2050

tiv

y

Te

ec

rg

2040

Ene

2035

es

2045 chn rs olo g y P e

p

Technology Roadmap
Solar Photovoltaic Energy

Secure

Sustainable

Together

2014 edition

INTERNATIONAL ENERGY AGENCY
The International Energy Agency (IEA), an autonomous agency, was established in November 1974.
Its primary mandate was – and is – two-fold: to promote energy security amongst its member countries through collective response to physical disruptions in oil supply, and provide authoritative research and analysis on ways to ensure reliable, affordable and clean energy for its 29 member countries and beyond. The IEA carries out a comprehensive programme of energy co-operation among its member countries, each of which is obliged to hold oil stocks equivalent to 90 days of its net imports.
The Agency’s aims include the following objectives: n Secure member countries’ access to reliable and ample supplies of all forms of energy; in particular, through maintaining effective emergency response capabilities in case of oil supply disruptions. n Promote sustainable energy policies that spur economic growth and environmental protection in a global context – particularly in terms of reducing greenhouse-gas emissions that contribute to climate change. n Improve transparency of international markets through collection and analysis of energy data. n Support global collaboration on energy technology to secure future energy supplies and mitigate their environmental impact, including through improved energy efficiency and development and deployment of low-carbon technologies. n Find solutions to global energy challenges through engagement and dialogue with non-member countries, industry, international organisations and other stakeholders.

IEA member countries:
Australia
Austria
Belgium
Canada
Czech Republic
Denmark
Estonia
Finland
France
Germany
Greece
Hungary
Ireland
Italy
Secure Sustainable Together
Japan
Korea (Republic of)
Luxembourg
Netherlands
New Zealand
Norway
Poland
Portugal
Slovak Republic
© OECD/IEA, 2014
Spain
International Energy Agency
Sweden
9 rue de la Fédération
Switzerland
75739 Paris Cedex 15, France
Turkey
www.iea.org
United Kingdom
United States
Please note that this publication

is subject to specific restrictions that limit its use and distribution.
The terms and conditions are available online at http://www.iea.org/termsandconditionsuseandcopyright/ The European Commission also participates in the work of the IEA.

Foreword
Current trends in energy supply and use are unsustainable – economically, environmentally and socially. Without decisive action, energy-related greenhouse-gas (GHG) emissions would lead to considerable climate degradation with an average
6°C global warming. We can and must change the path we are now on; sustainable and low-carbon energy technologies will play a crucial role in the energy revolution required to make this change happen. Energy Efficiency, many types of renewable energy, carbon capture and storage (CCS), nuclear power and new transport technologies will all require widespread deployment if we are to achieve a global energy-related CO2 target in 2050 of 50% below current levels and limit global temperature rise by 2050 to 2°C above pre-industrial levels.
This will require significant global investment into decarbonisation, which will largely be offset by reduced expenditures on fuels. Nonetheless, this supposes an important reallocation of capital. To address this challenge, the International Energy
Agency (IEA) is leading the development of a series of technology roadmaps which identify the steps needed to accelerate the implementation of technology changes. These roadmaps will enable governments, industry and financial partners to make the right choices – and in turn help societies to make the right decision.
Photovoltaic (PV) energy is one of the most promising emerging technologies. The cost of PV modules has been divided by five in the last six years; the cost of full PV systems has been divided by almost three. The levelised cost of electricity of decentralised solar PV systems is approaching or falling below the variable portion of retail electricity prices that system owners pay in some markets, across residential and commercial segments. For bulk power on grid, PV electricity can already be competitive at times of peak demand, especially in areas where peak electricity is provided by burning oil products. And there remains ample room for improvements, as this roadmap details.
Much has happened since our 2010 IEA technology roadmap for PV energy. PV has been deployed faster than anticipated and by 2020 will probably reach twice the level previously expected. Rapid

deployment and falling costs have each been driving the other. This progress, together with other important changes in the energy landscape, notably concerning the status and progress of nuclear power and CCS, have led the IEA to reassess the role of solar PV in mitigating climate change. This updated roadmap envisions PV’s share of global electricity rising up to 16% by 2050, compared with 11% in the
2010 roadmap.
As PV spreads beyond Europe, where most deployment was concentrated until 2012, it faces a number of barriers, economic and non-economic.
To help overcome such potential obstacles, this updated roadmap provides renewed proposals on technology, system integration, legislative and regulatory issues, based on analyses of the lessons learned by pioneering countries.
In mature PV markets – currently still only a handful of countries – greater market exposure is necessary as PV becomes more competitive. However, changes in legislative frameworks and support policies must be as transparent and predictable as possible. Like most renewable energy sources and energy efficiency improvements, PV is very capital-intensive: almost all expenditures are made up-front. Keeping the cost of capital low is thus of primary importance for achieving this roadmap’s vision. But investment and finance are very responsive to the quality of policy making. Clear and credible signals from policy makers lower risks and inspire confidence. By contrast, where there is a record of policy incoherence, confusing signals or stop‐and‐go policy cycles, investors end up paying more for their finance, consumers pay more for their energy, and some projects that are needed simply will not go ahead.
I strongly hope that the analysis and recommendations in this roadmap will play a part in ensuring the continued success of PV deployment and, more broadly, a decarbonised energy system.
This publication is produced under my authority as
Executive Director of the IEA.
Maria van der Hoeven
Executive Director
International Energy Agency

This publication reflects the views of the International Energy Agency (IEA) Secretariat but does not necessarily reflect those of individual IEA member countries. The IEA makes no representation or warranty, express or implied, in respect to the publication’s contents (including its completeness or accuracy) and shall not be responsible for any use of, or reliance on, the publication.

Foreword

1

Table of contents
Foreword

1

Table of contents

2

Acknowledgements

4

Key findings and actions

5

Key actions in the next five years

6

Introduction

7

Rationale for solar photovoltaic power in the overall energy context

7

Purpose of the roadmap update

7

Roadmap process, content and structure

8

Progress since 2009

9

Recent market developments

9

Technology improvements

11

Advances toward competitiveness

12

Barriers encountered, overcome or outstanding

16

Medium-term outlook

17

Vision for deployment

18

CO2 reduction targets from the ETP 2014 hi-Ren Scenario

18

Revised solar PV goals

20

Potential for cost reductions

22

Global investment to 2050

25

Beyond 2050

26

Solar PV technology development: Actions and milestones

27

Cells and modules

28

Non-module costs

30

Life-cycle analysis

31

System integration: Actions and milestones

32

Variability and uncertainty

32

System-friendly PV deployment

32

Integrating large PV shares

33

Decentralised PV generation

37

System level GHG emissions

41

Removing non-economic barriers

41

Setting predictable financial schemes and regulatory frameworks

42

Policy options

43

Financing

47

International collaboration

49

Roadmap action plan

51

Near-term actions for stakeholders

51

Implementation

52

Abbreviations and acronyms

53

References

2

39

Policy, finance and international collaboration: Actions and milestones

54

Technology Roadmap Solar photovoltaic energy

List of tables
Table 1. Progress in solar PV markets and installation since 2009

10

Table 2. Typical PV system prices in 2013 in selected countries

15

Table 3. PV capacities by region in 2030 and 2050 in the hi-Ren Scenario

20

Table 4. Projections for LCOE for new-built utility-scale PV plants to 2050 in the hi-Ren Scenario

24

Table 5. Projections for LCOE for new-built rooftop PV systems to 2050 in the hi-Ren Scenario

24

Table 6. CO2 prices in the climate-friendly scenarios of ETP 2014

26

List of figures
Figure 1. Global cumulative growth of PV capacity

9

Figure 2. PV manufacturing by countries

10

Figure 3. System prices in Italy, 2008-2013

14

Figure 4. Grid parity was reached in 2013 in various countries

15

Figure 5. Global electricity mix in 2011 and in 2050 in three ETP 2014 scenarios

18

Figure 6. Cumulative technology contributions to power sector emission reductions

in ETP 2014 hi-Ren Scenario, relative to 6DS, up to 2050

19

Figure 7. Additional CO2 emission reductions due to PV in 2050 in the hi-Ren Scenario (over the 6DS)

20

Figure 8. Regional production of PV electricity envisioned in this roadmap

21

Figure 9. Generation mix by 2050 in the hi-Ren Scenario by region (in annual energy)

22

Figure 10. Past modules prices and projection to 2035 based on learning curve

23

Figure 11. PV investments cost projections in the hi-Ren Scenario

23

Figure 12. The share of the costs of capital in the LCOE of PV systems

25

Figure 13. Commercial 1-sun module efficiencies (actual and expected)

27

Figure 14. How different climate zones in the United States affect the lifetime of PV modules

29

Figure 15. Hourly planned versus actual solar generation in Germany, 2013

33

Figure 16. Weekly production of solar and wind in Germany in 2013

34

Figure 17. Expected evolution of the net load of a typical spring day in California

34

Figure 18. Controlled versus uncontrolled EV charging effects on load net of PV

36

Figure 19 Self-consumption of stylised household and rooftop PV system during a sunny day

37

Figure 20. Hourly electricity consumption profiles for different building types in Germany

38

Figure 21 Increasing self-consumption with load management (+10%) and small storage (+10%)

39

Figure 22. Feed-in tariff levels and LCOE of small rooftop in Germany

44

List of boxes
Box 1. Efficiency and performance ratio

12

Box 2. Solar radiation relevant for PV

13

Box 3. ETP Scenarios. 6DS, 2DS, hi-Ren

19

Box 4. Sustainable PV energy for all

25

Box 5. “Soft costs” in the United States

42

Box 6. Value-of-solar tariffs in Minnesota

46

Box 7. Financing off-grid solar electrification

48

Table of contents

3

Acknowledgements
This publication was prepared by the Renewable
Energy Division (RED) of the International Energy
Agency (IEA). Cédric Philibert was the main author of this update, based on the original published in 2010. Paolo Frankl, Head of RED, provided important comments and inputs. Cecilia Tam, in her role as Technology Roadmap Co-ordinator, made significant contributions throughout the drafting process. Several other IEA colleagues also provided important contributions, in particular
Yasmina Abdelilah, Heymi Bahar, Quentin Marchais,
Simon Mueller, Uwe Remme, Michael Waldron and
Hoël Wiesner. The author is also grateful to Keisuke
Sadamori, Director of Energy Markets and Security at the IEA, for his guidance.
The author would also like to thank Andrew
Johnston for skilfully editing the manuscript; the
IEA Publication Unit, in particular Muriel Custodio,
Therese Walsh and Astrid Dumond; and Bertrand
Sadin for executing the layout.
Finally, this roadmap would not be effective without all of the comments and support received from the industry, government and non-government experts who attended the workshop at IEA headquarters in
Paris on February 3, 2013, reviewed and commented on the drafts, and provided overall guidance and support. The authors wish to thank all of those who contributed through discussions and early comments, in particular: Fabrizio Bizzarri (Enel),
Frédéric Bouvier (Soitec), Jenny Chase (BNEF),

4

Paul Denholm (NREL), Paul Gauché (Stellenbosch
University), Winfried Hoffmann (ASE), Véronique
Jolivet (Total Energies Nouvelles), Thierry Lepercq
(Solairedirect), Daniel Lincot (IPVF-IRDEP), Philippe
Malbranche (CEA-INES), Gaëtan Masson (EPIA),
Johannes Mayer (Fraunhofer-ISE), Paula Mints (SPV
Market Research), Stefan Nowak (PVPS) and Wim
Sinke (ECN).
Review comments were received from Luca
Benedetti (GSE), Simon Philipps (Fraunhofer-ISE),
Paul Denholm (NREL), Winfried Hoffmann (ASE),
Arnulf Jaeger-Waldau (EU JRC), Veronique Jolivet
(Total Energies Nouvelles), Jean-Pierre Joly (CEAINES), Daniel Kammen (University of California),
Manoël Rekinger (EPIA), Ioannis-Thomas Theologitis
(EPIA), Lou Trippel (First Solar), Wim Sinke (ECN),
Roberto Vigotti (IEA REWP) and Eric Westerhoff
(Soitec).
This publication was made possible also thanks to the support of Soitec and the French government through ADEME.
For more information on this document, contact:
Technology Roadmaps
International Energy Agency
9, rue de la Fédération
75739 Paris Cedex 15
France
Email: TechnologyRoadmapsContact@iea.org

Technology Roadmap Solar photovoltaic energy

Key findings and actions zz  ince 2010, the world has added more solar
S
photovoltaic (PV) capacity than in the previous four decades. New systems were installed in 2013 at a rate of 100 megawatts (MW) of capacity per day. Total global capacity overtook 150 gigawatts
(GW) in early 2014. zz  he geographical pattern of deployment
T
is rapidly changing. While a few European countries, led by Germany and Italy, initiated large-scale PV development, PV systems are now expanding in other parts of the world, often under sunnier skies. Since 2013, the People’s
Republic of China has led the global PV market, followed by Japan and the United States. zz  V system prices have been divided by three in
P
six years in most markets, while module prices have been divided by five. The cost of electricity from new built systems varies from USD 90 to
USD 300/MWh depending on the solar resource; the type, size and cost of systems; maturity of markets and costs of capital. zz  his roadmap envisions PV’s share of global
T
electricity reaching 16% by 2050, a significant increase from the 11% goal in the 2010 roadmap.
PV generation would contribute 17% to all clean electricity, and 20% of all renewable electricity.
China is expected to continue leading the global market, accounting for about 37% of global capacity by 2050. zz  chieving this roadmap’s vision of 4 600 GW of
A
installed PV capacity by 2050 would avoid the emission of up to 4 gigatonnes (Gt) of carbon dioxide (CO2) annually. zz  his roadmap assumes that the costs of electricity
T
from PV in different parts of the world will converge as markets develop, with an average cost reduction of 25% by 2020, 45% by 2030, and
65% by 2050, leading to a range of USD 40 to
160/MWh, assuming a cost of capital of 8%. zz  o achieve the vision in this roadmap, the total PV
T
capacity installed each year needs to rise rapidly, from 36 GW in 2013 to 124 GW per year on average, with a peak of 200 GW per year between
2025 and 2040. Including the cost of repowering
– the replacement of older installations – annual investment needs to reach an average of about
USD 225 billion, more than twice that of 2013. zz  tility-scale systems and rooftop systems will
U
each have roughly half of the global market.
Rooftop systems are currently more expensive but the value of electricity delivered on consumption sites or nearby is greater. However, as PV expansion is driven more and more by self-

consumption – the use of PV electricity directly at the same site where it is generated – grids may carry smaller amounts of traded electricity, raising concerns over how to recover the fixed costs of grids. Grid operators, regulators and policy makers should monitor the impact of rapid expansion of distributed PVs on distribution networks. Rate changes ensuring full grid cost recovery and fair allocation of costs might be considered but should be carefully designed in order to maintain incentives for energy efficiency and the deployment of rooftop PV. zz  he variability of the solar resource, as of wind
T
energy, is a challenge. All flexibility options
– including interconnections, demand-side response, flexible generation, and storage –need to be developed to meet this challenge so that the share of global electricity envisioned for PV in this roadmap can be reached by 2050. zz  V has to be deployed as part of a balanced
P
portfolio of all renewables. In temperate countries, wind power tends to be stronger during winter and hence compensate for low solar irradiance. In hot and wet countries, hydropower offers considerable resource in complement to solar PV. In hot and arid countries, solar thermal electricity with built-in thermal storage capabilities can generate electricity after sunset, complementing the variability of PV and thus adding more solar electricity to systems – potentially making solar the leading source of electricity by 2040. 1 zz  espite recent falls in the cost of PV electricity,
D
transitional policy support mechanisms will be needed in most markets to enable PV electricity costs to reach competitive levels, as long as electricity prices do not reflect climate change or other environmental factors. The vision in this roadmap is consistent with global CO2 prices of
USD 46/tCO2 in 2020, USD 115/tCO2 in 2030, and
USD 152/tCO2 in 2040. zz n the last few years, manufacturing of PV systems
I
has been concentrated in Asia, particularly in
China and Chinese Taipei, mainly based on economies of scale in large new production facilities. Future progress is likely to be driven mainly by technology innovation, which keeps open the possibility of global deployment of manufacturing capabilities if research and development (R&D) efforts and international collaboration are strengthened.
1. See the companion Technology Roadmap: Solar Thermal Electricity

(IEA, 2014a).

Key findings and actions

5

zz  ppropriate regulatory frameworks – and wellA designed electricity markets, in particular – will be critical to achieve the vision in this roadmap.
PV costs are incurred almost exclusively up-front, when the power plant is built. Once built, PV generates electricity almost for free. This means that investors need to be able to rely on future revenue streams so that they can recover their initial capital investments. Market structures and regulatory frameworks that fail to provide robust long-term price signals – beyond a few months or years – are thus unlikely to deliver investments in volumes consistent with this roadmap in particular and timely decarbonisation of the global energy system in general.

Key actions in the next five years zz  et or update long-term targets for PV
S
deployment, consistent with national energy strategies and national contributions to global climate change mitigation efforts. zz  upport these targets with predictable market
S
structures and regulatory frameworks to drive investment. zz  ddress non-economic barriers. Develop
A
streamlined procedures for providing permits. zz dentify the cost structure of current projects and
I
any anomalies in comparison with projects in other jurisdictions. Implement specific actions to reduce anomalous costs. zz  ork with financing circles and other
W
stakeholders to reduce financing costs for PV deployment, in particular involving private money and institutional investors. zz  educe the costs of capital and favour innovation
R
in providing loan guarantees, and concessional loans in emerging economies. zz  trengthen research, development and
S
demonstration (RD&D) efforts to further reduce costs. zz  trengthen international collaboration on RD&D
S
and exchanges of best practices. zz In emerging PV markets: zz mplement priority connection to the grid and
I
priority dispatch of PV electricity.

6

zz mplement support schemes with fair
I
remuneration for investors but predictable decrease of the level of support. zz  hen parity with retail electricity prices is
W
achieved in some market segments, provide incentives for distributed PV generation through net energy metering and/or tariffs for energy (total generation or only injections into the grid) based on a value of solar electricity determined through a transparent process open to all interested parties. zz In mature markets: zz  rogressively increase short-term market
P
exposure of PV electricity while ensuring fair remuneration of investment, for example with sliding feed-in premiums and/or auctions with time-of-delivery and locational pricing. zz  rovide incentives for generation at peak
P
times through time-of-delivery payments. zz  rovide incentives for self-consumption during
P
peaks through time-of-use electricity rates. zz mprove forecasts and reform energy-only
I
electricity markets for better synchronisation of supply and demand. zz  esign and implement investment markets for
D
new-built PV systems and other renewables, and markets for ancillary services. zz  rogressively reform rate structures to
P
encourage generation and discourage consumption during peak times, ensuring the recovery of fixed costs of the transmission and distribution grids while preserving the incentives for efficiency and distributed PV. zz  void retroactive legislative changes.
A
zz  ork with financing circles and other
W
interested parties to reduce financing costs for
PV deployment, in particular involving private money and institutional investors. zz  trengthen research, development and
S
demonstration (RD&D) efforts to further reduce costs. zz mprove quality via more diversified module
I
qualification, and certification of developers, designers and installers. zz  trengthen international collaboration on RD&D
S
and exchanges of best practices. zz  upport best practices in developing economies,
S
in particular for providing access to electricity based on off-grid and mini-grid PV systems.

Technology Roadmap Solar photovoltaic energy

Introduction
There is a pressing need to accelerate the development of advanced energy technologies in order to address the global challenges of clean energy, climate change and sustainable development. To achieve the necessary reductions in energy-related CO2 emissions, the IEA has developed a series of global technology roadmaps, under international guidance and in close consultation with industry. These technologies are evenly divided among demand-side and supply-side technologies and include several renewable energy roadmaps (www.iea.org/roadmaps/).
The overall aim is to advance global development and uptake of key technologies to limit the global mean temperature increase to 2 degrees Celsius
(°C) in the long term. The roadmaps will enable governments, industry and financial partners to identify and implement the measures needed to accelerate the required technology development and uptake.
The roadmaps take a long-term view, but highlight the key actions that need to be taken in the next five years, which will be critical to achieving long-term emission reductions. Existing conventional plants and those under construction may lock in CO2 emissions, as they will be operating for decades.
According to the IEA Energy Technology Perspectives
2014 (ETP 2014) (IEA, 2014b), early retirement of
850 GW of existing coal capacity would be required to reach the goal of limiting climate change to
2°C. Therefore, it is crucial to build up low-carbon energy supply today.

Rationale for solar photovoltaic power in the overall energy context
ETP 2014 projects that in the absence of new policies to accelerate the uptake of low-carbon solutions, CO2 emissions from the energy sector would increase by 61% over 2011 levels by 2050
(IEA, 2014b). The ETP 2014 model examines a range of technology solutions that can contribute to preventing this increase: greater energy efficiency, renewable energy, nuclear power and the neardecarbonisation of fossil fuel-based power generation. Rather than projecting the maximum possible deployment of any given solution, the
ETP 2014 model calculates the least-cost mix to achieve the CO2 emission reduction needed to limit climate change to 2°C (the ETP 2014 2°C Scenario

[2DS]). The hi-Ren Scenario is a variant of the 2DS with slower deployment of nuclear and carbon capture and storage (CCS) technologies, and more rapid deployment of renewables, notably solar and wind energy.
Based on the ETP 2014 hi-Ren Scenario, this roadmap envisions up to 16% of global electricity for solar
PV with 6 300 TWh generated in 2050, up from the 4 500 TWh foreseen in the 2010 roadmap. This increase in PV compensates for slower progress in the intervening years in CCS and higher costs for nuclear power. It also reflects faster-than-expected rollout and cost reductions for solar PV.
Solar energy is widely available throughout the world and can contribute to reduced dependence on energy imports. As it entails no fuel price risk or constraints, it also improves security of supply.
Solar power enhances energy diversity and hedges against price volatility of fossil fuels, thus stabilising costs of electricity generation in the long term.
Solar PV entails no greenhouse gas (GHG) emissions during operation and does not emit other pollutants (such as oxides of sulphur and nitrogen); additionally, it consumes no or little water. As local air pollution and extensive use of fresh water for cooling of thermal power plants are becoming serious concerns in hot or dry regions, these benefits of solar PV become increasingly important.

Purpose of the roadmap update
The solar PV roadmap was one of the first roadmaps developed by the IEA, in 2009/10. Since then, the world has added more PV capacity than it had in the previous four decades, and more rapidly than expected. The 210 GW of cumulative capacity expected to be reached by 2020 is now likely to be achieved five years earlier, and the capacity now expected for 2020 will be over twice what was foreseen in the 2010 roadmap. Moreover, the system cost milestones for 2020 in the original roadmap have already been reached in the most advanced markets, except for the smallest rooftop capacities.
This updated roadmap thus presents a new vision that takes into account this considerable progress of PV technologies, as well as changing trends in the overall energy mix. It presents a detailed assessment of the technology milestones that PV energy will need to reach to attain this

Introduction

7

ambitious vision. The key objective is to improve PV technology performance and reduce costs in order to achieve the competitiveness needed for the large investments foreseen.
The roadmap also examines numerous economic and non-economic barriers that hamper deployment and identifies policy actions to overcome them. About half of overall capacity is likely to be deployed on buildings, so this roadmap considers critical issues arising from the complex relationships between PV generation, on-the-spot consumption and electricity networks.
This roadmap thus identifies actions and time frames to achieve the higher PV deployment needed for global emission reductions. In some markets, certain actions have already been taken, or are under way. Many countries, particularly in emerging regions, are only just beginning to develop PV systems. Accordingly, milestone dates should be considered as indicative of urgency, rather than as absolutes. Each country will have to choose which actions to prioritise, based on its mix of energy sources and industrial policies.
This roadmap is addressed to a variety of audiences, including policy makers, industry, utilities, researchers and other interested parties. As well as providing a consistent overall picture of PV power at global and continental levels, it aims at providing encouragement and information to individual countries to elaborate action plans, set or update targets, and formulate roadmaps for PV power deployment. The roadmap also takes into account other regional and national efforts to investigate the potential of
PV, including: zz  he SunShot initiative of the US Department t of Energy zz  he EU Strategic Energy Technology Plan t (SET Plan) zz  he international technology roadmap t for PV (ITRPV) zz  he Chinese 12th five-year plan for the solar t PV industry.
This roadmap is organised into six major sections.
First, the current state of the PV industry and progress since 2009 is discussed, followed by a section that describes the vision for PV deployment between 2015 and 2050 based on ETP 2014. This discussion includes information on the regional distribution of PV generation projects and the associated investment needs, as well as the potential for cost reductions.
The next three sections describe approaches and specific tasks required to address the major challenges facing large-scale PV deployment in three major areas: PV technology development; system integration; and policy frameworks, public engagement and international collaboration. The final section sets out next steps and categorises the actions in the previous sections that policy makers, industry, power system actors and financing circles need to take to implement the roadmap’s vision for
PV deployment.

Roadmap process, content and structure
This roadmap was developed with the help of contributions from representatives of the solar industry, the power sector, research and development (R&D) institutions, the finance community and government institutions. An expert workshop was held in Paris in February 2014 at IEA headquarters in Paris, focusing on technology and
“vision” for both solar PV and STE 2. A draft was then circulated to experts and stakeholders for further contributions and comments.

2.  ee www.iea.org/workshop/solarelectricity
S
roadmapworkshop.html.

8

Technology Roadmap Solar photovoltaic energy

Progress since 2009
Recent market developments The PV industry has experienced a sea change in only five years, with considerable increases in manufacturing capacities, and a move of module manufacturing from European countries and the
United States to Asia, notably China and Chinese
Taipei. Market prices have been drastically reduced
– by factor of five for modules, and by a factor of almost three for systems. The global rate of annual new-built capacities, which was 7 GW in 2009, was
5 times higher in 2013.

In the last ten years, cumulative installed capacity has grown at an average rate of 49% per year
(Figure 1). In 2013, about 37 GW of new PV capacity was installed in about 30 countries – or 100 MW per day – bringing total global capacity to over
135 GW. For the first time since 2004, more new capacity was installed in Asia than in Europe. China alone installed more than all of Europe, with over
11 GW. Japan ranked second with almost 7 GW, and the United States third with over 4 GW. New investment in PV capacity in 2013 was assessed at
USD 96 billion.

Figure 1: Global cumulative growth of PV capacity
80%

140

70%

120

60%

100

50%

80

40%

60

30%

40

20%

20

10%

0

Annual growth

Cumulative PV capacity (GW)

160

0%
2003

2004
Germany
Japan

2005

2006

2007

Italy
United States

2008
Spain
China

2009

2010

2011

France
Rest of world

2012

2013

Rest of Europe
Annual growth

Source: Unless otherwise indicated, all tables and figures derive from IEA data and analysis.

KEY POINT: Cumulative PV capacity grew at 49%/yr on average since 2003.

Grid-connected PV systems continue to be built at all scales, from just a few kilowatts (kW) to hundreds of megawatts (MW). Off-grid systems can be even smaller while providing highly valued power far from electricity networks. At the opposite end, there are about 20 utility-scale plants of over
100 MW capacity in the world, mostly in China and in the United States.

consumption (Burger, 2014), and total capacity was rated at 36 GW at the end of 2013. In Italy,
PV systems generated 22 TWh in 2013, or 7% of electricity consumption, with total capacity rated at 17 GW at the end of 2013. PV generation has exceeded 3% of electricity demand in five other countries – Belgium, Bulgaria, Czech Republic,
Greece and Spain (PVPS IA, 2014; RED, 2014).

In Germany, more than 1.3 million solar power plants generated almost 30 TWh in 2013, equivalent to 5.3% of German electricity

Crystalline silicon (c-Si) modules, whether single(sc-Si) or multi-crystalline (mc-Si), currently dominate the PV market with around 90% share.

Progress since 2009

9

Thin films (TF) of various sorts now represent only about 10% of the market, down from 16% in 2009, and concentrating photovoltaics (CPV), although growing significantly, represent less than 1%.
Decentralised systems represent approximately 60% of the global market, while centralised, utility-scale systems represent close to 40%. Off-grid systems, which once dominated a much smaller market, now account for 1% at most.

In the last few years, the PV module manufacturing industry has witnessed a dramatic shift, from
Europe, particularly Germany, to Asia, mostly China and Chinese Taipei.

Table 1: Progress in solar PV markets and installation since 2009
End of 2009

End of 2013

Total installed capacity

23 GW

135 GW

Annual installed capacity

7 GW

37 GW

USD 48 billion

USD 96 billion

Number of countries with >1 GW installed

5

17

Number of countries with >100 MW yearly market

9

23

PV electricity generated during the year

20 TWh

139 TWh

PV penetration levels

% of yearly electricity consumption

Annual investment

Europe zz Germany zz Italy

2.6%
5.3%
7%

Figure 2: PV manufacturing by countries
45 000
40 000
35 000
30 000
25 000
20 000
15 000
10 000
5 000
0
2005
United States

2006
Europe

2007
Japan

2008
China

2009

Chinese Taipei

2010
Rest of world

2011

2012

2013

Total manufacturing capacity

Source: SPV Market Research (2014), Photovoltaic manufacturer Shipments: Capacity, Price & Revenues 2013/2013,
Report SPV-Supply 2, April.

KEY POINT: The manufacturing of PV modules shifted from OECD to non-OECD countries over just a few years.

10

Technology Roadmap Solar photovoltaic energy

In 2012, the PV industry employed approximately
88 000 people in Germany to manufacture materials (silicon, wafers, metal pastes, plastic films, solar glass), and intermediate and final products
(cells, modules, inverters, supports, cables, coated glass), and to build manufacturing plants and install
PV systems. About a quarter of this production has since been lost, as the German PV market declined and as manufacturing moved to Asia. This shift was primarily a result of huge investments in production capacity in Asia. Labour costs only played a marginal role, as PV production is highly automated; other variable costs, including energy costs, played a more important role. Fast technology transfer was facilitated by the availability of turn-key production lines producing very good quality PV modules.
Manufacturing PV cells and modules now accounts for less than half the value chain, however, although it remains the largest single element. Upstream activities, from research and development (R&D) to building production lines, and downstream activities linked to installation and services, together account for the largest part. In many
European countries, in particular those where PV manufacturing has never been important, such as Spain, thousands of PV-related jobs have been created and eliminated more because of on-again off-again renewable energy policies than because of competition from Asia.
At a global level, the PV industry has been estimated to represent about 1.4 million full-time jobs, including 300 000 to 500 000 in China, 312 000 in Europe, 112 000 in India and 90 000 in the
United States (REN21, 2014), but these figures must be taken with caution as data collection is not homogenous and, more important, may change rapidly as markets evolve.
A detailed analysis of the cost trends of c-Si modules and the shift of manufacturing to China suggests that the historical price advantage of a China-based factory over a US-based factory is driven not by country-specific factors, but by scale, supply-chain development and access to finance. Technology innovations may result in effectively equivalent minimum sustainable manufacturing prices for the two locations (Goodrich et al., 2013) – and this may hold true for many other locations.

Technology improvements
PV cells are semiconductor devices that generate direct current (DC) electricity. Silicon cells are usually sliced from ingots or castings of highly purified silicon. The manufacturing process creates a charge- separating junction, deposits passivation layers and an anti-reflective coating, and adds metal contacts. Cells are then grouped into modules, with transparent glass for the front, a weatherproof material for the back and often a surrounding frame.
The modules are then combined to form strings, arrays and systems.
PV can be used for on-grid and off-grid applications of capacities ranging from less than 1 watt to gigawatts. Grid-connected systems require inverters to transform DC power into alternating current (AC). The balance of system (BOS) includes inverters, transformers, wiring and monitoring equipment, as well as structural components for installing modules, whether on building rooftops or facades, above parking lots, or on the ground.
Installations can be fixed or track the sun on one axis (for non- or low-concentrating systems) or two axes (for high-concentrating systems).
Alternative PV technologies, including thin films, had been expected to gain an increasing share of the market, but instead their share shrank from 15% in 2009 to about 10% in 2013. Thin films (TF) are based on cadmium telluride (CdTe), copper-indiumgallium-selenide (CIGS), or amorphous silicon (a-Si), plus some variants. They are usually manufactured in highly automated processes to produce complete modules, with no need to assemble modules from individual cells. Multi-junction cells, which are the standard PV technology in space applications, recently entered the terrestrial market in concentrating photovoltaics (CPV) systems with several large-scale plants (50 MW each) in operation or under construction. Some manufacturers also sell hybrid PV-thermal panels that deliver both heat and electricity. Progress since 2009

11

Box 1: Efficiency and performance ratio
The output per watt (W) installed (sometimes termed “watt-peak”, or Wp) does not depend on nominal efficiency, which determines the required receptive area per watt. The nominal efficiency relates to the power generated under so-called “standard test conditions” (STC) – module temperature of 25°C, vertical irradiance of 1 000 W/m2, air mass of 1.5 (distance travelled through the atmosphere 50% greater than when the sun is exactly overhead) and a specific irradiance spectrum. For example, modules of 1 m2 would generate a maximum power of 150 W with 15% efficiency, and 200 W with 20% efficiency under the STC.

takes into account all efficiency losses resulting from actual module temperature, module mismatch, varying irradiance conditions, dirt, line resistance and conversion losses in the inverter. Well-designed PV plants achieve average PR of 80% to 90% throughout the year.
A PV system receiving 1 200 kWh/m2 per year would generate annually 1 020 kWh/kW with a PR of 85%, the equivalent of 1 020 full load hours, or a capacity factor of 11.6%. In regions with high insolation of 2 500 kWh/m2 on the collector surface, the same system, assuming the same PR of 85%, would produce up to
2 125 kWh/kW, or a capacity factor of 24.3%.

The actual output depends on the solar resource, the orientation of the modules and the
“performance ratio” (PR) of the system, which

The average efficiency of commercial silicon modules has improved in the last ten years by about 0.3% per year, reaching 16% in 2013. The best-performing commercial modules, based on back-junction, interdigitated back-contact (IBC) offer efficiencies of 21%, with heterojunction (HTJ) technologies close behind at over 19% efficiency, and excellent performance ratios. Modules are usually guaranteed for a lifetime of 25 years at minimum 80% of their rated output, and sometimes for 30 years at 70%. TF modules also saw increases in efficiencies, with commercial CdTe TF, in particular, reaching 15%. Moreover, CdTe modules, especially in hot and humid climates, and possibly
CIGS TF modules, have higher performance ratios than average c-Si modules of similar prices. CPV modules offer efficiencies of 25% to 35%, but only make use of direct normal irradiance (DNI), which is lower than global normal irradiance (see Box 2).
Therefore CPV performs best in high DNI locations.
Even more impressive progress has been made with respect to manufacturing. The amount of specific materials (silicon, metal pastes, etc.), the energy consumption and the amount of labour required to assemble modules were all significantly reduced.

12

Advances toward competitiveness The emergence of the global PV market has coincided with rapid reductions in the costs of modules and systems. The levelised cost of energy
(LCOE)3 from PV systems is already below retail electricity prices (per-kWh charge) in several countries, and rapidly approaching the level of generation costs from conventional alternatives, especially new coal with strict air pollutant emission standards, new nuclear plants with increased safety standards, or new gas plants in areas with high gas prices. Investment costs
From 2008 to 2012, PV module prices were divided by five, and PV system prices divided by three in mature markets such as Italy (Figure 3). In 2013 and 2014, module prices more or less stabilised, indicating that prices were not fully reflecting underlying costs. As too many modules began
3. The LCOE represents the present value of the total cost

(overnight capital cost, fuel cost, fixed and variable operation and maintenance costs, and financing costs) of building and operating a generating plant over an assumed financial life and duty cycle, converted to equal annual payments, given an assumed utilisation, and expressed in terms of real money to remove inflation.

Technology Roadmap Solar photovoltaic energy

to be produced, many were sold at prices too low to recover investment, as the deterioration of the balance sheets of most PV companies
(up to bankruptcy for some) demonstrated. But

improvements in technology and the scaling up of manufacturing were by far the main factors driving cost reductions.

Box 2: Solar radiation relevant for PV
Solar energy is the most abundant energy resource on earth, with about 885 million terawatt hours (TWh) reaching the surface of the planet every year – 6 200 times the commercial primary energy consumed by humankind in 2008, and 3 500 times the energy that humankind would consume in
2050 according to the ETP 2014 6-degree
Scenario (IEA, 2011; 2014b).
The solar radiation reaching the earth’s surface is about 1 kilowatt per square metre (kW/‌ 2) m in clear conditions when the sun is near the zenith. It has two components: direct or
“beam” radiation, which comes directly from the sun’s disk; and diffuse radiation, which comes indirectly after being scattered by the atmosphere. PV systems, with the exception of concentrating PV (CPV), make use of the
“global” irradiance, which is the sum of direct and diffuse radiations.
All places on earth receive 4 380 daylight hours per year — i.e. half the total duration of one year. Different areas, however, receive different yearly average amounts of energy from the sun.
When the sun is lower in the sky, its energy is spread over a larger area, and more is also lost when passing through the atmosphere, because of increased air mass; it is therefore weaker per horizontal surface area: inter-tropical areas should thus receive more radiation per land area than places north of the Tropic of Cancer or south of the Tropic of Capricorn. However,

Production of PV modules in China has stimulated competition and reduced prices. In the United
States, however, the installed price of Chinese and non-Chinese modules was roughly the same for any given module efficiency (Barbose et al., 2013).
In the first half of 2014, Chinese Tier 1 module players were selling at USD 0.59-0.60/W in China,

atmospheric absorption characteristics affect the amount of this surface radiation significantly, and the sunniest places on Earth are usually arid and semi-arid areas close to the tropics but distant from the Equator.
The average energy received in Europe, measured in global horizontal irradiance (GHI), is about 1 200 kilowatt hours per square metre per year (kWh/m2/y). This amount compares with 1 800 kWh.m2/y to 2 300 kWh/m2/y in the
Middle East. The United States, Africa, most of Latin America, Australia, most of India, and parts of China and other Asian countries also have good to excellent solar resource; these are broadly the regions where energy demand is expected to rise most in the coming decades.
Alaska, Northern Europe, Canada, Russia and
Southeast China receive less solar energy.
But tilting equator-facing modules can reduce disparities and increase the annual energy received on PV systems, especially at high latitudes, although this varies with meteorological patterns and the ratio of diffuse versus direct light. For example, modules in La Rochelle, in France, where the GHI is
1 300 kWh/m2/y, receive up to 1 500 kWh/m2/y if optimally tilted and oriented. Tracking the sun on one axis or two axes further increases the amount of energy receives by the modules.
Global normal irradiance (GNI) is the relevant resource for two-axis sun-tracking “1-sun” (i.e. non-concentrating) PV systems.

and USD 0.67-0.79/W in other countries (Bnef,
2014). German modules were selling at
EUR 0.69 (USD 0.95)/W.
The learning experience for complete PV systems is usually considered slower than that for modules and other hardware parts (inverters, support structures,

Progress since 2009

13

cables, etc) – a phenomenon with both national and global dimensions. However, in emerging markets, non-module costs often shrink rapidly as installers gain experience — and also as project density increases, saving significant travel times for sales and marketing staff, and skilled workers.

In 2013, the cheapest large-scale, ground-mounted
PV systems could cost less than USD 1.50/W, a price that most market analysts expected, just two years previously, to apply in 2019 or 2017 at the earliest.
Although module prices seem to have stabilised in 2013, system costs have continued to decline, with cost reductions in California, for example, ranging from 10% to 15% depending on system size in the first half of 2013 (Barbose et al., 2013).
Both the investment cost difference and the output gap between fixed-tilted PV systems and one-axis sun-tracking systems have narrowed in the last few years. In Japan, costs of residential PV systems fell from USD 5.9/W in 2012 to USD 4.64/W in 2013 – a
21% reduction.

The costs of PV systems have fallen considerably over the last six years in several markets. In Italy, prices for non-module components of PV systems dropped significantly (Figure 3). In other countries, notably in the United States, the reductions were much smaller, and the fall in module costs was the main driver of the decline in system costs up to 2012.

Figure 3: System prices in Italy, 2008-2013
6

EUR/W

5
4
3
2

1-3 kW

20-200 kW
Modules

Inverters

200-1000 kW

2013

2011

2012

2009

2010

2008

2013

2012

2011

2010

2009

2008

2013

2011

2012

2009

3-20 kW

2010

2013

2008

2011

2012

2009

2010

2008

2013

2012

2011

2010

2009

0

2008

1

Above 1000 kW

Others

Source: Gestore dei Servizi energetici (GSE) (2014), PV in Italy: Generation Costs and Value Chain, May, Rome.

KEY POINT: In 2013, PV systems in Italy cost 30% to 44% of what they cost in 2008.

Prices for entire PV systems range more widely than those of cells and modules, which tend to be global commodities. Small systems, such as rooftop systems, are usually more expensive than larger ones, especially ground-based, utility-scale systems
(Australia and China being possible exceptions due to connection costs). Prices vary significantly among countries for similar system types (Table 2).
Most of the gap comes from differences in “soft costs”, which include customer acquisition; permitting, inspection and interconnection; installation labour; and financing costs, especially

14

for small systems (Seel et al., 2013). Generous incentive frameworks in some countries keep prices higher than raw costs plus a reasonable margin.
Even greater differences are evident in the costs of commercial PV systems from country to country; such systems are more than twice as expensive in the United States than in Germany.

Technology Roadmap Solar photovoltaic energy

Table 2: Typical PV system prices in 2013 in selected countries (USD)
Australia

China

France

Germany

Italy

Japan

United
Kingdom

United
States

Residential

1.8

1.5

4.1

2.4

2.8

4.2

2.8

4.9

Commercial

1.7

1.4

2.7

1.8

1.9

3.6

2.4

4.5

Utility-scale

2.0

1.4

2.2

1.4

1.5

2.9

1.9

3.3

USD/W

Sources: Friedman et al. (2014), Comparing PV Costs and Deployment Drivers in the Japanese and U.S. Residential and Commercial
Markets, February, NREL/TP-6A20-60360; PV-PS IA (2014a), PV Cost Data for the IEA, personal communication, January.

Levelised cost of electricity (LCOE)

relative to system costs and costs of capital would be already able to produce power for EUR 60
(USD 80)/MWh and therefore have a LCOE that is comparable to power generated from oil and gas, or even new-built coal (Kost et al., 2013). However, in countries where PV deployment has barely begun, PV system costs and costs of capital may be significantly higher, preventing PV from being immediately competitive.

PV power plants reached LCOE of EUR 78
(USD 110)/‌ Wh to EUR 142 (USD 190)/MWh in
M
the third quarter of 2013 in Germany, depending on the type of power plant and irradiance (Kost et al., 2013). At higher irradiation ranges (e.g. 1 4502 000 kWh/m2/y), assuming same system costs but slightly higher costs of capital, the LCOE from
PV in 2013 lies under EUR 120 (USD 162)/MWh for all PV power plant types. At 2 000 kWh/m2/y, PV utility-scale power plants under similar assumptions

PV can be built and operated in millions of small, decentralised systems, often characterised as
“rooftop”. When the LCOE of decentralised solar

Figure 4: Grid parity was reached in 2013 in various countries
1 100
1 000
900

USD/MWh

800
700
600
500
400
300
200
100
0
2010

2013

California

2010

2013

Australia
LCOE

2010

2013

Germany
Variable portion of tariff

2010

2013
Italy

2010

2013

Netherlands

Tiered tariffs

Note: Household electricity tariffs exclude fixed charges. LCOEs are calculated using average residential system costs (including valueadded tax and sales tax in where applicable, and investment tax credit in California); ranges mostly reflect differences in financing costs. The tiered tariffs in California are those of Pacific Gas and Electric. Tiers 3 to 4 or 5 are tariffs paid on monthly consumption when it exceeds given percentages of a set baseline. All costs and prices are in 2012 USD.

KEY POINT: Grid parity underpins PV self-consumption in Germany, and net metering in California.

Progress since 2009

15

PV systems becomes lower than the variable portion of the retail electricity price (i.e. per kWh)
(Figure 4), the situation is known as “grid parity” or “socket parity”. Grid parity provides an incentive to electricity customers to build a PV system and to generate part of the electricity they consume, and to consume part of the electricity they generate
(as more extensively discussed in the System
Integration section below). In virtually all power systems, the variable, per-kWh portion of retail prices covers energy costs, most transmission and distribution (T&D) costs, utility or grid operator margins, and various fees and taxes. Grid parity already drives part of the PV deployment in several countries. In the ETP model, the electricity from rooftop PV systems has to compete with bulk power costs from competitors, which is augmented by the
T&D costs.

Barriers encountered, overcome or outstanding
The quality of PV products has generally increased over the last few years, with reduced variance in PR
(Nowak, 2014), but as competition has intensified, some manufacturers have been able to sell lowerquality modules at very low costs. Most common defects were broken interconnections, solder bonds and diodes, or encapsulant discoloration or delamination. Other problems arose because local installers lacked the required skills or the initial design was poor. Conceiving and building
PV systems requires a variety of skills, some very specific to PV.
Standards established by International Electricity
Commission (IEC 61215 for c-Si, IEA 61646 for TF,
IEC 62108 for CPV modules) have proven useful in reducing early failure – or “infant mortality”
– of PV modules, but they were not designed to identify how modules wear out or fail in different climates and system configurations, or differentiate between products with short or long lifetimes, or quantify module lifetime for different applications or climates.
There are no widely recognised standards, norms or labels that would tell customers about the behaviour, performance and longevity of various
PV products in specific environments. Most of today’s commercial modules pass qualification tests with minimum changes required, so the tests do not provide a means of rankings. Furthermore,

16

depending on the robustness of the quality assurance system, certification of a module type may only provide insurance with respect to one module out of millions.
In several countries, notably Germany and Italy, deploying PV rapidly has created technical issues
(as well as policy cost issues, which are discussed below in the section on policy and finance). When the concentration of significant PV capacities in rural areas created “hot spots”, low-voltage grids needed to be strengthened to evacuate the power.
To mitigate the problem, in 2012 Germany revised its Renewable Energy Sources Act to oblige new
PV plants to allow remote curtailment, except for systems below 30 kW, which could instead opt for reducing feed-in to 70% of peak capacity
(IEA, 2013a).
Grid codes have created other issues. The European power grid functions at a frequency of 50 Hertz
(Hz). When more energy is fed in to the power grid than is removed from it, the grid frequency increases. Excessively high frequencies render the grid unstable. Until 2011, inverters for PV systems were equipped with an automatic switch-off function triggered at a fixed frequency of 50.2 Hz.
As the number of PV systems in Germany increased, however, this requirement meant to protect the grid could have paradoxically put its stability at risk as PV systems switched off abruptly.
To ensure network security and handle the “50.2
Hz issue”, Germany’s System Stability Act of May
2012 scheduled the retrofit of PV systems until the end of 2014. Power inverters must be able to reduce output when frequency rises too high or to turn themselves off smoothly. If PV systems do not meet the technical requirements necessary to meet this obligation, the law requires that they be switched off at different frequencies. In March 2012, Italy required that PV systems over 50 kW and connected to the medium-voltage grid carry out retrofits by the end of March 2013 to solve a problem of under-frequency threshold for disconnection.
This resulted in the saturation of the market for interface protection of medium voltage, leading to a suspension of incentives for plants that did not met the deadline.

Technology Roadmap Solar photovoltaic energy

In March 2013, the European Network of
Transmission System Operators for Electricity
(ENTSO-E) released its latest network code on
“requirements for generators”. When the new network code is formally approved and turned into laws, it will apply to all new generators and address the key issues of fault ride-through, frequency stability, voltage stability and remote control.
Obtaining permits and, more specifically, getting access to the grid has remained an obstacle for PV in many countries, because PV is not allowed at various voltage levels, or because grid operators have instituted complex, slow or expensive (or all of the above) connecting procedures. The replacement of feed-in tariffs (FiTs) by auctions has sometimes led to increases in overall costs, especially for low system sizes, as uncertain results increase development and financing costs.

Medium-term outlook
Based on a detailed analysis of all main PV markets, the IEA Medium-Term Renewable Energy Market
Report (IEA 2014c) conservatively estimates that cumulative installed PV capacity will likely exceed
400 GW worldwide by 2020. China, which recently adopted a target of 70 GW PV capacity by 2017, would lead the world, with over 110 GW. Japan and
Germany would each reach around 50 GW, followed by the United States at over 40 GW. Italy and India would rank fifth and sixth with 25 GW and 15 GW, followed by the United Kingdom, France and
Australia, all nearing 10 GW.
With respect to annual markets, by 2020 China would be leading with about 14 GW/y, followed by the United States (5 GW/y) and Japan (3 to 4 GW/y).
In 2020, global PV capacity that had been installed by the end of 2019 would be generating 530 TWh to 580 TWh, or about 2% of global electricity consumption. In the “enhanced case”, global installed capacity could reach 465 GW to 515 GW by
2020 (IEA, 2014c).

Progress since 2009

17

Vision for deployment
Since the IEA’s original PV roadmap was published in 2010, technology has improved and costs have fallen more than expected, partly because PV systems have been rolled out faster than expected.
Meanwhile, because of slower progress in carbon capture and storage (CCS) and persistent increased costs for nuclear power, ETP 2014 envisages lower deployment of those technologies between now and 2050 than ETP 2012 estimated (IEA, 2014b).
While the original roadmap set a goal for PV of
11% of total electricity generation by 2050, this roadmap, based on the hi-Ren Scenario of ETP 2014, aims for as much as 16%. PV generation would contribute 17% to all clean electricity, and 20% of

all renewable electricity. In both variants of the
2DS, global electricity production in 2050 is almost entirely based on zero-carbon emitting technologies mostly renewables (65% in the 2DS, 79% in the hiRen), in sharp contrast with the unsustainable 6DS and 4DS (Figure 5).
It is also worth noting that the greater expansion of
PV is not expected to harm the deployment of solar thermal electricity (STE) generated in concentrating solar power (CSP) plants; in this year’s update of the 2010 STE/CSP roadmap, estimates of STE’s share of total electricity generation have barely changed
(IEA, 2014a).

Figure 5: Global electricity mix in 2011 and in 2050 in three ETP 2014 scenarios
60 000
50 000

Variable 8%
Renewables 27%

TWh

40 000

29%
30 000
20 000

38%

65%
79%

2%

20%

10 000
0
Oil
Waste
Solar PV

2011
Coal
Biomass
Land-based wind

6DS
Coal with CCS
Biomass with CCS
Wind offshore

2DS
Natural gas
Hydro
Ocean

hi-Ren
Natural gas with CCS
Geothermal

Nuclear
STE

KEY POINT: in the hi-Ren Scenario, renewables provide 79% of global electricity by 2050, variable renewables provide 38%, and PV provides 16%.

CO2 reduction targets from the ETP 2014 hi-Ren Scenario

are reduced to a mere 1 GtCO2 in 2050. Solar PV would be responsible for avoiding 4 GtCO2/yr of emissions, or 19% of the total power sector emission reductions, by 2050, and 20% of cumulative emission reductions over the entire scenario period.

PV systems installed by the end of 2013 are generating 160 TWh/yr of clean electricity and thus avoiding about 140 million tonnes of CO2 per year
(MtCO2/yr). Annual emissions from the power sector would increase from 13 GtCO2 in 2011 to about
22 GtCO2 in 2050 in the ETP 2014 6DS (IEA, 2014b, see Box 3). By contrast, in the hi-Ren Scenario, they

18

Technology Roadmap Solar photovoltaic energy

Box 3: ETP Scenarios: 6DS, 2DS, hi-Ren
This roadmap takes as a starting point the vision in the IEA ETP 2014 analysis, which describes several scenarios for the global energy system in 2050.
The 6°C Scenario (6DS) is a base-case scenario in which current trends continue. It projects that energy demand would increase by more than two-thirds between 2011 and 2050.
Associated CO2 emissions would rise even more rapidly, pushing the global mean temperature up by 6°C.
The 2°C Scenario (2DS) sees energy systems radically transformed to achieve the goal of limiting the global mean temperature increase to 2°C. The high-renewables Scenario (hi-Ren
Scenario), achieves the target with a larger share of renewables, which requires faster and stronger deployment of PV, wind power and
STE, to compensate for the assumed slower progress in the development of CCS and deployment of nuclear than in 2DS.

Figure 6: Cumulative technology contributions to power sector emission reductions in ETP 2014 hi-Ren Scenario, relative to 6DS, up to 2050
Biomass 7%
CCS 5 %

Hydro 6%

Nuclear 3%
Fuel switching and efficiency
3%
Solar PV 20%

Electricity efficiency improvements
23%

STE 9%

Other renewables 2%
Wind offshore 6%

Land-based wind 14%

The ETP 2014 analysis is based on a bottom-up
TIMES* model that uses cost optimisation to identify least-cost mixes of energy technologies and fuels to meet energy demand, given constraints such as the availability of natural resources. Covering 28 world regions, the model permits the analysis of fuel and technology choices throughout the energy system, representing about 1 000 individual technologies. It has been developed over several years and used in many analyses of the global energy sector. The ETP model is supplemented with detailed demand-side models for all major end-uses in the industry, buildings and transport sectors.
* TIMES = The Integrated MARKAL (Market Allocation)EFOM (energy flow optimisation model) System.

Concerns have been raised about CO2 emitted in the manufacturing of PV modules (see the Solar
PV Technology Development section) and the possibility that the variable nature of PV power may hinder CO2 emission reductions at power system level (see the System integration section). However, modelling by the IEA and others shows that the penalties incurred due to the manufacturing process and the variability of PV are minor compared with the emission reductions arising from fossil fuel displacement.
The regional distribution of additional CO2 emission reductions due to PV in the hi-Ren Scenario
(Figure 7) primarily reflects the share of PV in the electricity mix of each region (see below). However, it also depends on the carbon intensity of that mix in the 6DS. China, for example, has a carbon-intensive power mix today and in the 6DS. This explains why
China alone accounts for half the additional emission reductions in 2050 due to the large PV deployment in the hi-Ren Scenario, while accounting for “only”
35% in the total PV generation.

KEY POINT: Solar PV would equal wind power in cutting CO2 emissions over the next 35 years.

Vision for deployment

19

Figure 7:  dditional CO2 emission reductions due to PV in 2050
A
in the hi-Ren Scenario (over the 6DS)
Latin America
Eastern Europe and
2% OECD Europe 2%
Former Soviet Union 1%
OECD Asia Oceania 3%

United
States
9%

Africa 2%

Other developing Asia 13%

China 50%

Middle East 3%
India 15%

KEY POINT: China alone would account by 2050 for half the global emission cuts due to PV deployment.

Revised solar PV goals
To achieve the goals set out in the hi-Ren Scenario, this roadmap considerably increases the PV capacity deployment that was envisioned in the
2010 roadmap. The hi-Ren Scenario now sees a deployment of 1 700 GW of PV by 2030 (up from

870 GW in the 2010 roadmap), and of 4 670 GW by
2050 (up from 3 155 GW in the 2010 roadmap). This represents capacity additions of over 120 GW/yr on average, with a 15-year plateau above 200 GW/‌ r y between 2025 and 2040 (Table 3). Including repowering, annual installed capacities would be
185 GW on average.

Table 3: PV capacities by region in 2030 and 2050 in the hi-Ren Scenario (GW)
Eastern
Europe and Other former NonOther
Middle developing Soviet
OECD
OECD China India Africa East
Asia
Union Americas World

Year

US

Other
OECD
Americas

2013

12.5

1.3

78

18

18

2.3

0.3

0.1

1.4

3

0.2

135

2030

246

29

192

157

634

142

85

94

93

12

38

1721

2050

599

62

229

292

1738

575

169

268

526

67

149

4674

EU

Notes. Some numbers in this table, especially for 2030, differ from those provided in ETP2014. The ETP model was re-run after the publication of ETP 2014 with slightly updated assumptions.

20

Technology Roadmap Solar photovoltaic energy

As for electricity generation (Figure 8), the hiRen Scenario foresees 2 370 TWh by 2030 and
6 300 TWh by 2050 (a 39% increase over the 2010

roadmap), so that PV achieves a 16% share in the global electricity mix (up from 11% in the 2010 roadmap or the ETP 2014 2DS).

Figure 8: Regional production of PV electricity envisioned in this roadmap
18%
16%

6 000

14%
5 000

12%

4 000

10%

3 000

8%
6%

2 000

4%
1 000

2%

0

Share of total electricity generation

PV electricity generation (TWh)

7 000

0%
2015

United States
Middle East

2020

2025

Other OECD Americas
Other developing Asia

2030

2035

European Union
Eastern Europe and FSU

2040
Other OECD

2045

2050

China

Non-OECD Americas

India
2DS

Africa

Share of total

KEY POINT: in the hi-Ren scenario, PV provides 16% of global electricity by 2050, and China has a 35% share of the total PV electricity production.

China is expected to overtake Europe as the largest producer of PV electricity soon after 2020, with its share regularly increasing from 18% of global generation by 2015 to 40% by 2030 then slowly declining to 35% by 2050. From 2030 to 2050, the share of India and other Asian countries is expected to rise from 13% to 25%. By contrast, the United
States’ share is expected to remain at about 15% from 2020 on, and Europe’s share to decrease constantly from 44% in 2015 to 4% in 2045.
This reflects widely different situations with respect to the power mix, and more specifically differences in the mix of renewables by 2050 (see Figure 9), based on the variety of resources available in different parts of the world, but also on different electricity load profiles In non-OECD Americas, for example, the large availability of hydro power eases the integration of variable PV but combines with very competitive land-based wind power to limit PV penetration. In Europe, the solar resource is high in the south but significantly lower in the north, while electricity demand is on average greater in winter than in

summer (IEA, 2011). Demand peaks often occur in late afternoon or early evening, so the “capacity credit”4 of PV at winter peak times is close to zero in most countries. Wind power in Europe offers a better match with daily and seasonal variations in demand, at competitive costs, and thus limits the penetration of PV to about 8% by 2050 in ETP 2014 hi-Ren Scenario.
PV power could be more widely deployed in Europe if the costs of decentralised electricity storage
(beyond transportation uses) fell significantly
(IEA, 2014e). UBS estimates, for example, that if battery costs fell in line with the most optimistic assumptions, 14% to 18% of electricity demand in Germany, Spain and Italy could be met by self-produced solar electricity – with 6% to 9% of electricity demand replaced on these markets by
2020 (Hummel and Lekander, 2013)

4.  he capacity credit of variable renewables is the reduction of
T
peak capacity required to satisfy the power demand with the same loss-of-load probability, as a percentage of the variable renewable capacity installed.

Vision for deployment

21

Figure 9:  eneration mix by 2050 in the hi-Ren Scenario
G
by region (in annual energy)
Non-OECD Americas

10%

Eastern Europe and FSU 3%
19%

Other developing Asia

18%

Middle East

11%

Africa

22%

India

21%

China
13%

Other OECD

8%

European Union
Other OECD Americas

6%

United States

18%

0%
Solar PV

STE

10%

Wind

Hydro

20%
30%
40%
50%
60%
Biomass and waste
Other renewables
Oil

70%
Coal

80%
90%
100%
Nuclear
Natural gas

KEY POINT: PV shares vary with the solar resource and electricity load.

Where the solar resource is high and electricity demand is largely driven during many months of the year by air-conditioning from noon (or even before) to evening, the match between resource and demand is much better. The economics of
PV are improved if an actual capacity value is acknowledged and rewarded on electricity markets.
When PV has a large share of electricity generation, however, the capacity credit of additional PV generation will diminish in the absence of demand side response or storage options (IEA, 2014d). PV systems can be deployed close to consumption centres, and even directly on consumption sites, which allows for lower grid losses and lower grid investments in some cases.
In countries or regions with strong sunshine and clear skies, CSP plants with built-in thermal storage capabilities may be better placed than PV with storage to capture a large share of electricity demand when the sun is not shining (IEA, 2014a). This explains why PV does not fare better in Africa, India, the Middle East or the United States than it does in
China or other Asian developing countries, in the
ETP 2014 hi-Ren Scenario. Rooftop PV represents half of PV capacities in this roadmap. An indicative-only possible repartition among main market segments could be about 2% off-grid systems and 98% gridconnected systems, of which 20% residential and
30% commercial rooftop systems, 10% industrial and
40% utility ground-based systems.

22

Potential for cost reductions

As local markets develop, system costs are likely to converge towards the current lowest values, except in places where “soft” costs, such as the cost of obtaining permits, are higher. The cost range will thus narrow significantly. Costs will fall further as technology improves, for both utilityscale and rooftop PV systems. Utility-scale capital expenditures cost would fall below USD 1/W by
2030 on average, but the cheapest systems would reach that mark by about 2020. Average costs would then reach a level of USD 700/kW by 2050.
Rooftop prices would hit USD 1/W by 2025 for the cheapest systems and by 2040 on average
(Figure 11).

The prices of cells and modules fell rapidly from
USD 4/W in 2008 to USD 0.8/W in 2012, but have since stabilised. Prices in 2008 were higher than expected, given the long learning trend, because of a shortage in c-Si capacities. The lowest market prices in 2012 and 2013 may have been below full costs, including return on investment. However, there is considerable body of evidence that the costs of cells and modules, whether of c-Si or TF, will decline further as deployment increases and technology improves in the next two decades. This roadmap expects module costs to fall to USD 0.3/W to USD 0.4/W by 2035 (Figure 10).

Technology Roadmap Solar photovoltaic energy

Figure 10: Past modules prices and projection to 2035 based on learning curve
100

1976

pv module price (USD/W)

1980

1990

10

2000

2015

1

2020
2013

2030
2025
2035

0.1
0.00

0.00

0.01

0.10
1.00
10.00
Cumulative manufactured capacity (GW)
Experience curve
Observations

100.00

1 000.00

10 000.00

Targets

Notes: Orange dots indicate past module prices; purple dots are expectations. The oval dots correspond to the deployment starting in
2025, comparing the 2DS (left end of oval) and 2DS hi-Ren (right end).

KEY POINT: This roadmap expects the cost of modules to halve in the next 20 years.

Figure 11: PV investments cost projections in the hi-Ren Scenario
USD/kW
5 000

Utility-scale PV system

Rooftop PV system

4 000

3 000

2 000

1 000

2015

2020

2030

2040

2050
Range

0

2015

2020

2030

2040

2050

Weigthed average

KEY POINT: The price ranges of PV systems will narrow, and the average cost will be halved by 2040 or before.

Vision for deployment

23

As capital expenditures fall, performance ratios increase, and the bulk of PV moves from Europe to sunnier skies, the average PV LCOE will continue to diminish – and the range of LCOE across countries will continue to narrow. The average LCOE of

new-built, large-scale, ground-based PV plants is expected to fall on average below USD 100/MWh by 2025, and to gradually reach USD 60/MWh
(Table 4).

Table 4:  rojections for LCOE for new-built utility-scale PV plants to 2050
P
(USD/MWh) in the hi-Ren Scenario
USD/MWh

2013

2020

2025

2030

2035

2040

2045

2050

Minimum

119

96

71

56

48

45

42

40

Average

177

133

96

81

72

68

59

56

Maximum

318

250

180

139

119

109

104

97

Note: All LCOE calculations in this table rest on 8% real discount rates as in ETP 2014 (IEA, 2014b). Actual LCOE might be lower with lower WACC.

Table 5:  rojections for LCOE for new-built rooftop PV systems to 2050
P
(USD/MWh) in the hi-Ren Scenario
USD/MWh

2013

2020

2025

2030

2035

2040

2045

2050

Minimum

135

108

80

63

55

51

48

45

Average

201

157

121

102

96

91

82

78

Maximum

539

422

301

231

197

180

171

159

Note: All LCOE calculations in this table rest on 8% real discount rates as in ETP 2014 (IEA, 2014b).

The LCOE of new-built rooftop PV systems will fall on average below USD 100/MWh soon after 2030, and gradually reach USD 80/MWh (Table 5).
For a given site and irradiation, setting aside performance ratio and its evolution over time, the most important levers for cost reductions are capital expenditures and costs of capital (Figure 12). When the weighted average capital cost (WACC) exceeds
9%, more than half the LCOE represents the burden of financing.
The LCOE projections in this roadmap rest on a
WACC of 8%. More optimistic assumptions lead to lower costs. For example, a WACC of 2.4% to

24

2.8% in Germany and 4.7% in high solar irradiation countries could reduce the LCOE of PV power plants to EUR 55 to 94/MWh (USD 74 to 127/MWh) by
2020, so that “even small rooftop PV systems will be able to compete with onshore wind power and the increased LCOE from brown coal hard coal and combined cycle gas turbine (CCGT) power plants”
(Kost et al., 2013). Where solar irradiation is high
(2 000 kWh/m2/.m ‑2.y‑1), the same study computes the LCOE of utility-scale PV at EUR 59/‌ Wh
M
(USD 80/MWh) today, and EUR 43 to 64/MWh
(USD 58 to 87/MWh) by 2030.

Technology Roadmap Solar photovoltaic energy

Figure 12: The share of the costs of capital in the LCOE of PV systems

70%
LCOE USD/kWh

0.20

60%
50%

0.15

40%
0.10

30%
20%

0.05

10%
0.00

Share of cost of capital in LCOE

80%

0.25

0%
0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10% 11% 12% 13% 14% 15% 16% 17% 18% 19% 20%
WACC
Investment

O&M

Cost of capital

Percentage cost of capital

Notes: This example is based on output of 1 360 kWh/kW/y, investment costs of USD 1 500/W, annual operations and maintenance
(O&M) of 1% of investment, project lifetime of 20 years, and residual value of 0.

KEY POINT: When the WACC exceeds 9%, over half the LCOE of PV is made of financial expenditures.

Box 4: Sustainable PV energy for all
Nearly 1.3 billion people did not have access to electricity in 2011, mostly in Africa and developing Asia. By 2050, although population growth will concentrate in cities, hundreds of millions of people will still live in sparsely
-populated rural areas where off-grid solar systems would likely be the most suitable solution for minimum electrification. The IEA
Energy for All case of the 2012 World Energy
Outlook (IEA, 2012a) assumes grid extension for all urban zones and around 30% of rural areas, and for the remainder, mini-grids and standalone solutions.

improve life considerably for those who earn
USD 1 to USD 2 per day and spend as much as USD 0.4 per day on dry batteries, kerosene and other energy products (IEA, 2011). PV could prove competitive if financing costs can be reduced, given the high share of up-front investment costs. This roadmap thus assumes by 2030 a PV capacity of 200 W per capita for
500 million people lacking other access to electricity. Mini-grids and off-grid PV capacity would thus total 100 GW, representing about
5% of total capacity by 2030 (2% by 2050), a significant increase from current trends.

In both on-grid and off-grid situations, solar
PV has considerable merits, sometimes in combination with other energy sources. It can

Global investment to 2050
To decarbonise the entire energy system in the
2DS by 2050 will require about USD 44 trillion of additional spending. This investment is more

than offset by over USD 115 trillion in fuel savings, resulting in net savings of USD 71 trillion. Even with a 10% discount rate, the net savings are more than
USD 5 trillion (IEA, 2014b).

Vision for deployment

25

The 2DS hi-Ren requires cumulative investments for power generation of USD 4.5 trillion more than in the 2DS, including notably PV but also wind power and STE. The lower consumption of fossil fuels in this variant saves USD 2.6 trillion, however, partly offsetting the additional investment needs, so that overall the 2DS hi-Ren variant results in additional costs of USD 1.9 trillion. This represents

a 3% increase in total cumulative costs for power generation compared with the 2DS, and only a 1% increase over the 6DS.
However, investments are more significant in the next two decades of the hi-Ren Scenario. This is reflected in the implicit carbon prices in both variants, which differ significantly by 2030 (Table 5).

Table 6: CO2 prices in the climate-friendly scenarios of ETP 2014
USD/CO2

2020

2030

2040

2050

2°C Scenario

46

90

142

160

Hi-Ren Scenario

46

115

152

160

Total investments in PV over the modelling period of the hi-Ren Scenario would be about USD 7.8 trillion
(undiscounted) for a total of 6 600 GW manufactured capacity, including repowering.

Beyond 2050
Limits to PV deployment mentioned above – such as 8% in Europe – depend on the modelling assumptions built into ETP modelling and (in part) on its time horizon of 2050, and may not be hard limits in a world that will strive to further reduce
GHG emissions while ensuring energy security and access at affordable costs. The ultimate objective of the United Framework Convention on Climate
Change (UNFCCC) is to stabilise GHG atmospheric concentrations at a level that would prevent dangerous anthropogenic interference with the

26

climate system. Whatever this exact level turns out to be, CO2 stabilisation will require net emissions of zero or below to compensate for the rebound effect, i.e. the release in the atmosphere of CO2 from natural reservoirs that accumulate some of current anthropogenic emissions (IPCC, 2014, chapters 6 and 12).
Longer-term climate change mitigation studies tend to show significantly higher PV deployment beyond 2050 – or even by 2050, due to their longer- term perspective. Germany alone could install 105 GW of PV without additional storage, to cover 19% of its electricity consumption (Giesler et al., 2013). A longer-term post-2060 hypothetical scenario envisages reducing global energy-related
CO2 emissions to about a tenth of current levels by installing 2.6 times as much PV capacity as this roadmap assumes by 2050 (IEA, 2011).

Technology Roadmap Solar photovoltaic energy

Solar PV technology development:
Actions and milestones
The era of rapid price decreases for PV cells and modules is probably over. All types of PV modules still have significant room for improvement, however, starting with c-Si modules, which dominate the market with a share of 90%. TF manufacturers plan to increase efficiencies and durability. Low- and high-concentrating PV providers strive to reduce costs and compete with
“1-sun” PV (i.e. PV without concentration) in highirradiance areas.

cell efficiencies5 achieved on very small surface areas do not immediately translate into affordable commercial high-efficiency modules. Nevertheless, the PV industry has constantly demonstrated it can reduce costs while increasing the efficiency of commercial modules (Figure 13). Furthermore, while greater deployment has driven most cost reductions over the past decade, technology improvements are likely to return as a major factor behind future cost reductions (Zheng and Kammen, 2014).

The lowest PV costs are not necessarily achieved with the highest efficiencies, and small improvements in efficiency can come at too high a cost to be worthwhile, even accounting for lower
BOS costs driven by higher efficiencies. Record

5. See for example the well-known graph of best research-cell

efficiencies that the NREL regularly updates and makes available at www.nrel.gov/ncpv/images/efficiency_chart.jpg.

Figure 13: Commercial 1-sun module efficiencies (actual and expected)

Average module efficiency (%)

25.0

20.0

15.0

10.0

5.0

0.0
1997

1999

2001

2003

2005

SPW X-series
SPW E-series
SPW 1st
HIT S/P ms-Si (Photon Int)
CdTe, FirstSolar
CiGS, Solar Frontier

2007

2009

2011

sc-Si Yingli mc-Si Yingli a-Si/µc-Si, Oerlikon Solar Fabs

2013

2015

sc-Si (Photon Int) a-Si, Tsolar, ES

Note: SPW stands for SunPower, HIT S/P stands for Heterojunction Intrisic Thin layer Sanyo/Panasonic.
Source: De Wild-Scholten, M. (2013), “Energy payback time and carbon footprint of commercial PV systems”, Solar Energy Materials &
Solar Cells, No. 119, pp. 296-305.

KEY POINT: PV efficiencies have been rising constantly, and still have room for improvement.

Solar PV technology development: actions and milestones

27

Cells and modules
This roadmap recommends the following actions

Time frames

1. ncrease module efficiencies to 40% (HCPV), 24% (sc-Si), 19%(mc-Si; CdTe; CIGS) or
I
12% (a-Si/µc-Si; organic; dye-sensitised).

Complete by 2017.

2. Increase performance ratios and decrease degradation rates.

Complete by 2020.

3. Diversify module specifications for variable environments.

Complete by 2020.

4.  educe Si consumption to 3 g/W while increasing module longevity. Reduce silver
R
consumption.

Complete by 2020.

5. Enlarge wafer size. Generalise reusable moulds.

Complete by 2020.

6.  evelop low-cost high-efficiency high-output bifacial 1-sun tandem cells, and
D
design specific systems around them.

Complete by 2020.

7.  evelop specific PV materials for building integration, road integration and other
D
specific supports.

Complete by 2025.

8. urther reduce Si consumption below 2 g/W and increase efficiencies to 50%
F
(HCPV), 28% (tandem cells), 22% (mc-Si, CdTe, CIGS), 16% (a-Si/µc-Si; organic; dye-sensitised cells).

Complete by 2025.

c-Si modules
The efficiency of the best commercial c-Si modules now exceeds 21% and manufacturers such as USbased SunPower are targeting 23% efficiency by
2015, together with significant cost reductions.
The major cost in c-Si PV cells is for pure polysilicon feedstock, which dropped from USD 67 per kilogram (kg) in 2010 to USD 20/kg in 2012 and has remained below this price since then.
Continued progress in the production processes, and reduction in the use of consumables, will keep the price under USD 20/kg in the next few years.
Cost of ingot growth, wafer (cell precursors) sawing and cleaning will also improve. Efforts to reduce the amount of purified silicon in cells – which is now as low as 5 grams (g) per watt for the best cells – will continue towards 3 g/W or less, with thinner wafers.
Diamond wire sawing and improved slurry-based sawing will reduce losses in slicing c-Si wafers, while kerf-less technologies may or may not offer an alternative to the traditional wafer-based c-Si process. Manufacturers are also striving to use less silver and other expensive materials (maybe replacing silver with copper), while maintaining or even extending the technical life of cells and modules.
Manufacturing automation is progressing for both

28

cells and modules. For modules, higher throughput could be achieved for the interconnection and encapsulation processes. Energy efficiency improvements over the whole manufacturing process are being sought. “Mono-like” mc-Si ingots, and reusable ingot moulds, could bring sc-Si performances at mc-Si costs. Back contact and metal wrap-through technologies, which reduce shading and electric losses, have been successfully introduced to markets by various manufacturers.
The historical learning rate of 20% could be maintained over the next few years by introducing new double- and single-sided contact cell concepts with improved Si-wafers, as well as improved cell front and rear sides and better module technologies
(IRTPV, 2014).
The heterojunction (HTJ) cell design combines two materials – often c-Si wafer and a-Si TF – into one single junction, resulting in higher efficiencies and performance ratios, thanks to a better resistance to high temperatures. The leader in HTJ technology,
Sanyo/Panasonic, now develops HTJ cells with back contacts, and has announced a record efficiency of
25.6% in April 2014 on a research cell of “practical size” (over 100 cm2). Bifacial solar cells offer another emerging option, to be used in glass/glass modules enabling an increase in performance ratio and energy output of up to 15% using the light

Technology Roadmap Solar photovoltaic energy

reflected by the ground or buildings through the rear face. HJT technology, possibly combined with back-contact designs, may be further improved by using alternatives to amorphous silicon, in order to increase the overall spectral response (Sinke, 2014).

Costs could also be reduced by producing a greater variety of modules, adapted to a wider variety of conditions, including snow, hail, salt, humidity and heat. In the United States, for example risks vary greatly across the country (Figure 14).

Figure 14:  ow different climate zones in the United States
H
affect the lifetime of PV modules
Snow load

Mod A
Dry A

Hail

Mod A, Damp C
Damp B

.

Marine (salt spray test)

Damp A

Snow load > 4 800 Pa
Hail testing > 4 cm

This map is without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area.

Notes: Estimated minimum ratings needed for: 25 years estimated service life, open-rack mounting, retention of 80% power and pass high pot testing for 90% of modules. Climate ratings are indicated for Mod (Moderate), Damp (Warm damp, equable), Dry (Extremely warm dry).
Source: Kurtz, S. (2013), Photovoltaic Module Reliability Workshop 2013, NREL.

KEY POINT: Tailoring the resistance of PV modules and systems to their environment could reduce their cost without reducing their longevity.

Thin films
Three technologies dominate the TF area. The leading company in CdTe technology, First Solar, recently revised upward its efficiency objectives, targeting 25% for research cells and over 19% for commercial modules in three years. The USbased firm also claims that its latest generation of technology reduces degradation of performance to
0.5%/year in all climates.
Copper-indium-gallium-selenide (CIGS) technology, with efficiency of 12% to 14%, lags behind c-Si but offers a slightly higher performance ratio. The largest supplier, the Japanese firm Solar Frontier, which exceeded 20% efficiency in relatively large

research cells produced with mass production technology, aims to increase the efficiency of its commercial modules on this basis. It also aims to halve its costs from end-of-2012 levels by 2017. a-Si technology offers traditionally the lowest efficiency among commercial modules, and its deployment has long been impeded by concerns about the longevity of its modules and degradation rates. When these issues were solved, the cost gap with c-Si was no longer sufficient to warrant strong deployment. The combination of amorphous and micro-crystalline silicon (a-Si/µc-Si) allows higher efficiencies. Solar PV technology development: actions and milestones

29

Multi-junction cells
Multi-junction cell design involves superposing several cells in a stack. In the case of two cells, it will form a double junction, also called a tandem cell. Stacking more cells together forms a triple or a quadruple junction. In all cases, the upper cell(s) must be as transparent as possible to enable the lower cells to still be active. This approach enables a broader spectrum of sunlight to be captured, and overall efficiency to be increased. Record researchcell three-junction efficiencies of 38.8% under 1-sun exposure, and 44.4% under high concentration
(>300 suns), have been achieved by Spectrolab, a
Boeing subsidiary, and Sharp, respectively. Fouror five-junction cells could reach even higher efficiencies, such as the 44.7% achieved by Soitec and Fraunhofer ISE. Multi-junction cells have been so far mainly used for space applications and highconcentration solar cells (see below), reducing the semiconductor area to a small fraction and therefore allowing the use of more expensive materials. Niche markets exist however, for 1-sun multi-junction cells, such as unmanned aerial vehicles.
The rapid cost decline of c-Si, however, opens the door for mass-production of high-efficiency tandem cells, where TF would be deposited on c-Si wafers. Reviewing a broad range of material options
– “III-V” alloys, chalcogenides and perovskites
– and relevant production processes, Green et al. (2013) suggest that Si wafer-based tandem cells could represent a very cost-effective longterm combination. Such cells would probably be best used under 1-sun or low concentration with simplified tracking devices.

CPV
CPV technologies include low-concentrating PV
(LCPV), which tracks the sun on one axis with a concentration ratio of around 10, and highconcentrating PV (HCPV), which tracks the sun on two axes with a concentration ratio in the hundreds. LCPV can be based on best-in-class c-Si cells. HCPV uses very high concentration factors, at or above 300, which allows the use of more expensive but highly efficient multi-junction cells.
HCPV requires more precise tracking devices than
LCPV or 1-sun devices. At locations with high DNI, both LCPV and HCPV already today compete with
PV. Unlike dispatchable STE, CPV output varies like that of other sun-tracking PV systems while the higher efficiency significantly reduces the installed module surface per MW.

30

Advanced solar cells under development
Novel PV devices, such as quantum dots, dyesensitised cells, organic cells and thermoelectric devices hold great promise for the future, but for mainstream applications they need to reach specific performance and cost levels to enter the market. That is why “stepping stones” in the form of markets that require specific properties (such as low weight, transparency, flexibility, colour and freedom of form) may help these new options to enter the market successfully. Efficiencies of 11% for organic cells and 12% for dye-sensitised cells have recently been achieved by Mitsubishi Chemical and Sharp, respectively. Non-module costs
Non-module costs relate to non-module hardware, including fixed supports or tracking systems, cables, inverters and soft costs, including customer acquisition, permitting, installation, connection, and financing. The latter are investigated in the
Policy and Finance section of this roadmap.
Inverters have followed an impressive learning curve, similar to that of PV modules. The reduction in material has been dramatic in the last ten years, from 12 kg/W to 2 kg/W. Manufacturers expect this trend to continue. Other hardware costs – materials, such as support and cables, or labour, such as installation – relate to the area of the solar PV systems and thus depend mostly on the efficiency of the modules. Increased efficiencies thus drive system cost reductions, which become progressively more important as the cost of PV modules diminishes and other costs rise above half the total system costs.
Systems designed to be integrated into the envelopes of buildings, or building-integrated PV
(BIPV), currently cost more than standard rooftop systems. The BIPV concept raises the possibility, however, that a thin layer of PV-active material, possibly deposited as a paint, could become a standard feature of building elements such as roof tiles, façade materials, glasses and windows, just as double-glazed windows have become standard in most countries. Given that such elements comprise a large part of building envelopes, mass production could enable the cost of PV to almost vanish in this market segment where it currently costs the most.

Technology Roadmap Solar photovoltaic energy

Thin films and advanced solar cells are the primary candidates for such applications. Other integrated applications, such as PV materials for roads and similar surfaces, are also under development.

Life-cycle analysis
Manufacturing PV cells, modules and installing systems consumes energy and results in greenhouse gas emissions. Questions have been raised about the life-cycle assessment of PV systems with respect to climate change. Often-quoted estimates of energy pay-back times of two to five years, and
GHG emissions of about 50 gCO2-eq/kWh for mc-Si modules and 75 gCO2-eq/kWh for sc-Si modules
(e.g. IPCC, 2011, p.372) are already outdated and on steep downward trends. Technical improvements rapidly decrease energy consumption in the PV manufacturing process, while efficiency and performance ratios of new PV systems continuously increase. Recent studies show energy payback times of commercial PV systems under Southern
Europe sunshine of 0.7 to 2.5 years, depending on technology and the power mix in manufacturing countries; the carbon footprint of PV electricity ranges from 20 to 81 gCO2-eq/kWh, one order of magnitude below electricity from fossil fuels (de
Wild-Scholten, 2013).

Some companies, such as FirstSolar, have longestablished recycling schemes with facilities operational at all manufacturing plants and recovery rates of up to 95% of the semiconductor material and 90% of the glass. PV producers set up a manufacturer-independent recycling system in June 2010 (PV Cycle), which currently has more than 300 members. The version of the European
WEEE Directive (Waste Electrical and Electronic
Equipment Directive), which came into force on
August 13, 2012, must be implemented in all EU states by the end of February 2014. This directive makes it compulsory for manufacturers to take back and recycle at least 85% of their PV modules free of charge. Task 12 of the Implementing Agreement for a
Co-operative Programme on Photovoltaic Power
Systems of the IEA (PVPS IA) includes ongoing work on recycling of manufacturing waste and spent modules, as well as further work on life-cycle inventories and assessment.

During the production of thin-film PV and flat screens, nitrogen trifluoride (NF3) is still used by some manufacturers to clean the coating systems.
Residues of this gas can escape into the atmosphere.
NF3 is more than 17 000 times as harmful to the environment as carbon dioxide. Current emission quantities are not known. As of 2013, NF3 emissions are to be determined in 37 countries according to the revised Kyoto Protocol.

Solar PV technology development: actions and milestones

31

System integration: Actions and milestones
PV generation is variable and uncertain. This does not create serious issues when PV has a low share of the power mix, but system-friendly deployment

is necessary to allow PV’s share to progressively increase, as well as other measures that transform broader power systems.

Facilitate PV integration into electric grids

Time frames

1.  evelop meteorological PV forecast, with feedback loop from PV power plant
D
online data to weather forecasting.

2015-20

2.  laborate and enforce grid codes that will drive inverters to provide voltage
E
control and frequency regulation.

2015-30

3.  revent PV hot spot emergence in ensuring geographical spread, e.g. through
P
spatial remuneration differentiation.

2015-40.

4.  acilitate rapid market reactions by shortening gate closure times and trading
F
block length.

2015-20

5. ncentivise generation during demand peaks through time-of-delivery payments
I
and/or limitation to instantaneous injection except at peak times.

2015-30 depending on countries

6. ncentivise load management and flexibility from existing generating capacities;
I
ensure fair remuneration of ancillary services.

2015-30 depending on countries

7. nvestigate options for new PHS plants; anticipate the need for more flexible
I
power capacities.

2020-30

8.  evelop new storage capabilities.
D

2030-50

Variability and uncertainty
The output of solar PV depends on daylight patterns and the weather, notably the cloud cover and atmospheric turbidity. Clouds are only partially predictable over small areas, but the uncertainty regarding aggregate cloud coverage – which must be distinguished from variability – is reduced at larger geographic scales. Indeed, the generation of solar power is now easier to plan thanks to increasingly reliable forecasts. Changes in cloud cover are usually not able to create unpredicted sudden changes in generation at some level of aggregation (Figure 15) – this explains why the resource should be termed “variable” rather than
“intermittent”. Unexpected episodes of fog can cause significant forecast errors, however.
Variability has two distinct effects on electric systems. Balancing effects, which relate to rapid short-term changes in load net of PV generation, from minutes up to a timescale of one or two days, must be addressed to avoid outages. Utilisation effects relate to how often a certain net load level

32

– defined as gross load minus variable generation from wind and PV – occurs over the course of a longer period of time. This relates to adequacy – the long-term transformation of the entire electrical system needed to keep pace with demand costeffectively, i.e. re-arranging the shares of the different energy sources and electric-generating capacities to match likely utilisation rates, the various technologies and sources being best used for peak, mid-peak or base-load generation.

System-friendly
PV deployment
Integrating low shares of PV power (of just a few percent) usually does not raise significant challenges, provided some pitfalls are avoided. For example, it is important to avoid concentrating PV capacities in areas with low power demand and relatively weak distribution grids, where variability may cause voltage problems, create reverse power flows, and lead to large grid congestions.

Technology Roadmap Solar photovoltaic energy

Changes in PV system design can better match supply and demand. PV developers can opt for sun-tracking systems. Developers can adjust the tilt and orientation of panels to maximise output at certain times of day or year, instead of maximising the annual output, if they receive appropriate timeof-delivery (TOD) price signals. Another option, which recent module cost cuts have made possible, is to design fixed-tilted PV systems with panels at different orientations and a greater DC/AC ratio —
i.e. increased total capacity of modules (generating
DC current) with respect to the capacity of the inverters (delivering AC current to the grid).
Modules would not all face the equator; some could be oriented southeast and others southwest, thus delivering a more regular output throughout the day and increased capacity factors for the inverter.6
Multiple orientation systems, while not necessarily optimal, also offer additional opportunities for deployment on buildings.
System-friendly PV deployment should also be part of a broader system-friendly deployment of renewables, especially variable renewables. The balance between wind power and PV must be considered carefully, to take advantage of these complementary resources over time, as illustrated by the German case (Figure 16) – and possibly also the balance between PV and STE, if the latter is available at reasonable distance.

6. W ind power follows a similar path with higher hubs and greater

swept area/capacity ratios (IEA, 2013b).

G

-5

G

W

25 000

+5

+1
0

G

W

30 000

W

Figure 15: Hourly planned versus actual solar generation in Germany, 2013

G

W

20 000

-1
0

Planned production solar (MW)

Modern electronics allow PV systems, via their inverters, to perform a number of tasks autonomously, however, such as riding through wide ranges of voltage and frequency fluctuations, actively counteracting voltage changes (volt‐ var control) in providing reactive power, and reconnecting softly to avoid sharp spikes when disconnecting during power outages (SIWG, 2014).
Telecommunication skills of modern inverters would greatly expand the possibilities and enable decentralised PV systems to support the grid. The system services capabilities of PV systems (and wind turbines), and the costs associated with providing these services, are the focus of the EU-funded
REserviceS project (Kreutzkamp et al., 2013).

15 000
10 000
5 000

0

5 000 10 000 15 000 20 000 25 000 30 000
Real production solar (MW)

Source: Burger, B. (2014), Electricity Production from Solar and
Wind in Germany in 2013, Fraunhofer Institute for Solar Energy
Systems, January, Freiburg, Germany.

KEY POINT: Over country-sized areas, solar energy generation is largely foreseeable.

Integrating large PV shares
When PV makes up a large share of electricity generation, systems may require more reserves to ensure balancing than would otherwise be developed on the basis of the unpredictability and variability of electricity demand, and the risks of failures of some generating plants or connecting lines. The possibility of long periods with little solar resource — more frequent in winter — calls for adequate firm capacities. Eventually, integrating large shares of PV electricity requires technical and economic flexibility from the rest of the system (IEA,
2014). This need for flexibility can be illustrated by the foreseen evolution of the net load curve of spring days in California, nicknamed the “duck chart” (Figure 17), which reveals how PV is expected to modify the curve during daytime but keep almost unchanged the demand peak of the early evening
– unless other measures, such as demand-side management, are taken (see e.g., Lazar, 2014).
Load management, including electricity efficiency improvements and load shifting, offers affordable options for integrating variable PV output. This strategy has great potential, but is not infinite — people will always need light at night. Electricity

System integration: actions and milestones

33

Figure 16: Weekly production of solar and wind in Germany in 2013
2.5

TWh

2.0

1.5

1.0

0.5

0.0
1

4

8

16

12

20

24

28
Week

32

36

40

44

48

52

Solar

Wind

Source: Burger, B. (2013), Electricity Production from Solar and Wind in Germany in 2013, Fraunhofer Institute for Solar Energy Systems, January,
Freiburg, Germany. Data: EEX Transparency Platform.

KEY POINT: Wind energy is often strong when sunshine is weak and vice versa.

Figure 17: Expected evolution of the net load of a typical spring day in California
28 000
26 000
24 000
2012 actual

22 000

MW

20 000

2013 actual

18 000

2014
2015

16 000
14 000

2019
2018

12 000

Ramp need
~13 000 MW in three hours

2016
2017
2020

Over generation risk

10 000
0
0

3

6

9

12
Hour

15

18

21

Source: California ISO (Independent System Operator) (2013), What the Duck Chart Tells Us About Managing a Green Grid, Fast Facts,
Folsom, CA, accessed 4 June 2014.

KEY POINT: The “duck chart” illustrates how large PV generation requires flexibility from the rest of the power system.

34

Technology Roadmap Solar photovoltaic energy

savings, especially targeting nocturnal peak consumption, would help integrate more PV in the mix. Load management would not only reduce the annual electricity demand that PV cannot supply, but also reduce the minimum load level, during daytime, of the conventional plants required to cover the peak at night, in particular if they have long starting times for hot starts. Savings on lighting is the obvious example.
Interconnections are important, because they allow smoothing out to some extent of the variability of PV plants over large areas, and enable the sharing of flexible generation, demand-side management and storage. Integration with other energy forms and energy networks, such as district heating or gas networks through hydrolysis and methanation, could also help increase PV shares in the electricity mix.
The flexibility of electricity-generating plants other than PV and wind has two aspects that are interlinked but distinct: one is purely technical, the other economical. Conventional thermal plants often take time to start or stop; not all can change pace quickly, and many have minimum loads in the 40%-50% range (in particular base-load plants, including older designs of CCGT power plants), but depending on plant design, minimum generation levels can be 25% or lower, even for coal-fired plants. Cold starts take a long time, especially for nuclear and coal plants. Economically, some technologies represent high investments, and their cost-effectiveness is contingent on continuous running; other plants are cheaper to build but usually burn more expensive fuels, and are preferably used as “peaking” or “mid-merit” plants.
The business model takes into account the fact that they will operate with fewer full-load hours.
Many plants run more economically at minimum load than if stopped for a few hours. Dispatchable renewables, such as reservoir hydropower and
STE, where available, offer better prospects for complementing PV generation, because their electrical capacity can be adjusted by design, for a given energy input (solar or water inflows) to be run as mid-merit or peaking capacities.
Storage would be needed to shift more PV electricity to other consumption times. However,
99% of grid-tied electricity storage capabilities today are pumped-storage hydropower (PHS) plants, with 150 GW in service worldwide and another 50 GW under construction or in development. Global storage capacities are

estimated to reach 600 GW in the 2DS hi-Ren, with
PHS providing most of the growth (IEA, 2014b). PHS will, in particular, be developed in areas with large penetration of wind power and little room for CSP plants, such as temperate regions. The potential for new PHS is important, because these plants do not require the large surface areas that characterise reservoir hydropower plants (IEA, 2012b; JRC,
2013). Storage at intermediate voltage levels can help address “hotspot” and grid congestion issues
— providing these issues occur frequently enough to make sufficient use of the storage capacities.
Decentralised battery storage is currently more expensive than PHS but may also have higher locational value, exactly like distributed PV generation, and for electricity storage competition with retail prices would be more favourable than with wholesale prices. Inevitably however, storage capacity optimisation would let some PV curtailment take place. California and Germany have already engaged in providing subsidies for distributed storage.
Further electrification of transport could also play a role in integrating variable PV output, because it offers storage (which is needed for driving autonomy in any case) and a potential means of reducing peak load (because most cars are stationary most of the time, offering time flexibility for charging). The external surface area of passenger cars and freight trucks is too small for embedded PV systems to provide a significant energy contribution. At present, these vehicles remain dependent on oil, a primary source of greenhouse gases (GHG) and other polluting emissions. Electric vehicles (EVs), whether partially electrified, such as plug-in hybrids, or full-fledged battery electric vehicles, are major options of reducing both oil dependence and environmental impacts. These vehicles offer electricity storage as an “absorbing capacity”, in what is termed the grid-to-vehicle (G2V) configuration. Provided charging can take place in the middle of the day,
G2V could help flatten the net load curve, i.e. the load curve minus PV (Denholm et al., 2013).
Otherwise, the risk is that uncontrolled EV charging may take place during evening peaks and increase the crest factor (i.e. the ratio of peak over average load) of the net load curve (Figure 18). Appropriate time-of-use (TOU) price signals could presumably provide incentives for appropriate behaviour. EV batteries connected to the grid may also have significant upward reserve value for the grid, even if the economics of routinely using them to provide energy to the grid are not favourable.

System integration: actions and milestones

35

Figure 18: Controlled versus uncontrolled EV charging effects on load net of PV
Solar PV
50
40

GW

30
20
10
0
0

24

48

72

96

120

72

96

120

72

Hours

96

120

Solar PV and uncontrolled EV charging
50
40

GW

30
20
10
0
0

24

48
Hours

Solar PV and controlled EV charging
50
40

GW

30
20
10
0
0

24

48
Hours

Electricity demand

Solar PV generation

EV charging

Net load

Notes: stylised electricity system for a five-day period with PV generation (top figure), with additional uncontrolled PV charging
(middle figure) and controlled charging (lower figure).

KEY POINT: Controlled charging of electric vehicles would reduce the volatility of net load and thus facilitate the integration of solar PV.

EV charging can increase PV self-consumption.
Experiments in southeast France have shown that a PV system installed over the parking space for one car could produce enough electricity to run a

36

four-passenger car over 10 000 kilometres per year.
Midday charging is more likely to happen at offices and other work sites using PV charging stations.
Charging during daylight would also increase if

Technology Roadmap Solar photovoltaic energy

proposals were implemented to feed electricity into electric vehicles, notably heavy trucks, while they are on the move, via induction or trolleys (see transport chapter in IEA, 2014b).

Grid parity holds potential but may also create illusions and raise concerns. The variability of the solar resource, together with the variability of electricity demand, limits self-consumption and its related benefits for electricity consumers who are also PV producers (known as “prosumers”), especially in the residential sector. For example, in temperate countries, most PV electricity generated in winter will be self-consumed, but the bulk of electricity consumption will still be drawn from the grid. On sunny summer days, the opposite holds true: less than half of PV electricity is self-consumed, but some electricity must still be drawn from the grid, especially during the evening peak (Figure 19).
In practice, even reaching the suggested levels of self-consumption may require the development of
“mini-grids” or “solar gardens”, where very shortterm demand variations of dozens of customers would largely cancel out, while communities, including renters and owners, would share the investment in larger and cheaper, well-oriented, well-designed and well-managed PV systems.

A recent IEA publication, The Power of Transformation
(IEA 2014d), investigates in detail the economics of integrating large shares of variable renewables such as wind power and solar PV into power systems. It shows that with timely re-optimisation of power systems, less inflexible base-load power, and more flexible mid-merit and peaking generation, total electricity costs at 45% of variable renewables would be increased by about 10% to 15% with current wind and PV technology costs. With assumptions relative to the decline of costs of these technologies consistent with the level of deployment in the hi-Ren
Scenario, total electricity costs would eventually increase by about 3% only as a result of the expansion of solar PV (IEA, 2014b).

Decentralised PV generation

The prospects for self-consumption are higher in sunnier countries, where consumption is partly driven by air-cooling loads, and for buildings other than residential. The load profile of office buildings or supermarkets suggests a better match with the solar resource, which reaches its maximum in the middle of the day (Figure 20).

About half the large PV deployment considered in this roadmap would take place on buildings or nearby (such as over parking lots). It rests in part on the concept of grid parity – when the cost of distributed PV generation is equal or below the perkWh component of retail electricity prices – and on self-consumption. Figure 19:  elf-consumption of stylised household
S
and rooftop PV system during a sunny day kW Without self-consumption

With self-consumption

2.5

2.0

1.5

1.0

0.5

0.0
0

4

8

12
Solar PV

16

20

24

Consumption

0

4

To the grid

8

12

16

20

24

From the grid

KEY POINT: Variability of both solar power and electricity demand limits self-consumption.

System integration: actions and milestones

37

Figure 20:  ourly electricity consumption profiles
H
for different building types in Germany
Household

Bakery

Office

Supermarket

120%

Capacity

100%
80%
60%
40%
20%
0%
0

8

16

24 0

8

16
Winter

24 0

8

16

24 0

8

16

24

Summer

Note: The reported annual consumption of these buildings spans three orders of magnitude. To allow for an easy comparison of load profiles, these curves have been scaled down with respect to the annual consumption of a typical German household.

KEY POINT: Many buildings other than residential offer better prospects for self-consumption.

Load management offers a significant opportunity to increase self-consumption — simply by shifting the use of some devices to hours of high solar generation. Chilled water, ice and other frozen media can be produced during the sunniest hours, and cheaply stored for hours to provide air conditioning, or cold for food and beverage storage and display. Decentralised battery storage could further increase self-consumption, but its exact role in the span of the scenarios depends on cost reductions that remain uncertain. This is not a go or no-go issue, though. The value of each marginal kWh of storage capacity decreases with its utilisation rate, but as battery costs decline with mass production and experience, progressively greater storage capacities will find their business model Building on the PV output and load profiles of Figure 19, Figure 21 illustrates how load management and small storage could each increase self-consumption by 10 percentage points. It also makes inflows to the grid more predictable.
The economic viability of PV systems depends on both the value of electricity savings due to selfconsumption, and the remuneration of injections into the grid. If the electricity injected into the grid were not remunerated at all, only small PV systems (e.g. of about 1 kW in Germany for a singlefamily household) would have a sufficient selfconsumption ratio to be economically viable.

38

For apartment buildings in an urban environment, small PV systems may be close to what is possible given the available space, especially if limited to roofs. A five-storey building housing ten families, with an average apartment surface of 85 m2, is likely to have a roof surface of about 170 m2. Assuming that only one-quarter is free for PV systems with acceptable tilt and orientation, and assuming by
2030 an efficiency of 20% (already exceeded by the best commercial modules), the maximum power from a rooftop system would be about 8.5 kW — less than 1 kW per family. Neither the small capacity nor the small available surface area in urban environments should thus be considered obstacles from a system perspective; on the contrary, they fit very well with each other to support PV selfproduction and self-consumption. By 2050, onethird of 3 billion families with a 1 kW system would represent 1 000 GW, over 20% of global capacity in the 2DS hi-Ren.
In sum, in urban areas the available roof or façade surface area is likely to be the limiting factor, and all PV electricity generated will likely be either self-consumed or consumed in the immediate whereabouts of generation, whether through specific cables, “mini-grids”, or simply the existing distribution infrastructure.
In less densely populated areas, an appropriate framework for self-consumption may incentivise

Technology Roadmap Solar photovoltaic energy

Figure 21:  ncreasing self-consumption with load management
I
(+10%) and small storage (+10%) kW Load management

Load management and storage

2.5

2.0

1.5

1.0

0.5

0.0
0

4
Solar PV

8

12
Consumption

16

20

24

To storage

0

4

From storage

8

12

16

To the grid

20

24

From the grid

KEY POINT: Load management and decentralised storage can increase self-consumption.

load management and thus ease integration (see next section). Curtailment can help avoid excess injection in distribution grids – and avoid the need for grid strengthening that would only serve in a few occurrences (typically, during sunny summer
Sundays). In Germany, new-built small-scale PV systems (