoy 21.92 78.9 100 10/10/5
Eltek Theia 4300 TL 22.44 79.2 102 17/6/6

Theia 4800 TL 22.46 79.2 102 17/6/6
Theia 5300 TL 22.04 79.3 100 20/5/5
Theia 2.9 HE-t 22.84 79.0 104 13/8/8
SMA Tripower 8000TL 22.55 79.5 102 17/6/3
Danfoss TLX Pro 6k 22.77 78.8 104 13/8/4
TLX Pro 8k 22.43 79.1 102 17/6/3
TLX Pro 12.5k 22.77 78.8 104 13/8/4
inverter was undersized for the monocrystalline alternative and could, therefore,
not be used in a simulation. It is, however, seen from the monocrystalline case
that the performance ratio is higher than both of the other polycrystalline cases
even when the same inverter is used. This could be explained by the high mod-
ule efficiency of the monocrystalline modules. Due to the high power of the
monocrystalline modules it is also seen that approximately only half of the in-
verters within the chosen series of inverters were possible to use when taking into
consideration the voltage, power and current conditions where the modules and
the inverters should match. Sizing the systems such that each string should have
its own MPPT input also limited the amount of possible inverters, since several
inverters have a set up which allows parallel couple strings such that the parallel
coupling has a MPPT instead of each string.
Comparing the base cases when using REC modules and ITS modules it is seen
that the performance ratio in the ITS simulation is 78.9 % while 80.1 % for the
REC simulation. This can be explained by Table 4.4.4 and the I-V curves of the
two modules, illustrated in Figure 4.4.1 and Figure 4.4.2. The two simulations are
identical except for the inverters and the module types, there are equally many
modules in series, number of strings and number of inverters. The reason for the
deviation of performance ratio is due to the array loss caused by a reduction in
the irradiance level, which differs with 3.3 %. As explained in Section 3.4.4 this
loss is caused by a reduction in the nominal efficiency when the irradiance level
falls below 1000 W/m2 . According to the technical data on the two modules,
Table 4.4.3: The main results from simulations with inverters from Eltek, SMA
and Danfoss with the monocrystalline SunPower module: SPR-327NE-WHT-D

SMA SunnyBoy
Eltek Theia 2.0 HE-t 30.71 81.2 104 8/13/13

SMA Tripower 8000TL 31.06 82.1 104 13/8/4
Danfoss TLX Pro 6k 29.60 81.4 100 10/10/5
TLX Pro 8k 30.89 81.7 104 13/8/4
Table 4.4.4: Main results and main differences from base case simulations with
the polycrystalline REC and ITS modules
Produced Specific Performance Irradiance

energy production ratio (PR) level loss
MWh/year kWh/kWp/year % %
REC base 22.28 891 80.1 -3.8
ITS base 21.92 877 78.9 -7.1
the REC module has a module efficiency of 15.1 % while the ITS module has a
module efficiency between 15.2 − 15.8 %. It is also seen from Figure 4.4.1 and
Figure 4.4.2 that the ITS module has a higher MPP than the REC module for
irradiations of both 1000 and 800 W/m2 . However, when the incident irradiation
decreases to 600 W/m2 and below, the REC module has the higher MPP.
Concluding from the results of the simulation the irradiance level is most often in
the lower range of measured irradiance in W/m2 , when disregarding the influence
on the I-V curve with changing temperatures. The mean annual global horizontal
irradiance in Oslo and Ås estimated from cloudiness during 1931 − 1960 was 109
and 104 W/m2 [44] which corresponds to 954.84 kWh/m2 and 911.04 kWh/m2 .
At least for Oslo these numbers correspond nicely to the value in the new mete-
orological data set.
The maximum produced energy per year for the heterogeneous case is 22.826 MWh
(102 modules, corresponding to 25.5 kWp), while for the homogeneous case it
is 22.40 MWh (100 modules, corresponding to 25 kWp). The performance ratio
of the heterogeneous case would lie between 80.5 and 80.6 %, while it could be
4.4. INVERTERS 75
Figure 4.4.1: The REC module I-V curve with different irradiation, from PVsyst
80.6 % as for the homogeneous case. As seen, there is not a considerable dif-
ference between these two cases. Nevertheless, it was chosen not to go further
with other heterogeneous cases due to the limited amount of inverters which were
chosen for this study. The inverters from SMA and Danfoss which are chosen
are e.g. oversized for field A. A wider range of inverters should be taken into
account in order to optimize the production from each sub-field. An advantage
of designing a heterogeneous case where the field is divided into smaller fields on
account of physical location is that it could increase the practicality when cabling
and connecting the modules into strings.
Since the simulation with the ITS modules revealed that these modules most
likely are not as well suited for the irradiation levels in Skøyen an alternative
including ITS modules will not be considered further. On account on the high-
est performance ratio, the Eltek inverter Theia 5300 TL and the SMA inverter
used in the homogeneous REC case are taken further. The SMA inverter in the
monocrystalline system is also taken further on account of the high performance
ratio.
Figure 4.4.2: The ITS module I-V curve with different irradiation, from PVsyst
Figure 4.4.3: Overview of the roof divided into three fields. Field A = blue, field
B = green, field C = yellow, black objects = shading objects
4.5. SHADE ANALYSIS AND MODULE LAYOUT 77
Table 4.4.5: Main results when simulating a heterogeneous system with different
Eltek inverters with the polycrystalline REC module

Field A Theia 2.0 HE-t 2.24 80.6 10 10/1/1
Field B Theia 4300 TL 16.1 80.4 72 18/4/4

Theia 4800 TL 16.11 80.5 72 18/4/4
Field C Theia 4800 TL 4.476 80.5 20 20/1/1

Theia 5300 TL 4.475 80.5 20 20/1/1
4.5 Shade Analysis and Module Layout

The three final alternatives are systems consisting of a total of 100, 102 and 104
modules. In order to assign all modules to their string, four and two modules
would have to be removed from the original 104 module maximum on the roof.
In order to decide which redundant modules should be removed, the results from
the shade analysis was taken into account.
Table 4.5.1: The main result from the shade analysis, displaying the modules
with the longest duration (in hours) of shade from the three fields. * marks
modules that are fully shaded a period during the day
Module no. Shade duration

A-11 3.0*
A-8 2.5
B-18 4.0*
B-19 2.5*
B-28 3.0
C-17 2.5*
C-19 2.5*
Table 4.5.1 presents the main results from the shade analysis of the field with
20◦ modular tilt and 2 meter pitch distance. For a fully detailed analysis see Ap-
pendix C. For the polycrystalline alternative with Eltek inverters there are four
redundant modules. There are 20 modules per string in this alternative. For the
second polycrystalline alternative with SMA inverters there are two redundant
modules which has to be removed. For the last alternative, the monocrystalline
alternative with SMA inverters there is no need to remove modules. Figure 4.5.1
to Figure 4.5.3 illustrate the module layout.
4.5.1 Discussion
For the first polycrystalline alternative with Eltek inverters four modules were
removed: B-18, B-28, A-11 and A-8. Module B-18, B-28 and A-11 were the
modules with longest duration of occurring shade, while the fourth module had
several alternatives which could be removed. Modules A-11, B-19, C-17 and C-19
were candidates which could be removed. Since field C consists of 20 modules
with a similar shading pattern, it was considered best if these were to be placed
in the same string. Modules C-17 and C-19 were, therefore, not eliminated.
Module A-11 was simulated without being fully shaded during the day and the
string from field A would have to connect to field B due to the low amount of
modules on field A. Removing B-10 would give a module layout which perhaps
is not practical when considering the coupling of the modules. Module A-8 was,
therefore, the fourth module which was removed.
For the second polycrystalline alternative with the REC modules and the SMA
inverters only two modules were removed based on the same arguments when
removing four: B-18 and A-11. Module B-28 was considered removed. However,
A-11 was chosen since this module is fully shaded during the day. Removing
module A-11 would give a longer coupling distance from field A to field B. A more
thorough analysis of which modules that are to be removed could be reconsidered
in a possible next step in this project.
Figure 4.5.1: Module layout of the REC alternative with Eltek inverters, 100
modules in total
Figure 4.5.2: Module layout of the REC alternative with SMA inverters, 102
modules in total
Figure 4.5.3: Module layout of the SunPower alternative with SMA inverters,
104 modules in total
4.6 Cable Sizing

As mentioned, cable sizing is not done explicitly in PVsyst. However, it is based
on the resulting simulations done in the program. Table 4.6.1 to 4.6.4 show
the minimum cross sectional area (CSA) of proposed DC cable paths for the
three alternatives. Only the DC cables have been sized in this study. The loss
in percent is of the standard size cable which is found closest in size with the
calculated CSA. The inverters are assumed placed in the stair house on the roof.
The proposed cable paths and more detailed calculations of the cable lengths are
found in Appendix E.
Table 4.6.1: Minimum CSA (mm2 ) and loss for the polycrystalline alternative
with Eltek inverters
Eltek String 1 String 2 String 3 String 4 String 5

Minimum CSA 2.96 3.25 2.53 2.43 2.24
Closest CSA 4 4 4 4 4
Loss 1.11 % 1.22 % 0.95 % 0.91 % 0.84 %
Table 4.6.2: Minimum CSA (mm2 ) and loss for the polycrystalline alternative
with SMA inverters
SMA String1 String2 String 3 String 4 String 5 String 6

Minimum CSA 3.89 3.42 3.31 2.80 2.77 2.19
Closest CSA 4 4 4 4 4 2.5
Loss 1.46 % 1.28 % 1.24 % 1.05 % 1.04 % 1.31 %
Table 4.6.3: Minimum CSA (mm2 ) and loss for the five first strings of the
monocrystalline alternative
SunPower String1 String2 String 3 String 4 String 5

Minimum CSA 1.69 1.63 1.45 1.33 1.10
Closest CSA 2.5 2.5 1.5 1.5 1.5
Loss 1.01 % 0.98 % 1.45 % 1.33 % 1.10 %
4.6.1 Discussion
The CSA depend on the cable path and the amount of modules that are to
be connected in series. The minimum cross section of the two polycrystalline
4.6. CABLE SIZING 83
Table 4.6.4: Minimum CSA (mm2 ) and loss for the three last strings of the
monocrystalline alternative
SunPower String 6 String 7 String 8

Minimum CSA 1.21 1.20 1.03
Closest CSA 1.5 1.5 1.5
Loss 1.21 % 1.20 % 1.03 %
alternatives is in the same range since they have 20 and 17 modules in series
(see Table 4.6.1 and 4.6.2). The monocrystalline alternative with 13 modules
in a string has the smallest CSA (illustrated in Table 4.6.3 and 4.6.4). Both
the SunPower and REC modules are supplied with 1 − 2 meters of solar cables
and MC4 connectors (se Appendix F). These connectors match solar cables with
a CSA of 4 mm2 . Using higher cross section in the cables would diminish the
ohmic loss in the system. However, higher cross section would result in higher
price. These calculations apply only for one proposed cable path for each of the
alternatives, altering the cable paths would alter these calculations. If all of the
cables were to have a cross section of 4 mm2 , all strings in all alternatives would
have a loss below the defined maximum of 1.5 % in PVsyst.
Chapter 5
Customer Status and

Economics
In order to determine whether the company in the building can be classified as

a surplus customer, the production from the PV installation has to be compared
with the consumption within the building. It is possible to compare production
and consumption since the building has a meter which measures the energy and
power each hour and since PVsyst generates hourly plots of the production of
simulated system. Furthermore, the results of the economic calculations are
presented in this chapter.
5.1 Production and Consumption

The production in kWh from the three alternatives is drawn per month in Fig-
ure 5.1.1. The yearly production from the polycrystalline alternative with Eltek
inverters is 22.4 MWh, 22.9 MWh for the polycrystalline alternative with SMA
inverters while the yearly production is 31 MWh for the monocrystalline alter-
native. For further details see Appendix H. Table 5.1.1 presents the production
as percentage of the energy use within the building. The consumption figures
are from the year 2011. Consumption vary on account of several factors, for
example how many people that use the building, the weather situation etc. The
consumption figures from 2011 were chosen since this is the last full year with
consumption data that will visualise to some extent the relation between energy
production and consumption.
85
86 CHAPTER 5. CUSTOMER STATUS AND ECONOMICS
Figure 5.1.1: Simulated energy production per month of the three alternatives
Table 5.1.1: The energy production as percentage of the total energy use in the
building, 2011
REC REC SunPower

Eltek SMA SMA
January 0.08 % 0.07 % 0.11 %
February 0.02 % 0.02 % 0.42 %
March 1.06 % 1.08 % 1.45 %
April 1.71 % 1.76 % 2.38 %
May 1.78 % 1.82 % 2.47 %
June 1.49 % 1.53 % 2.08 %
July 1.44 % 1.48 % 2.02 %
August 1.03 % 1.05 % 1.43 %
September 0.89 % 0.91 % 1.23 %
October 0.51 % 0.52 % 0.69 %
November 0.19 % 0.19 % 0.26 %
December 0.07 % 0.07 % 0.10 %
Yearly 0.92 % 0.94 % 1.27 %
5.1.1 Discussion
Both the energy and power production have been compared to the energy and
power consumption in the building in order to establish the effect of the PV
5.1. PRODUCTION AND CONSUMPTION 87
Table 5.1.2: The energy production as percentage of the energy used for cooling
in the building, 2011
REC REC SunPower

Eltek SMA SMA
January 14.04 % 13.43 % 19.34 %
February 62.06 % 14.91 % 83.50 %
March 180.46 % 184.73 % 247.00 %
April 106.14 % 108.82 % 147.21 %
May 25.90 % 26.57 % 36.06 %
June 9.23 % 9.46 % 12.89 %
July 7.31 % 7.49 % 10.20 %
August 6.11 % 6.26 % 8.51 %
September 8.26 % 8.46 % 11.43 %
October 6.26 % 6.39 % 8.48 %
November 3.42 % 4.70 % 4.70 %
December 2.64 % 2.57 % 3.67 %
Yearly 10.83 % 11.08 % 15.02 %
installation. The customer status would only change if there is an export of

energy out on the grid, which can been observed when evaluating the power
production and consumption. Evaluating the energy production first, it is seen
from Figure 5.1.1 that the monocrystalline alternative produces most energy.
The production from the polycrystalline alternatives show that the SMA inverter
produces the most, as mentioned in Section 4.4. The SMA inverter has the
highest efficiency, which could be on account of the galvanic isolation that the
Eltek inverters provide. An average household in Norway uses approximately
20.4 MWh a year, where approximately 16 MWh/year is for electricity [1]. Table
5.1.1 presents the energy production of the PV installation in percentage of the
energy consumption in the building. As seen, The production per year is less
than 1 % in the polycrystalline cases and about 1.3 % in the monocrystalline
case. Since the building is a six storey building, it has a considerable energy
consumption. For the monocrystalline case it is at most covering approximately
2.5 % of the buildings consumption (in May).
As this is a commercial building its weekly consumption profile would be high
during the weekdays and low during the weekends. The most probable time
when production would exceed the consumption would, therefore, be during the
weekend in summer. Evaluating the average and maximum power production in
Table 5.1.3 and Table 5.1.4 it is seen that the consumption in the building does
not go below 78 kW, while the highest power production for the PV installation
Table 5.1.3: The average and maximum power produced during the simulated
year
REC/Eltek REC/SMA SunPower/SMA

Average Max Average Max Average Max
kW kW kW kW kW kW
January 0.67 3.15 0.68 3.21 0.96 3.81
February 2.20 9.38 2.28 9.63 2.98 12.57
March 5.58 17.66 5.78 18.19 7.64 24.19
April 7.00 18.88 7.39 19.43 9.86 26.41
May 7.24 19.73 7.53 20.31 10.13 27.80
June 6.82 19.68 7.27 20.26 9.72 28.00
July 6.25 20.21 6.52 20.81 8.85 28.72
August 6.20 19.95 6.48 20.54 8.69 28.41
September 5.52 17.24 5.75 17.75 7.72 24.22
October 3.88 16.88 4.03 17.38 5.28 23.07
November 1.99 9.27 2.11 9.52 2.75 12.33
December 0.79 3.12 0.78 3.18 1.11 4.41
is approximately 29 kW (for the monocrystalline alternative). Consequently the

PV installation would not change the customer status to a surplus customer. As
seen in Table 5.1.4 the minimum power during a year is within a range of 21 kW,
while the maximum power during a year is in a range of 195 kW. The minimum
load in the building is, therefore, much more stable than the maximum load
which has its power peak during the summer. Analysing the consumption of the
building (Appendix H) shows that there is an increase in the main consumption.
However, the cooling measurements show that this is the post that is increasing
the most during summer, which could only be considered natural. Comparing
the energy production with the cooling consumption during a year it is seen from
Table 5.1.2 that the production from the PV installation cannot cover the cooling
consumption in the building either. The simulated production would cover the
cooling demand during March and April 2011, since little cooling is used and
the production is increasing. When reaching June the production can only cover
about 7 − 10 % of the cooling demand.
5.2. ECONOMICS 89
Table 5.1.4: The minimum and maximum power consumed for each month in
2011
Min Max
kW kW
January 81 369
February 84 378
March 81 384
April 81 435
May 87 480
June 93 537
July 99 510
August 99 564
September 96 471
October 84 462
November 84 420
December 78 429
5.2 Economics
Table 5.2.1 shows the approximate evaluation of the investment cost for the three
alternatives. An enquiry regarding prices were sent out to REC, SunPower, Eltek
and SMA. The module price for the REC modules were obtained from REC and
the inverter price for the Eltek inverter was provided from Eltek. The module
price for the SunPower modules were obtained from Photon International [45]
as well as the price for the SMA inverters [46]. The remaining BoS costs for
the different alternatives were considered to be equal and are based on previ-
ously calculated figures from Multiconsult. The difference in BoS costs for the
monocrystalline alternative is because of the price being in cost/kWp, which de-
pends on the efficiency of the modules used. For an equal mounting system only
with modules with higher efficiency and higher Wp would have BoS cost/kWp
which would be lower since the kWp of the entire system would be higher.
The power variable in the tariff paid today is calculated by using the maximum
load power peak within the building during one month. In order to be able to
reduce the power variable the production from the PV installation would have
to reduce the load power peak in the building at that exact time and day when
the load peak occurs. Since the PV installation does not produce more than the
building consumes during a year (in this case the consumption in 2011) the income
generated from the PV installation is the same as saved costs from importing the
energy from Hafslunds grid (seen in Table 5.2.2). A theoretical alternative was
Table 5.2.1: Investment cost for the three alternatives. Rate per May 11th, 2012:
7.62 EUR/NOK
REC REC SunPower

Eltek SMA SMA Unit
Module price 750 750 1140 e/kWp
Length 1.665 1.665 1.559 m
Width 0.991 0.991 1.046 m
Power 250 250 327 Wp
Efficiency 15.15 15.15 20.05 %
Inverter 297.14 210.00 210.00 e/kWp
Mounting system 250.00 250.00 188.90 e/kWp
Montage 200.00 200.00 151.12 e/kWp
DC cables 214.00 214.00 161.69 e/kWp
AC cables 131.20 131.20 99.13 e/kWp
Communication 40.00 40.00 30.22 e/kWp
Total BoS cost 1 132.34 1 045.20 841.06 e/kWp
Module + BoS cost 1 882.34 1 795.20 1 981.06 e/kWp
14 347.12 13 682.93 15 099.56 NOK/kWp
made illustrating the income if the energy production was sold for market price,
which is illustrated in Table 5.2.3. The simulated results from the three PV
alternatives represent the production based on the meteorological data set which
should represent a normal year. Since the meteorological data set represents
normal weather it would not be realistic to compare the production data with
the exact time and dates when a load power peak occurs in the building. During
2011 the load power peaks occurred between 11:00 and 15:00 (see Appendix H).
The PV modules produce the most within this time range. In Table 5.2.2 there is
presented a best and a worst case scenario with respect to reducing the maximum
power production in the building. The best case scenario is considering that
the total amount of the maximum power production peak that month reduces
the maximum load power peak that month. The scenario where the minimum
reduction in effect is considered is when all the simulated power measurements
during a day is averaged. And then averaging these values to get one value for
each month. The value for each month is then the reduction in the maximum
load power peak occurring that month. The real reduction in load power peaks is
considered to be located in between the minimum and maximum scenarios taken
into account in this study.
Table 5.2.4 illustrates a present value calculation over 20 years. The same as-
5.2. ECONOMICS 91
sumptions for duration of the analysis period, inflation rate, discount rate, etc.,
made for the life cycle costing are used in this calculation. For more details see
Appendix I.
Table 5.2.2: Calculated income (NOK) for a worst (minimum) and a best case
scenario (maximum) when considering the power production
Energy Power Total

Min. Max. Min. Max. Unit
REC/Eltek 1 243 1 904 6 465 3 147 7 708 NOK
REC/SMA 1 271 1 985 6 647 3 256 7 918 NOK
SunPower/SMA 1 722 2 649 8 906 4 370 10 627 NOK
Table 5.2.3: The theoretical income if the total amount of energy was sold at
spot price on Nord Pool
Income (NOK)
REC/Eltek 7 648
REC/SMA 7 826
SunPower/SMA 10 605
Table 5.2.4: Main results of the present value calculations
REC/Eltek REC/SMA SunPower/SMA

Best Worst Best Worst Best Worst
Investment 358 678 358 678 348 915 348 915 513 506 513 506
Maintenance 23 595 23 595 21 591 21 591 43 769 43 769
Repair 5 970 5 970 5 621 5 621 6 032 6 032
Income 114 163 46 607 117 284 48 227 157 411 64 733
-274 080 -341 636 -258 842 -327 899 -405 895 -498 573
5.2.1 Simple Payback
Table 5.2.5 presents the payback time for the two scenarios with the tariff savings
and the payback time if all the produced energy were to be sold in the market.
Table 5.2.5: Payback time in years when using the simple payback method
Best case Worst case Market price

Years Years Years
REC/Eltek 46.5 114.0 46.9
REC/SMA 44.1 107.2 44.6
SunPower/SMA 48.3 117.5 48.4
5.2.2 Life Cycle Costing
Table 5.2.6 presents the cost per kWh for the three alternatives, when the yield is
assumed degraded 13.3 % during the 20 year analysis period (0.75 % degradation
each year).
Table 5.2.6: The cost per kWh of the three alternatives
NOK/kWh
REC/Eltek 0.93
REC/SMA 0.88
SunPower/SMA 0.97
5.3 Discussion
There are limited investment figures based on previous calculations from (similar)
PV installations in Norway. The prices for the BoS components excluding the
inverters in Table 5.2.1 are, therefore, figures based on previous calculations from
other European countries. The figures of the DC cables are e.g. from a German
roof installation of 60 kWp. The figures used in this table are, therefore, highly
uncertain. In the cases where an offer has been made it should be kept in mind
the fluctuations in the solar market could cause the offer to change in a later
stage. The figures do, however, give an impression of what the costs could be for
these alternatives, which was the intention. Evaluating the calculated income for
a worst and a best case scenario (Table 5.2.2) it is seen that with the current tariff
with Hafslund there is most to save in reducing the power variable. Comparing
5.3. DISCUSSION 93
with the theoretical income where all of the energy is sold on the spot market it
is seen that the income here incidentally matches the best case scenario in the
current tariff agreement. As seen in Table 5.2.4 there is a negative present value
for all of the alternatives with the current tariff agreement. This is when using
the tariff of 2012 and considering it to be fixed only to be adjusted by inflation.
As seen in the table, it is the polycrystalline alternative with SMA inverters that
results in the least negative present value of the alternatives. There are a lot of
variables influencing the electricity market, amongst others the global electricity
market which could possibly increase profitability of the PV alternatives. Na-
tional regulations could also change and subsidies for PV could be given. As it
is today there are no subsidies for PV installations in Norway. Disregarding pos-
sible political changes as subsidies, it is seen that in the worst cases the income
during the 20 years barely covers the maintenance and repair costs.
The simple payback method is inaccurate since it does not take into account the
maintenance, repair costs, inflation and discount rate. However, it gives an idea
of the payback time, which is above the assumed lifetime of the modules (20
years), seen in Table 5.2.5. During 50 years, some reinvestment costs would have
to be taken into consideration, which would then further increase the payback
time. As seen there is a variation of 50 − 70 years depending on alternative and
whether the income is a best or worst case scenario. The theoretical income
alternative, selling the energy on the spot market, is comparable to the best
case income scenario. The life cycle costing method takes into consideration the
factors that the simple payback does not and, therefore, give a more realistic view
of the economics in the alternatives. Comparing Table 5.2.6 to the average spot
price of energy delivered in Oslo the last three years of 0.34 NOK/kWh (44.8
e/MWh), it is seen that the energy from the PV installation is approximately
triple the spot price. Comparing with the energy variable in the Hafslund tariff,
which is on average 0.06 NOK/kWh, would not be correct since this is only one
variable within the tariff that results in a part of the overall energy price which
is paid today.
Chapter 6
Further Work
This thesis should be considered a preliminary study of the possibilities of in-

stalling a PV system on a commercial building, in Oslo. In this thesis, there has
been an overall focus on the electrical aspects of PV systems. There are, there-
fore, electrical aspects which could be investigated further as well as mechanical
considerations that also should be taken into account in possible further studies
of this installation. For example, the wind pressure should be calculated when
evaluating the modular angle tilt and the edge zone of 2 meters. If ballasted
mounting systems were to be chosen for the installations, an investigation should
be performed on how much weight the roof holds. With regards to the electrical
aspects it would be interesting to simulate the systems with different types of
modules and inverters. It could be investigated further the existing circuits in
the building and evaluating if the connection point could be shifted from the
main switchboard to one of the circuits providing electricity to the top floors.
Other interesting aspects would be fire safety when having a PV installation on
the roof, and protection equipment and settings.
In this thesis only the consumption of 2011 was compared with the production of
the PV installations. There is complete yearly data for the consumption in 2009
and 2011 which could be considered if a more thorough study should take place.
A comparison study of similar PV installations (with regards to installed Wp)
would also be of use in order to evaluate the difference in production due to me-
teorological conditions (location). Further, it would be interesting to investigate
the production and consumption trends and the system which is already built in
the Hafslund grid and match this with a simulation done in PVsyst in order to
see the deviation. Evaluating the system in this study in a different simulation
program could shed light on other aspects which are not considered in PVsyst.
With regards to economics it would be interesting to examine the socio-economic

benefits of a PV installation and including these figures into the calculations.
95
96 CHAPTER 6. FURTHER WORK
Furthermore, it would be interesting to investigate subsidies in Norway and the

effect green certificates would or could have on a PV installation.
In case this study is taken further it should be considered to invest in mea-
surement equipment such as reference cells or pyranometers. A pyranometer
measures irradiance (global and diffuse irradiation can be measured) and could
be placed with the same tilt angle as the modules or placed horizontally in order
to evaluate the horizontal irradiance. Such measuring equipment could be used
after some time (months or years) in order to evaluate the simulation results
provided by PVsyst (with the measured irradiance and temperature) with the
output and actual production from the installation.
Chapter 7
Conclusion
The optimum tilt angle of the fixed modules considering the annual yield of a
system is 40◦ at Skøyen, Oslo. However, due to the limited amount of space on the
roof it was found that the tilt angle resulting in both an acceptable performance
ratio and production was 20◦ . Designing the distance between modular rows
such that there would not occur shading from other modular rows during spring
equinox resulted in a pitch distance of approximately 2 meters. Increasing the
pitch distance further would increase the performance ratio somewhat, except
result in reducing of the total amount of modules and, therefore, also reducing
the production and the installed capacity of the installation. The final installation
oriented towards geographical south with an edge zone of 2 meters has in this
study a capacity of 104 modules on the roof. With the choice of three inverter
series to match the maximum amount of modules and three types of modules,
the three alternatives which provided the best production and performance ratio
were:
1. A polycrystalline alternative with REC modules and Eltek inverters
2. A polycrystalline alternative with REC modules and SMA inverters
3. A monocrystalline alternative with SunPower modules and SMA inverters
Since Norway is located so far north the incident irradiation will, therefore, not
be very high. The choice of modules with a higher MPP at lower irradiance is,
therefore, more suited here than modules with high MPP at the highest incident
irradiation. The REC modules were on account of this chosen for both the
polycrystalline alternatives instead of one ITS polycrystalline alternative.
The three final alternative PV installations produce 22.4, 22.9 and 31 MWh/year,
which is more than an average household in Norway consumes during a year
(20.4 MWh/year). The monocrystalline alternative produces approximately 1.5
times that value. However, this production is not enough to cover the consump-
97
98 CHAPTER 7. CONCLUSION
tion of a six storey commercial building. The production from the three systems
only covers about 1 % of the total energy consumption in the building. The
production would have covered approximately 11 % of the cooling load during
2011. On account of the high consumption in the building, there are no situations
where the installation produces more power than the building uses (consumption
of 2011). The customer status will, therefore, not change to surplus customer on
account of the PV installation on the roof.
With the assumptions and simplifications done regarding the economic situations
investing in a PV installation is found unprofitable. Using the simple payback
method, it would take about 50 years for the system to pay itself back. The
cost per kWh is calculated to be about triple the spot price on electricity. The
preferred alternative to proceed with, would with the information and the as-
sumptions of this study, be polycrystalline modules with SMA inverters. This
alternative produces most of the two polycrystalline alternatives and has a lower
investment cost. The monocrystalline alternative would be the preferred system
to proceed with when regarding the production, however, not with regards to
costs.
References
[1] Magnussen I.H. Killingland, M. and D. Spilde. Energibruk, energi-

bruk i fastlands-norge. Report 9/2011, Norges vassdrags- og en-
ergidierktorat, Middelthunsgate 29, Postboks 5091 Majorstua, 0301
Oslo, May 2011. http://www.nve.no/Global/Publikasjoner/Publikasjoner%
202011/Rapport%202011/rapport9-11.pdf.
[2] U.S. Energy Information Administration. International energy out-

look 2011. http://205.254.135.7/forecasts/ieo/pdf/0484%282011%29.pdf,
September 2011. Accessed: June 5th, 2012.
[3] Heck S. Aanesen, K. and D. Pinner. Solar power: Darkest before dawn. Can
be accessed at http://www.mckinsey.com/Insights, May 2012.
[4] F. Drevon. Solar power (pv) covered half of ger-
manys energy demand. http://www.tu.no/energi/2012/05/29/
solkraft-dekket-halvparten-av-tysklands-energibehov, May 2012. Ac-
cessed: June 5th, 2012.
[5] F. Drevon. This dwelling has an energy surplus. http://www.tu.no/bygg/
2012/03/19/dette-bolighuset-gar-i-pluss, March 2012. Accessed: May
5th, 2012.
[6] J. Seehusen. Kraftbygg skal sette verdensrekord. http://www.tu.no/bygg/
2012/03/16/kraftbygg-skal-sette-verdensrekord, March 2012. Accessed:
June 6th, 2012.
[7] B. Neill S. Garrett S., McLean and Stapleton G. Grid-Connected PV Systems
Design and Installation. Global Sustainable Energy Solutions Pty Ltd, 2010.
[8] The German Solar Energy Society. Planning and Installing Photovoltaic
Systems - A Guide for installers, architects and engineers. Number ISBN:
1-84407-131-6. James and James Ltd, 2005.
[9] S. Hegedus and A. Luque. Handbook of Photovoltaic Science and Engineer-
ing. Number 978-0-470-72169-8. John Wiley and Sons, 2nd edition edition,
September 2011.
99
100 REFERENCES
[10] University of Geneva. PVsyst Contextual Help, 1994-2010. Also available on

the website: www.pvsyst.com.
[11] Spagnuolo G. Teodorescu R. Veerachary M. Petrone, G. and M. Vitelli.
Reliability issues in photovoltaic power processing systems. Industrial Elec-
tronics, IEEE Transactions on, 55(7):2569 –2580, july 2008.
[12] M.A. Eltawil and Z. Zhao. Grid-connected photovoltaic power systems:
Technical and potential problems—a review. Renewable and Sustainable
Energy Reviews, 14(1):112 – 129, 2010.
[13] www.pvsyst.com.
[14] T. Markvart. Solar Electricity. John Wiley and sons, ltd., 2000.
[15] R.A Messenger and J. Ventre. Photovoltaic Systems Engineering. Number
0-8493-1793-2. CRC Press LLC, 2nd edition edition, 2004.
[16] Nielsen H.K. Norum L.E. Saetre T.O. Midtgård, O.M. and G.H. Yordanov.
Over-irradiance (cloud enhancement) events in high latitudes, 2011. Ex-
tended abstract submitted for the 38th IEEE PVSC, 3-8 of June 2012.
[17] E. Barstad. The solar resource in norway. http://www.fornybar.no/
sitepageview.aspx?sitePageID=1648. Accessed: April 3rd, 2012.
[18] Ossenbrink H.A. Dunlop E.D., Huld T.A. and Šúri M. Potential of so-
lar electricity generation in the european union member states and candi-
date countries. http://re.jrc.ec.europa.eu/pvgis/, 2007. Solar Energy, 81,
1295-1305.
[19] Norwegian Institute of Meteorology. eklima. www.eklima.met.no. Statistical
database.
[20] M. Behnia and S. Odeh. Improving photovoltaic module efficiency using
water cooling. Heat Transfer Engineering, 30(6):499–505, 2009. cited By
(since 1996) 8.
[21] Meteo data sources. http://www.pvsyst.com/en/publications/
meteo-data-sources. Accessed: April 2nd, 2012.
[22] Müller S. Remund J. Kunz, S. and C. Schilter. Handbook part 1: Software.

Meteonorm, version 6 edition, December 2012.
[23] Meteo data comparisons. http://www.pvsyst.com/en/publications/
meteo-data-comparaisons$\#$fig$\_$yearly$\_$globh$\_$diff. Accessed:
April 2nd, 2012.
[24] T.P. Chang. The gain of single-axis tracked panel according to extraterres-
trial radiation. Applied Energy, 86(7–8):1074 – 1079, 2009.
[25] C. Breyer and J. Schmid. Global distribution of optimal tilt angles for fixed
tilted pv systems. 3-936338-26-4, pages 4715–4721, September 2010. 25th
REFERENCES 101
European Photovoltaic Solar Energy Conference and Exhibition/ 5th World

Conference on Photovoltaic Energy Conversion.
[26] W.D. Lubitz. Effect of manual tilt adjustments on incident irradiance on
fixed and tracking solar panels. Applied Energy, 88(5):1710 – 1719, 2011.
[27] P. Ineichen. Global irradiance on tilted and oriented planes: Model valida-
tions. February 2011. University of Geneva.
[28] J.E. Hay. Calculation of monthly mean solar radiation for horizontal and
inclined surfaces. Solar Energy, 23(4):301 – 307, 1979.
[29] T. Lejeune and A. Mermoud. Partial shadings on pv arrays: By-pass diode
benefits analysis. Institute for the Environmental Science, University of
Geneva, September 2010.
[30] Google Maps. maps.google.no, March 2012.
[31] Knubix. Image gallery. http://www.knubix.com/english/
flachdach-solaranlagen/index.asp.
[32] DPW solar. Ballasted roof mounts. http://www.dpwsolar.com/pdf/lit/

SL-SS-1094-1DPWBRM.pdf.
[33] J. Appelbaum and Bany J. The effect of shading on the design of a field of
solar collectors. Solar Cells, 20(3):201 – 228, 1987.
[34] International Electrotechnical Commission (IEC), 3, rue de Varembe
Geneva, Switzerland. Photovoltaic system performance monitoring - Guide-
lines for measurement, data exchange and analysis, first edition: 1998-04
edition, 1998. International Standard IEC 61724.
[35] U. Jahn and W. Nasse. Operational performance of grid-connected pv sys-
tems on buildings in germany. Progress in Photovoltaics: Research and
Applications, 12(6):441–448, 2004. cited By (since 1996) 16.
[36] Norsk elektroteknisk norm (NEK). Electrical low voltage installations. Num-
ber NEK 400:2006. NEK, third edition edition, 2006.
[37] Jensen Lund P.T. and T.A Johansen. Annual Report 2012, The Norwe-
gian Energy Regulator. Norwegian Water Resources and Energy Directorate,
2010.
[38] Hafslund Nett AS. Connection of production plants, 230-400 v. Guideline:
01-06-01.
[39] Nord Pool. Elspot prices. http://www.nordpoolspot.com/Market-data1/
Elspot/Area-Prices/ALL1/Hourly/. Accessed: 11th of May, 2012.
[40] Hafslund Nett AS. Landets første plusskunde. http://www.hafslundnett.no/

omoss/artikkelarkiv/les_artikkel.asp?artikkelid=2156, 2012. Accessed:
June 13th, 2012.
102 REFERENCES
[41] Hafslund Nett AS. Tariffs for firms and commerce - 2012. http://
www.hafslundnett.no/nett/artikler/les_artikkel.asp?artikkelid=239. Ac-
cessed: 14th of March, 2012.
[42] Norges Bank Central bank of Norway. Inflation. http://www.norges-

bank.no/no/prisstabilitet/inflasjon/, June 2006. Accessed: 27th of May,
2012.
[43] Jensen T. Haugen, S. and I. Magnussen. Hand-

book of socio-economic analysis of energy projects.
http://www.nve.no/Global/Publikasjoner/Publikasjoner2003. Accessed:
27th of May, 2012.
[44] J.A. Olseth and A. Skartveit. The solar radiation climate of norway. Solar
Energy, Vol. 37(No. 6):pp. 423–428, June 1986.
[45] J. Herron. Stay tuned. Photon International, the solar power magazine,
2012-03, March 2012.
[46] Spot market prices. Photon International, the solar power magazine, 2012-
04, April 2012.
[47] G.C. Schwartz and K.V. Srikrishnan. Handbook of Semiconductor Intercon-

nection Technology. Number 978-1-57444-674-6. CRC Press, second edition
edition, May 2006, 2012.
[48] R.A. Freedmand and H.D. Young. University Physics, with modern physics.
Pearson Addison-Wesley, 12th edition edition, 2008.
[49] REC. Rec peak energy series. http://www.recgroup.com/en/products/

modules/recpeakenergyseries/. Accessed: May 30th, 2012.
[50] Innotech Solar. Product downloads. http://innotechsolar.com/eng/

product_downloads/. Accessed: May 30th, 2012.
[51] SunPower. New sunpower e20 series: the world’s first 20% effi-
ciency solar panel. http://us.sunpowercorp.com/homes/products-services/
solar-panels/e20/. Accessed: May 30th, 2012.
[52] Eltek. Solar products. http://www.eltek.com/wip4/products.epl?cat=&k1=

25509&k2=&k3=&k4=. Accessed: May 30th, 2012.
[53] Danfoss. Tlx inverter range. http://danfoss.ipapercms.dk/Drives/DSI/

Factsheets/UK/TLXrange/. Accessed: May 30th, 2012.
[54] SMA. Sunny tripower 8000tl/10000tl/12000tl/15000tl/17000tl.

http://www.sma.de/en/products/solar-inverter-without-transformer/
sunny-tripower-8000tl-10000tl-12000tl-15000tl-17000tl.html. accessed:
May 30th, 2012.
REFERENCES 103
[55] SMA. Sunny boy 3000tl/4000tl/5000tl. http://www.sma.de/en/products/

solar-inverter-without-transformer/sunny-boy-3000tl-4000tl-5000tl.
html. Accessed: May 30th, 2012.
104 REFERENCES
Appendix A
Data Collection Process
The search engines that were used in order to find articles were Scopus and IEEE.
The information regarding the geographical coordinates and distances between
different locations was retrieved from Google Earth. Cases where Google Earth
was used are for example:
• Locating the project site and measuring the distance between this site and
the site of the default meteorological file for Oslo and the local weather
stations Lier and Ås.
• Retrieving information concerning the size and dimensions of the roof where
the PV installation is proposed installed.
Information concerning the size of the roof etc. was provided by Multiconsult.
This information was used alongside with the information of the roof given from
Google Earth. Information regarding the size of the twelve storey building which
could shade for the PV installation was collected from ”plan- og bygningsetaten”
in Oslo, Brit Eikholt.
Publications on the PVsyst web page and the contextual help of the program gave
insight into how PVsyst functions. In order to understand details, the author of
the PVsyst software and founder André Mermoud was contacted by mail. Two
of the mounting system suppliers were contacted by mail, since there was not
specified an edge zone explicitly in the technical data on their web page. Eric
Gilliland replied from DPW, and Kate Bayard replied from Solar Dock.
Hafslund, and Elvedin Grudic was contacted when the grid connection was inves-
tigated. The project work of master student Håkon Tranøy as well as his main
sources has been looked into. His project concerned the technical requirements
of end-users with energy surplus (surplus plus customers).
105
106 APPENDIX A. DATA COLLECTION PROCESS
Appendix B
Meteorological Data
Meteorological data collection
When searching for local information concerning temperature and global hori-
zontal irradiance, MET was contacted. In the Oslo area they have no functioning
irradiation measuring station operative at the moment. It was, however, found
information from Blindern (59.9423◦ N, 10.7201◦ E at an altitude of 94 meters),
Oslo which showed they measured global horizontal irradiance up till the year
2006. The accuracy of the irradiation measurements were considered insufficient
due to the fact that it had been measured at only four times a day at hour 1, 7, 13
and 19. Using these measurements would give maximum three values a day, re-
sulting in an unrealistic high value of irradiation during a day. The temperature
information from MET are based on daily normals from 1961 to 1990.
The irradiation measurements used in order to make the second meteorological

set was collected from two Bioforsk stations; Lier and Ås, from the web site http:
//lmt.bioforsk.no/agrometbase/getweatherdata.php. Bioforsk is the Norwegian
institute of agricultural and environmental research which was established in
2006. The weather stations in Lier and Ås are the two closest to Oslo, which
measure irradiation. Figure B.0.1 show the measuring stations closest to Skøyen,
where the green marker illustrates Lier and the blue illustrates Ås.
What could be an error in converting data from hourly to monthly values was
discovered, with the help of Georgi H. Yordanov from the University of Agder,
when working with the monthly extracted irradiance data from the two stations.
The values resulted in annual irradiation of 1300 up to 1600 kW/m2 . Converting
the hourly irradiance values to monthly values, the annual values were below
960 kW/m2 , which is considered a more reasonable value in southern Norway.
107
108 APPENDIX B. METEOROLOGICAL DATA
Figure B.0.1: The closest Bioforsk measuring stations to Skøyen.
Global Irradiation
Table B.0.1 and B.0.2 show the global horizontal irradiation of the free online
databases Meteonorm, NASA-SSE, PVGIS and RETScreen and the global hori-
zontal irradiation from the weather stations at Lier and Ås respectively.
109
Table B.0.1: Global irradiation (kWh/m2 )
Source: Meteonorm D NASA-SSE PVGIS RETScreen Meteonorm

January 11 12.09 8.03 9.30 11
February 28 30.79 21.81 24.30 28
March 66 71.61 53.01 68.20 63
April 106 111.00 95.70 102.00 107
May 172 164.92 140.74 156.55 142
June 166 166.20 154.80 147.30 149
July 169 170.81 146.32 142.91 136
August 131 131.13 108.19 108.19 119
September 78 82.80 66.00 63.60 78
October 37 40.30 31.00 40.61 39
November 15 16.20 11.07 11.10 19
December 8 7.44 5.52 5.89 7
Yearly 987 1005.29 842.19 879.95 898
Table B.0.2: Global irradiation (kWh/m2 ) from the local weather stations in Lier
and Ås
Lier Ås
January 10.10296 12.26918
February 26.88368 26.65028
March 71.41158 69.47032
April 118.21084 115.97058
May 158.7409 159.05088
June 175.62776 175.66038
July 144.92432 147.63778
August 114.33588 114.48102
September 76.56544 77.68156
October 40.30082 39.9209
November 12.83348 12.54814
December 6.24626 7.34862
Yearly 956.18392 958.68964
Temperature
Table B.0.3 shows the monthly average and the mean yearly temperatures from
Meteonorm, NASA, PVGIS and RETScreen.
Table B.0.3: Temperatures (◦ C)
Meteonorm NASA PVGIS RETScreen

Default Open City Fornebu Blindern
January -1.7 -4.4 -3.3 -4.3 -3.1 -3.3 -4.3
February -2.4 -4.1 -3.1 -3.9 -2.6 -2.8 -4
March 0.3 -0.2 0.6 -1.1 -0.2 1 -0.2
April 4.7 4.5 5.3 3.4 4.7 5.2 4.5
May 10.1 10.8 11.6 9.3 9.7 11.6 10.8
June 13.9 15.2 16 13.2 14 15.5 15.2
July 17 16.4 17.2 15.4 16.8 17.7 16.4
August 16.2 15.2 16.1 14.4 16.6 16.4 15.2
September 13.2 10.8 11.7 9.7 12.3 11.5 10.8
October 8.7 6.3 7.2 5.3 6.8 6.8 6.3
November 3.7 0.7 1.8 0.1 1.8 1.3 0.7
December 0.8 -3.2 -2 -3.3 -2.3 -2.2 -3.1
Yearly 7.1 5.7 6.6 4.9 6.3 6.6 5.7
111
Table B.0.4 shows the monthly normal values based on the daily normal values
in between 1961 and 1990 measured by MET at Blindern in Oslo. Figure B.0.2
shows the lowest and highest temperature measured at Blindern between 2000-
2011 in January and July respectively.
Table B.0.4: Monthly normal temperature values (◦ C), Blindern, Oslo
MET
January -4.30
February -3.99
March -0.21
April 4.49
May 10.80
June 15.19
July 16.40
August 15.19
September 10.80
October 6.30
November 0.70
December -3.10
Yearly average 5.70
Figure B.0.2: Temperature maximums and minimums from 2000 − 2011 in Jan-
uary and July
113
Diffuse irradiation
Table B.0.5 shows the diffuse irradiation data which was possible to retrieve. As
seen there are only data from two of the databases, the Meteonorm and NASA.
Table B.0.5: Diffuse irradiation data (kWh/m2 ), D = default
Meteonorm NASA-SSE
D
January 8 7 8.4
February 18 16 19.5
March 39 33 41.5
April 59 48 60.3
May 72 81 77.8
June 84 86 83.1
July 80 79 81.5
August 66 60 69.1
September 44 41 44.4
October 25 24 25.1
November 10 9 11.1
December 5 5 5.6
Yearly 510 489 527.4
Wind
Table B.0.6 shows the wind data possible to derive from the databases.
Table B.0.6: Wind data (m/s)
Meteonorm NASA-SSE RETScreen

Fornebu Blindern
January 5.6 5.52 2.2 2.5
February 5.4 5.11 2.2 2.6
March 5.7 5.01 2.5 2.6
April 5.8 4.87 2.7 3
May 4.7 4.66 2.7 2.8
June 4.6 4.38 2.8 2.8
July 4.2 4.37 2.7 2.5
August 4.6 4.37 2.6 2.6
September 4.8 4.96 2.5 2.6
October 4.6 5.26 2.5 2.6
November 5 5.36 2.1 2.6
December 5.5 5.4 1.9 2.6
Yearly 5.1 5 2.5 2.7
Appendix C
Shade
Distance Between Modules

The distances between the module rows were calculated by using Equation 3.3.1,
five different tilt angles (20◦ , 30◦ , 40◦ , 53◦ and 59◦ ) at 10:00, 11:00 and 12:00 a
clock. Table C.0.1 shows the azimuth and altitude numbers that was used from
USNO and NASA. The numbers from USNO were calculated from the specific day
and time in year 2011, while the NASA numbers are monthly averaged numbers.
Table C.0.1: Azimuth and altitude numbers from USNO and NASA
Azimuth Altitude
USNO NASA USNO NASA
March 21st
10:00 139.7◦ 140◦ 24.4◦ 22.0◦
11:00 155.9◦ 155◦ 28.4◦ 26.0◦
◦ ◦ ◦
12:00 172.9 172 30.4 28.0◦
June 21st
10:00 130.9◦ 131◦ 46.2◦ 46.0◦
11:00 150.5◦ 151◦ 51.0◦ 50.7◦
◦ ◦ ◦
12:00 172.7 173 53.4 53.0◦
December 21st
10:00 149.3◦ 150◦ 2.5◦ 3.03◦
11:00 162.7◦ 163◦ 5.4◦ 5.95◦
◦ ◦ ◦
12:00 176.5 177 6.7 7.15◦
115
116 APPENDIX C. SHADE
Table C.0.2: Spring equinox distances, D in meters, between rows, polycrystalline

modules, USNO
March 21st 20◦ 30◦ 40◦ 53◦ 59◦

10:00 0.570 0.834 1.071 1.331 1.429
11:00 0.573 0.837 1.076 1.337 1.435
12:00 0.574 0.839 1.078 1.339 1.437
Table C.0.3: Spring equinox distances, D in meters, between rows, polycrystalline

modules, NASA
March 21st 20◦ 30◦ 40◦ 53◦ 59◦

10:00 0.643 0.939 1.208 1.501 1.611
11:00 0.630 0.921 1.184 1.471 1.578
12:00 0.631 0.923 1.186 1.474 1.582
As seen from Table C.0.2 and C.0.3 the distance between the module rows are
increasing as the clock is reaching noon when USNO numbers for the azimuth
and altitude were used, in contrast to when the NASA numbers for azimuth
and altitude were used. When considering June 21st (the day when the least
amount of distance between the rows are necessary in order to have no shade on
other rows, see Table C.0.4) both the numbers from USNO and NASA increases
with increasing hour of the day. Then considering the day when the distance
between the rows are as large as they can be, December 21st (see Table C.0.5),
the distance is decreasing with increasing hour of the day. Only the numbers from
NASA are shown in Table C.0.4 and C.0.5 since these numbers are considered
more conservative. The numbers from NASA are consequently the ones used to
set the pitch distance.
Table C.0.4: Distance D between rows June 21st, polycrystalline modules, NASA
June 21st 20◦ 30◦ 40◦ 53◦ 59◦

10:00 0.215 0.314 0.404 0.502 0.539
11:00 0.243 0.355 0.457 0.567 0.609
12:00 0.254 0.371 0.477 0.592 0.636
During summer increasing altitude will decrease the shade, however, not much
since the sun has an altitude above 45◦ . The azimuth is almost the same whether
it is June or December. Increasing azimuth angle increases the length of the shade
until its maximum length at approximately 180◦ is reached, before the shade
decreases again. As the altitude is above 45◦ the altitude does not diminish
117
Table C.0.5: Distance D between rows December 21st, polycrystalline modules,

NASA
December 21st 20◦ 30◦ 40◦ 53◦ 59◦

10:00 5.545 8.107 10.422 12.949 13.898
11:00 3.11 4.547 5.845 7.262 7.794
12:00 2.698 3.945 5.071 6.301 6.762
the length caused by the azimuth to the same extent as in the autumn or winter
months (September and March). The shade at noon will, therefore, be the largest
in the time between March and September. The reason why the distances are
increasing in Table C.0.2 and decreasing in Table C.0.3 are, therefore, due to the
small differences in azimuth and altitude from the two sources USNO and NASA
on the day of spring equinox.
Table C.0.6 illustrate the distance D in meters for both monocrystalline modules
and polycrystalline modules, with azimuth and altitude numbers from NASA.
Table C.0.7 show the pitch distance between monocrystalline and polycrystalline
modules based on the azimuth and altitude numbers from NASA.
Table C.0.6: Spring equinox distances, D in meters, NASA
March 21st 20◦ 30◦ 40◦ 53◦ 59◦

Mono
10:00 0.678 0.992 1.275 1.584 1.7
11:00 0.665 0.972 1.249 1.552 1.666
12:00 0.666 0.974 1.252 1.556 1.67
Poly
10:00 0.643 0.939 1.208 1.501 1.611
11:00 0.63 0.921 1.184 1.471 1.578
12:00 0.631 0.923 1.186 1.474 1.582
Table C.0.7: Spring equinox pitch distances in meters, NASA
March 21st 20◦ 30◦ 40◦ 53◦ 59◦

Mono
10:00 1.661 1.897 2.076 2.213 2.239
11:00 1.648 1.878 2.051 2.182 2.205
12:00 1.649 1.878 2.053 2.185 2.209
Poly
10:00 1.574 1.78 1.967 2.097 2.121
11:00 1.561 1.779 1.943 2.067 2.089
12:00 1.562 1.781 1.946 2.07 2.092
Shade Analysis
Figure C.0.1 illustrates the results of the shade analysis seen in Tables C.0.8 to
Table C.0.13. The shade analysis was performed with a module tilt of 20◦ , 60 %
electrical shading effect with a pitch distance of 2 meters. The module size is
set in the size of polycrystalline modules (1.665 x 0.991 x 0.043 m). The results
are, however, considered to be equally valid for the monocrystalline case (1.559 x
1.046 x 0 046 m), because of to the small deviation in modular size. Sunrise was
simulated to occur at 06:30 while sun set was at 18:30. Since the shade analysis
is done by manually observing which modules that are shaded each half hour
in PVsyst, there could occur faults. The resolution of the field could also have
been too small in order to see exactly when modules are shaded which is another
source of error. However, since the meteorological data set could be interpreted
as a general year there will be deviations from the analysis in a live field.
Table C.0.8: The time and modules shaded in field A during March 21st, x =
partially shaded module, f = fully shaded module
FIELD
A 6 7 8 9 10 11
06:30 x x x x f f
07:00 x x x f
07:30 x x
08:00 x x
08:30 x x
09:00 x
119
Table C.0.9: The time and modules shaded in field B, during March 21st, x =
FIELD
B 5 6 7 8 9 10 11 12 13 14 15 16 17
06:30
07:00
07:30
14:30
15:00
15:30
16:00
16:30 x
17:00 x x
17:30 f f x x
18:00 x x x x x f f f f f f f f
Table C.0.10: The time and modules shaded in field B, during March 21st, x =
FIELD
B 18 19 20 21 22 23 24 25 26 27
06:30
07:00
07:30
14:30 x
15:00 x
15:30 x
16:00 x x
16:30 x x
17:00 x x x
17:30 f f x x x
18:00 f f f f f f f x x x
Figure C.0.1: Overview of the results of the shading analysis for March 21st with
2 meter pitch distance. The dark red modules are the ones that are fully shaded
on some point during the day, while the peach coloured modules are partially
shaded during the day.
121
Table C.0.11: Time and modules shaded in field B, during March 21st, x =
FIELD
B 28 29 30 31 32 33 34 35 36 37 38 39 73
06:30
07:00
07:30 x
14:30
15:00
15:30 x
16:00 x
16:30 x
17:00 x x
17:30 x x x x x
18:00 x x x x x f f x x x x x
Table C.0.12: Time and modules shaded in field B, during March 21st, x =
FIELD
C 1 2 3 4 5 6 7 8 9 10
16:00
16:30
17:00 x x x
17:30 x x x x x x x x x x
18:00 x x x x x x x f f
Table C.0.13: Time and module number shaded in field C during March 21st, x
= partially shaded module, f = fully shaded module
FIELD
C 11 12 13 14 15 16 17 18 19 20
16:00 x x
16:30 x f x f x
17:00 x f x f f f f
17:30 x f f f f f f f f f
18:00 f f f f f f f f f f
Appendix D
Simulation Settings
Orientation Parameters
Field type: fixed tilted plane
Plane tilt: 40◦ , 30◦ , 25◦ , 20◦
Azimuth: 0◦
Loss Parameters
The three modules tested were one module from REC (REC250PE), one from
Innotech Solar (ITS EcoPlus, poly-Up 250Wp ) and one from SunPower (E20,
SPR-327NE-WHT-D).
The thermal parameter was defined by using the NOCT coefficient, and then the
corresponding constant loss factor was given by PVsyst (seen in Table D.0.1).
The DC ohmic losses were the only one defined in PVsyst during the simulations,
the option of defining the AC ohmic losses are as mentioned present in PVsyst.
Table D.0.1: Thermal parameter simulation settings
REC ITS SunPower

◦ ◦ ◦ ◦
NOCT coefficient 47.9 C (± 2 C) 47.9 C (± 2 C) 45◦ C (± 2◦ C)
Constant loss factor 25.8 (24.7) 25.8 (24.7) 28.8 (26.7)
123
124 APPENDIX D. SIMULATION SETTINGS
Table D.0.2: Ohmic losses parameters
DC circuit REC ITS SunPower

Global wiring resistance, mOhm 225 93.8 105.5
or loss fraction at STC, % 1.5 1.5 1.5
Voltage drop across series diode, V 0 0 0
Table D.0.3: Module quality and mismatch parameters
REC ITS SunPower

Module quality:
Module efficiency loss 0% 0% 0%
Mismatch losses:
Power loss at MPP 2% 2% 2%
Loss when running at fixed voltage 4% 4% 4%
Table D.0.4: Soiling losses parameters
January February March April May June

50 % 35 % 0% 0% 0% 0%
July August September October November December

0% 0% 0% 0% 0% 35 %
Appendix E
Cable sizing
After each of the modules where assigned to their string, the lengths were mea-
sured in the 3D representation of the building with the modules placed on the
roof. In order to find the total lengths Equation E.0.1 was used.
Cable length = Nmodules (Mlength +2Mf rame )+Nspaces (2height+pitch) (E.0.1)
where Nmodules are the number of modules in the string, Mlength is the length
of one module, Mf rame is the frame length, Nspaces are the number of spaces
between the rows (two rows equals one space) and height is the height from the
junction box to the ground. The junction boxes on the modules are assumed to
be located on module width divided by two.
The minimum cross section of the cables were calculated by using Equation 3.4.10
from Section 3.4 with a cable loss of 1.5 %. The metal resistivity is highly
dependent on temperature and is described by the following Equation
ρ = ρ0 [1 + α0 (T − T0 )] (E.0.2)
where ρ0 is the metals resistivity (Ωcm) defined at a reference temperature (T0 )

which is usually 0◦ C or 20◦ C and α0 is the temperature coefficient of resis-
tivity (◦ C−1 ). At a temperature T ρ is the resistivity. The bulk value of ρ0
is 1.678 µΩcm at 20◦ C [47, p.320], resulting in T0 = 20◦ C. The approximate
temperature coefficient of copper close to room temperature is 0.00393◦ C−1 [48,
p.852]. Calculating the copper resistivity at a temperature of 50◦ C results in
ρ = 18.75 mΩ/mm2 /m. A ρ of copper of 18.3 mΩ/mm2 /m is used in [7, p.221].
If the computational tool in PVsyst is used in order to calculate the wiring loss,
a ρ may be defined by the user or a default value of 22 mΩ/mm2 /m can be used
for a temperature of 50◦ C.
125
126 APPENDIX E. CABLE SIZING
REC modules, Eltek inverters

The module distribution which was made for the REC modules with the Eltek
inverter is seen in Figure E.0.1. The measured cable lengths are also displayed
in this figure. Table E.0.1 shows the total lengths of the cables per string, with
a 10 % safety margin added in the end due to uncertainty of the accuracy of the
measuring tool in the 3D representation in PVsyst.
Figure E.0.1: Module string distribution with the REC module and the Eltek
inverter
Table E.0.2 shows the minimum cross section area of the cables with 1.5 % loss,
the nearest normal cross section area of PV module cables and the loss calculated
with this cross section in percent. Standard cross sectional areas for PV module
cables are 1.5 mm2 , 2.5 mm2 , 4 mm2 and 6 mm2 .
127
Table E.0.1: Cable lengths for the alternative with the REC module and Eltek
inverter
String1 String2 String 3 String 4 String 5

Modules in a string 10 20 20 20 20
Number of spaces 0 4 3 2 7
Module tilt 20 20 20 20 20
Module length 1.665 1.665 1.665 1.665 1.665
Module width 0.991 0.991 0.991 0.991 0.991
Module frame 0.086 0.086 0.086 0.086 0.086
Pitch 2 2 2 2 2
Height, junction box 0.1695 0.1695 0.1695 0.1695 0.1695
Cable length 17.51 44.376 42.037 39.698 51.393

String colour Peach Blue Red Green Purple
String part 1 24.527

String part 2 17.51
Length between fields 4.7
Min lengths 46.737 44.376 42.037 39.698 51.393
Length to the door: 35.2 45.8 28.1 27.7 10.6
Total lengths 81.937 90.176 70.137 67.398 61.993
Safety margin, 10 % 90.1307 99.1936 77.1507 74.1378 68.1923
Table E.0.2: The minimum cross section area (CSA) with the measured lengths
(ρ = 18.3mΩ/mm2 /m)
String 1 String 2 String 3 String 4 String 5

2
Minimum CSA, mm 2.96 3.25 2.53 2.43 2.24
Closest CSA, mm2 4 4 4 4 4
Loss 1.11 % 1.22 % 0.95 % 0.91 % 0.84 %
REC modules, SMA inverters

The module distribution which was made for the REC modules with the SMA
inverter is seen in Figure E.0.2. The measured cable lengths are also displayed
in this figure. Table E.0.3 and Table E.0.4 show the total lengths of the cables
per string, with a 10 % safety margin added in the end due to uncertainty of the
accuracy of the measuring tool in the 3D representation in PVsyst.
Figure E.0.2: Module string distribution with REC modules and SMA inverter
129
Table E.0.3: Cable lengths for the alternative with the REC module and SMA
inverter

Module tilt 20 20 20 20 20
Module length 1.665 1.665 1.665 1.665 1.665
Module width 0.991 0.991 0.991 0.991 0.991
Module frame 0.086 0.086 0.086 0.086 0.086
Pitch 2 2 2 2 2
Cable length 14.596 36.784 32.106 32.106 7.592

String colour Peach Blue Red Grey Green
String part 1 24.527 29.192

String part 2 14.596 7.592
Length between fields 4.7 20.88
Min cable length 43.823 36.784 32.106 32.106 57.664
Lengths to the door 47.9 43.9 45.9 33.8 7.7
Total length 91.723 80.684 78.006 65.906 65.364
Safety margin, 10% 100.8953 88.7524 85.8066 72.4966 71.9004
Table E.0.4: Cable lengths for the alternative with the REC module and SMA
inverter, string 6
String 6
Modules in a string 17
Number of spaces 6
Module tilt 20
Module length 1.665
Module width 0.991
Module frame 0.086
Pitch 2
Height, junction box 0.1695
Cable length 43.801

String colour Purple
String part 1
String part 2
Length between fields
Min lengths 43.801
Lengths to the door 7.7
Total length 51.501
Safety margin, 10 % 56.6511
Table E.0.5 shows the minimum cross section area of the cables with 1.5 % loss,
the nearest normal cross section area of PV module cables and the loss calculated
with this cross section in percent. Standard cross sectional areas for PV module
cables are 1.5 mm2 , 2.5 mm2 , 4 mm2 and 6 mm2 .
Table E.0.5: The minimum cross section area (CSA) with the measured lengths
(ρ = 18.3mΩ/mm2 /m)
String1 String2 String3 String4 String5 String6

Minimum CSA 3.89 3.42 3.31 2.80 2.77 2.19
Closest CSA 4 4 4 4 4 2.5
Loss 1.46 % 1.28 % 1.24 % 1.05 % 1.04 % 1.31 %
131
SunPower modules, SMA inverters

The module distribution which was made for the SunPower modules with the
SMA inverter is seen in Figure E.0.3. The measured cable lengths are also dis-
played in this figure. Table E.0.6 and Table E.0.7 show the total lengths of the
cables per string, with a 10 % safety margin added in the end due to uncertainty
of the accuracy of the measuring tool in the 3D representation in PVsyst.
Figure E.0.3: Module string distribution with SunPower modules and SMA in-
verter
Table E.0.6: Cable lengths for the alternative with the SunPower module and
SMA inverter, to string 5
String1 String2 String 3 String 4 String 5

Module tilt 20 20 20 20 20
Module length 1.559 1.559 1.559 1.559 1.559
Module width 1.046 1.046 1.046 1.046 1.046
Module frame 0.092 0.092 0.092 0.092 0.092
Pitch 2 2 2 2 2
Cable length 27.5922 28.5364 23.8208 23.8208 23.8208

String colour Peach Blue green Grey Dark grey

String part 2 3.302
Min cable lengths 35.5942 28.5364 23.8208 23.8208 23.8208
Lengths to the door 39.2 43.8 40.2 35.3 25.1
Total length 74.7942 72.3364 64.0208 59.1208 48.9208
Safety margin, 10 % 82.27362 79.57004 70.42288 65.03288 53.81288
Table E.0.8 and Table E.0.7 show the minimum cross section area of the cables
with 1.5 % loss, the nearest normal cross section area of PV module cables and
the loss calculated with this cross section in percent. Standard cross sectional
areas for PV module cables are 1.5 mm2 , 2.5 mm2 , 4 mm2 and 6 mm2 .
133
Table E.0.7: Cable lengths for the alternative with the SunPower module and
SMA inverter, to string 8
String 6 String 7 String 8

Modules in a string 13 7 13
Number of spaces 2 2 5
Module tilt 20 20 20
Module length 1.559 1.559 1.559
Module width 1.046 1.046 1.046
Module frame 0.092 0.092 0.092
Pitch 2 2 2
Height, junction box 0.1789 0.1789 0.1789
Cable length 26.1786 16.2726 33.252

String colour Red Pink Purple
String part 1 9.906

Min cable lengths 26.1786 47.0586 33.252
Lengths to the door 27.5 6 12.4
Total length 53.6786 53.0586 45.652
Safety margin, 10 % 59.04646 58.36446 50.2172
Table E.0.8: The minimum cross section area (CSA) with the measured lengths,
SunPower modules, SMA inverters. String 1 to string 5 (ρ = 18.3mΩ/mm2 /m)

Minimum CSA, mm2 1.688189 1.632713 1.445021 1.334422 1.104197
Closest CSA 2.5 2.5 1.5 1.5 1.5
Loss 1.01 % 0.98 % 1.45 % 1.33 % 1.10 %
Table E.0.9: The minimum cross section area (CSA) with the measured lengths,
SunPower modules, SMA inverters. String 6 to string 8 (ρ = 18.3mΩ/mm2 /m)
String 6 String 7 String 8

Minimum CSA, mm2 1.211586 1.197592 1.030417
Closest CSA 1.5 1.5 1.5
Loss 1.21 % 1.20 % 1.03 %
Appendix F
Technical Data
Modules
Figure F.0.1 illustrates the technical data of the Peak Energy EU series by REC.
Figure F.0.2 illustrates the technical data of the Eco Plus series by ITS while
Figure F.0.3 illustrates the technical data of the E20 series by SunPower.
Inverters
Figure F.0.4 and Figure F.0.5 illustrate the technical data of the two Eltek in-
verters Theia TL String and Theia HE-t. Figure F.0.6 illustrates the technical
data of the Danfoss inverter series TLX. Figure F.0.7 to Figure F.0.12 illustrate
the technical data of the SMA Sunny Tripower series. Figure F.0.13 to Figure
F.0.16 illustrate the technical data of the SMA Sunny Boy series without reactive
power control. This Sunny Boy is the base inverter.
135
136 APPENDIX F. TECHNICAL DATA
Figure F.0.1: REC, Peak Energy EU Series [49]

137
Figure F.0.2: ITS, Eco Plus series [50]

Figure F.0.3: SunPower, E20 series [51]

139
Figure F.0.4: Eltek, Theia TL String series [52]

Figure F.0.5: Eltek, Theia HE-t series [52]

141
Figure F.0.6: Danfoss, TLX series [53]

Figure F.0.7: SMA, Tripower input specifications [54]
Figure F.0.8: SMA, Tripower output specifications [54]

143
Figure F.0.9: SMA, Tripower efficiency and protection specifications [54]
Figure F.0.10: SMA, Tripower, general data [54]

Figure F.0.11: SMA, Tripower, features [54]
Figure F.0.12: SMA, Tripower, footnotes [54]

145
Figure F.0.13: SMA, Sunny Boy input and output specifications [55]
Figure F.0.14: SMA, Sunny Boy efficiency and protection specifications [55]
147
Figure F.0.15: SMA, Sunny Boy general specifications [55]

Figure F.0.16: SMA, Sunny Boy features and footnotes [55]

Appendix G
Simulation Results
A number of simulations were performed when varying the fraction of electrical

effect, the module tilt angles and simulating with several inverters. Attachments
of the complete result reports would be to extensive in this appendix. The result
reports are, therefore, found in an attached CD.
Two meteorological datasets
Comparing the two data sets
Figure G.0.1 illustrates the global horizontal irradiation of the default and the
new meteorological data set performed on the same system with a module tilt of
20◦ and an electrical effect of 60 %. Table G.0.1 and Table G.0.2 shows the hori-
zontal global irradiation and incident global irradiation in the tilted plane and the
difference between these two factors for the default and the new meteorological
data sets respectively.
149
150 APPENDIX G. SIMULATION RESULTS
Table G.0.1: The horizontal and incident global irradiation of the default mete-
orological data set
DEFAULT Horizontal Incident Difference

2 2
kWh/m kWh/m kWh/m2
January 11 19.2 8.2
February 28 43.0 15.0
March 66 83.7 17.7
April 106 122.1 16.1
May 172 186.3 14.3
June 166 172.0 6.0
July 169 178.8 9.8
August 131 145.2 14.2
September 78 96.0 18.0
October 37 50.3 13.3
November 15 27.2 12.2
December 8 19.2 11.2
Yearly 987 1143 156
Table G.0.2: The horizontal and incident global irradiation of the new meteoro-
logical data set
METEO2 Horizontal Incident Difference

kWh/m2 kWh/m2 kWh/m2
January 11.1 20.8 9.7
February 27.4 42.9 15.5
March 67.5 88.8 21.3
April 111.8 131.3 19.5
May 157.9 169.6 11.7
June 166.6 172.6 6.0
July 149.4 157.8 8.4
August 119.7 132.8 13.1
September 77.6 95.3 17.7
October 39.1 56.3 17.2
November 14.8 27.2 12.4
December 7.2 16.5 9.3
Yearly 950.1 1111.9 161.8
151
Figure G.0.1: The horizontal global irradiation for the default and new meteoro-
logical data set
152 APPENDIX G. SIMULATION RESULTS
Appendix H
Production and Demand
Energy Production
Table H.0.1 presents the produced energy of all three alternatives for each month
of the year.
Table H.0.1: The monthly and yearly energy production for all three alternatives
in kWh
REC REC SunPower

Eltek SMA SMA
January 127.62 122.07 175.73
February 508.04 512.73 683.57
March 1 961.94 2 008.37 2 685.34
April 2 883.48 2 956.53 3 999.27
May 3 627.53 3 720.74 5 050.32
June 3 596.34 3 687.90 5 021.18
July 3 273.12 3 353.64 4 568.55
August 2 779.12 2 847.56 3 871.60
September 2 031.46 2 080.11 2 811.11
October 1 099.82 1 121.53 1 489.71
November 391.26 392.93 537.25
December 117.28 113.91 162.90
Yearly 22.40 22 918.01 31 056.54
153
154 APPENDIX H. PRODUCTION AND DEMAND
Consumption
Figure H.0.1 illustrates the consumption in the building from the main meter,
where the consumptions from Figure H.0.2 are withdrawn.
Figure H.0.1: Consumption of the main meter, 2011

155
Figure H.0.2: Consumption for four part-meters in the building, 2011

156 APPENDIX H. PRODUCTION AND DEMAND
Table H.0.2: The average time of occurring power peaks and the average max-
imum load peak value of the 15 highest power peaks occurring in each month,
2011
Average time Average

of occurrence max peak
January 11.93 329.50
February 11.27 340.50
March 12.27 343.00
April 12.87 350.75
May 14.80 406.50
June 14.67 460.25
July 14.20 439.50
August 14.47 488.00
September 14.33 401.75
October 12.80 380.50
November 11.87 370.50
December 12.20 362.25
Appendix I
Economics
Investment costs
Table I.0.1 illustrates the investment cost for the three final alternatives. The
figures marked with 1 are from Photon Magazine [45]. The figures marked with
2
are experience based on previous calculations from Multiconsult. The figures
marked with 3 are figures based on calculations from a German roof installation
of installed capacity of 60 kWp (from Multiconsult).
Income
Figure I.0.2 presents the Hafslund tariff agreement which is in effect today (2012).
It is from these numbers the income from the savings when producing energy is
calculated. Table I.0.3 presents the simulated energy production form the three
alternatives, while Table I.0.4 presents two power generating situations for the
three alternatives. Table I.0.5 presents the calculated income when considering
both power situations and the energy variable. Table I.0.6 presents the income
if all of the energy was sold on the spot market.
157
158 APPENDIX I. ECONOMICS
Table I.0.1: Investment cost for the three alternatives. Rate per May 11th, 2012:
7.622 EUR/NOK
REC REC SunPower

Eltek SMA SMA Unit
Module price 750 750 1140 1
e/kWp
Length 1.665 1.665 1.559 m
Width 0.991 0.991 1.046 m
Power 250 250 327 Wp
Efficiency 15.15 15.15 20.05 %
Inverter 297.14 210.001
210.00 1
e/kWp
Mounting system 250.00 2
250.002
188.90 e/kWp
Montage 200.002 200.002 151.12 e/kWp
DC cables 214.00 3
214.003
161.69 e/kWp
AC cables 131.20 2
131.202
99.13 e/kWp
Communication 40.00 2
40.002
30.22 e/kWp
Total BoS cost 1 132.34 1 045.20 841.06 e/kWp
Module + BoS cost 1 882.34 1 795.20 1 981.06 e/kWp
14 347.12 13 682.93 15 099.56 NOK/kWp
Discounted Costs and Income

Figure I.0.1 shows the discounted module maintenance and repair costs of the
three alternatives. Module maintenance is in this case considered necessary once
a year, while the inverter repair is considered necessary every five years. The
period of analysis is 20 years, the discount rate is 6 % and the inflation rate is
2.5 %. Figure I.0.2 presents the calculated discounted income in a best case and
worst case scenario and the discounted yield in kWh for the three alternatives
with 0.75 % degradation in production per year. A 0.75 % degradation per year
results in a total degradation of the modules to be 13.3 % during the analysis
period of 20 years.
159
Table I.0.2: The tariff in 2012 for low voltage connections (230 V and 400 V) to
Hafslunds grid [41]. Summer = April - October, winter = November - March.
Set price Power variable Energy variable

NOK/mth NOK/kW/mth NOKcent/kWh
Summer Winter Summer Winter
415 25 74 5.25 7.4
Table I.0.3: The simulated energy production from the three alternatives
Eltek SMA SunPower

Energy kWh/mth kWh/mth kWh/mth
January 128 122 176
February 508 513 684
March 1 962 2 008 2 685
April 2 883 2 957 3 999
May 3 628 3 721 5 050
June 3 596 3 688 5 021
July 3 273 3 354 4 569
August 2 779 2 848 3 872
September 2 031 2 080 2 811
October 1 100 1 122 1 490
November 391 393 537
December 117 114 163
Yearly 22 397 22 918 31 057
Table I.0.4: The power production in an average situation and maximum situa-
tion for the three alternatives
Eltek SMA SunPower

Min. Max. Min. Max. Min. Max.
Power kW kW kW kW kW kW
January 0.67 3.15 0.68 3.21 0.96 3.81
February 2.20 9.38 2.28 9.63 2.98 12.57
March 5.58 17.66 5.78 18.19 7.64 24.19
April 7.00 18.88 7.39 19.43 9.86 26.41
May 7.24 19.73 7.53 20.31 10.13 27.80
June 6.82 19.68 7.27 20.26 9.72 28.00
July 6.25 20.21 6.52 20.81 8.85 28.72
August 6.20 19.95 6.48 20.54 8.69 28.41
September 5.52 17.24 5.75 17.75 7.72 24.22
October 3.88 16.88 4.03 17.38 5.28 23.07
November 1.99 9.27 2.11 9.52 2.75 12.33
December 0.79 3.12 0.78 3.18 1.11 4.41
Table I.0.5: The calculated income in an worst and a best case scenario when
the PV installation is not producing more than the building consumes
Energy Power Total

Min. Max. Min. Max.
Eltek 1 243 1 904 6 465 3 147 7 708
SMA 1 271 1 985 6 647 3 256 7 918
SunPower 1 722 2 649 8 906 4 370 10 627
Table I.0.6: The theoretical income if the total amount of energy was sold at spot
price
Spot Price
Eltek 7 647.76
SMA 7 825.66
SunPower 10 604.67
161
Figure I.0.1
Figure I.0.2

Design of Roof PV Installation in Oslo: Siv Helene Nordahl

Uploaded by

Copyright:

Available Formats

Design of Roof PV Installation in Oslo: Siv Helene Nordahl

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Design of Roof PV Installation in Oslo: Siv Helene Nordahl

Uploaded by

Copyright:

Available Formats

Design of Roof PV Installation in Oslo

Siv Helene Nordahl

Master of Energy and Environmental Engineering

Norwegian University of Science and Technology

The international photovoltaic industry has experienced a powerful growth, and

Assignment given: Trondheim, 23.01.2012

This thesis is centered around the design of a grid-connected photovoltaic (PV)

Denne oppgavens hovedfokus er å designe et solcelleanlegg tilknyttet nettet på et

Siv Helene Nordahl

Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

5 Customer Status and Economics 85

Appendix A Data Collection Process 105

Appendix B Meteorological Data 107

Appendix C Shade 115

Appendix D Simulation Settings 123

Appendix E Cable sizing 125

Appendix F Technical Data 135

Appendix H Production and Demand 153

Appendix I Economics 157

2.1.1 Schematic of a stand-alone and a grid-connected PV system . . . 5

3.1.1 Solar radiation in the atmosphere . . . . . . . . . . . . . . . . . . 13

4.0.1 Main results of the base case system . . . . . . . . . . . . . . . . 64

5.1.1 Simulated energy production per month of the three alternatives 86

3.1.1 Overview of usual albedo values given in PVsyst . . . . . . . . . . 21

1.2 Objectives and Limitations

The objective of this thesis is to design a grid-connected PV system for a specific

A PV system as a whole is complicated as there are many variables and several

1.3 Outline of the Thesis

2.1 PV Systems in General

Figure 2.1.1: Schematic of a stand-alone PV system to the left and a grid-

A grid-connected PV system comprises of PV modules and Balance of Sys-

2.2 The Electrical Characteristics of a PV Cell,

Figure 2.2.1: PV cell, PV module, PV string, PV array

Figure 2.2.2: I-V characteristics of a diode. F = forward, R = reversed

cell is an extended equivalent circuit called the single-diode model (illustrated

Figure 2.2.3: The single diode model

Figure 2.2.4: I-V and power characteristics of a PV cell

Figure 2.2.5: Interconnection of PV modules in series and parallel

in parallel the current increases as seen in the figure.

2.3 The Inverter

There are several PV system simulation programs. Some are:

3.1 Project: Geographical Location and

3.1.1 Solar Radiation and Meteorology

The amount of effect produced by a PV module relies on how much radiation it

Figure 3.1.1: Solar radiation in the atmosphere

Figure 3.1.2: AM 1 solar spectrum after atmospheric absorption effects [15]

Figure 3.1.3: Air mass

• Cell temperature = 25◦ C

The characterisation of the PV module is further done by measuring the Nominal

• Solar radiation = 800 W/m2

Figure 3.1.4: Solar irradiation map of Norway on a horizontal in January and

Figure 3.1.5: Yearly global irradiation on optimally inclined south oriented PV

Position of the sun

The location of the sun is defined by two angles:

sin α = sin δ sin φ + cos δ cos φ cos ω (3.1.2)

sin α sin φ − sin δ

Temperature and irradiance influence on the PV cell

The nominal operation of PV systems is at STC conditions, however, the PV

3.1.2 PVsysts Geographical and Meteorological