ENERGY

Electricity generation
technology cost projections
2017-2050

Jenny A Hayward and Paul W Graham

December 2017
Citation
Hayward, J.A. and Graham, P.W. 2017, Electricity generation technology cost projections: 2017-
2050, CSIRO, Australia.

Copyright
© Commonwealth Scientific and Industrial Research Organisation 2017. To the extent permitted
by law, all rights are reserved and no part of this publication covered by copyright may be
reproduced or copied in any form or by any means except with the written permission of CSIRO.

Important disclaimer
CSIRO advises that the information contained in this publication comprises general statements
based on scientific research. The reader is advised and needs to be aware that such information
may be incomplete or unable to be used in any specific situation. No reliance or actions must
therefore be made on that information without seeking prior expert professional, scientific and
technical advice. To the extent permitted by law, CSIRO (including its employees and consultants)
excludes all liability to any person for any consequences, including but not limited to all losses,
damages, costs, expenses and any other compensation, arising directly or indirectly from using this
publication (in part or in whole) and any information or material contained in it.
CSIRO is committed to providing web accessible content wherever possible. If you are having
difficulties with accessing this document please contact [email protected].
Electricity generation technology cost projections | i
Contents

Acknowledgments ...........................................................................................................................iv
Executive summary ......................................................................................................................... v
1 Introduction ........................................................................................................................ 1
2 Updates and changes to GALLM-E...................................................................................... 2
2.1 GALLM-E overview................................................................................................. 2
2.2 GALLM-E and GALLM-T interactions to project battery costs .............................. 3
2.3 Extension to a multi-region model ........................................................................ 3
2.4 Regional constraints on renewable resources ...................................................... 4
2.5 Longer learning period for selected technologies................................................. 5
3 Updated technology cost projections ................................................................................ 7
3.1 Global climate policy scenarios ............................................................................. 7
3.2 Discussion and comparison of cost projections .................................................... 7
3.3 Technology deployment by region ...................................................................... 11
3.4 Battery storage cost projections ......................................................................... 14
3.5 Trends in cost projections ................................................................................... 15
4 Levelised costs of electricity ............................................................................................. 17
4.1 Conventional levelised cost of electricity estimates ........................................... 17
4.2 Extended levelised cost of electricity estimates ................................................. 19
Appendix A Projection data tables.......................................................................................... 22
5 References ........................................................................................................................ 26

ii | Electricity generation technology cost projections

Figures
Figure 2-1 – Upper limit on percentage of demand that can be met by various renewable
technologies in 2050 ....................................................................................................................... 5
Figure 3-1: Brown coal with CCS (left) and black coal with CCS (right) .......................................... 9
Figure 3-2: Gas with CCS (left) and nuclear (right) ......................................................................... 9
Figure 3-3: Solar thermal with 6 hours storage (left) and wind (right) .......................................... 9
Figure 3-4: Rooftop solar photovoltaics (left) and large scale solar photovoltaics (right) ............. 9
Figure 3-5: Brown coal with CCS (left) and black coal with CCS (right) ........................................ 10
Figure 3-6: Gas with CCS (left) and nuclear (right) ....................................................................... 10
Figure 3-7: Solar thermal with 6 hours storage (left) and wind (right) ........................................ 10
Figure 3-8: Rooftop solar photovoltaics (left) and large scale solar photovoltaics (right) ........... 10
Figure 3-9 – Projected share of generation by broad technology category under the 4-degree
scenario for selected regions for the years 2030 and 2050 ......................................................... 12
Figure 3-10 – Projected share of generation by broad technology category under the 2-degree
scenario for selected regions for the years 2030 and 2050 ......................................................... 12
Figure 3-11 – Projected 2050 global share of generation by technology category under the 4-
degree (left figure) and 2-degree (right figure) scenarios. ........................................................... 13
Figure 3-12: Comparison of battery only cost projections: 2017 update and previous
projections, 2017 AUS dollars ....................................................................................................... 14
Figure 3-13: Previous CSIRO and other publicly available electricity generation technology cost
projections for 2030 ...................................................................................................................... 16
Figure 4-1: Conventional LCOE estimates for selected technologies ........................................... 18
Figure 4-2: Extended LCOE for combined wind and solar field for a scenario with increasing
variable generation share ............................................................................................................. 21

Tables
Table 4-1: High and low values for key LCOE assumptions .......................................................... 18

Apx Table A.1: Projections data table for 4-degrees climate ambition scenario, 2017 Australian
dollars sent-out ............................................................................................................................. 23
Apx Table A.2: Projections data table for 2-degrees climate ambition scenario, 2017 Australian
dollars sent-out ............................................................................................................................. 24

Electricity generation technology cost projections | iii

Acknowledgments

The authors acknowledge the input of the CSIRO reviewers. Any remaining errors or omissions
remain the responsibility of the authors.

iv | Electricity generation technology cost projections

Executive summary

The continuing importance of electricity futures to the nation and the fast rate of change in costs
of some technologies, means that regularly updated electricity generation technology cost
projections remain an important public resource. CSIRO’s approach to cost projections is based on
using the most objective and data driven method available which simultaneously projects both
global and local deployment of electricity generation technology and the cost reductions achieved
due to the well-known learning by doing effect.
In the past, CSIRO has provided standalone projections or contributed to studies led by other
organisations at irregular intervals since 2011, with the last being the Australian Power Generation
Technology Report in 2015. Those projections are significantly out of date and therefore it is
appropriate that CSIRO provide this 2017 update to cost projections.
For this 2017 update, we have improved the regional detail of our Global and Local Learning
Model for Electricity (GALLM-E) and have constructed a similar model for transport (GALLM-T) to
assist in projecting battery storage costs which, while potentially very important for the electricity
sector, are being deployed at a faster rate in the transport sector (making this sector the strongest
source of learning and thus cost reductions). We have also updated our assumptions on global
renewable resources availability for each region and extended the learning rate for selected
technologies which have demonstrated an extended period of higher learning.
The updated projections indicate the trend we have seen in the last few years whereby solar
photovoltaics, wind and battery storage technologies reduce their costs at a faster rate than most
other technologies continues. Compared to previous projections, these substantial cost reductions
in renewables have also begun to impact the cost and adoption of other technologies. For
example, in our 4-degrees climate ambition scenario we project very low adoption of carbon
capture and storage (CCS) and subsequently very little cost reductions in that technology. In effect,
the ongoing cost reductions in renewables has crowded out investment in other technologies.
In the 2-degrees climate ambition scenario the renewable energy resource constraints, together
with a stronger carbon price signal, do allow for investment in a wider diversity of electricity
generation technologies. Significant adoption and cost reductions in CCS are projected. However,
they are less than those that we projected for a similar scenario in the 2015 projections. This is
because in either climate ambition scenario, the world deploys a greater share of low cost
renewables in response to emission constraints compared to other options.
The capital cost projections are primarily designed to be included in Australian electricity
modelling studies as scenario inputs. However, we also provide levelised cost of electricity (LCOE)
estimates which are useful for providing a quick indication of what the capital cost projections
mean for the future cost of electricity. LCOE estimates have been justifiably criticised for an
absence of standardised plant performance and financial assumptions and failing to take into
account the additional cost of supporting variable generation technologies to be reliable. These
two flaws mean that LCOE comparisons within (i.e. between technologies) and across alternative
LCOE projection studies can be meaningless.

Electricity generation technology cost projections | v

To address these issues we provide information about our LCOE calculation assumptions and
present a preliminary estimate of the cost of supporting variable renewable generation to be
reliable as their share of electricity generation increases. This “extended” LCOE estimate, which
includes the additional costs of very high renewable shares, indicates that the LCOE of renewables
begins to approach that of CCS and nuclear power (i.e. increasing towards $120/MWh at 90%
variable renewable share compared to $30-50/MWh under the conventional LCOE calculation).
However, given variable renewable electricity generation shares in most Australian states are low,
the cost of supporting renewables also remains low and therefore variable renewables are not
expected to face a significant low emission competitor for some considerable time.

vi | Electricity generation technology cost projections

1 Introduction

CSIRO has been publishing electricity generation technology cost projections since 2011 which
coincided with the development of the Global and Local Learning Model for Electricity (GALLM-E)
which is a quantitative modelling framework encapsulating CSIRO’s projection methodology
(Hayward, et al., 2011) (Hayward & Graham, 2012) (Hayward & Graham, 2013). We will discuss the
GALLM-E methodology and changes made for this 2017 update in the next section.
Since 2011, CSIRO has applied GALLM-E in coordination with collaborative cost assessment studies
including the 2012 and 2013 Australian Energy Technology Assessment (Bureau of Resource and
Energy Economics (BREE), 2013) (Bureau of Resource and Energy Economics (BREE), 2012) and the
2015 Australian Power Generation Technology Report (CO2CRC, 2015). These all included a
stakeholder engagement process, mainly focussed on seeking agreement around the current level
of a technology’s capital costs but sometimes providing advice on other GALLM-E data
assumptions. CSIRO added a battery cost forecast to GALLM-E’s capabilities which was published
in (Brinsmead, et al., 2015).
The continuing importance of energy futures and the fast rate of change in costs of some
technologies, means that the 2015 projections are out of date. Therefore, it is appropriate that
CSIRO provide this 2017 update to cost projections. While on this occasion no existing stakeholder
engagement process was available, we have sought to include updated current cost data and
other inputs from available literature or public announcements.
Because GALLM-E has been described in previous publications we will only provide a brief
explanation of the model and focus our methodology discussion on changes and improvements
that have since been made. Following that, we provide a discussion and comparison of the
updated cost projections to previous work. We conclude with a calculation of the levelised costs of
electricity (LCOE), including an estimate of the extended LCOE of variable renewable generation
technologies when the cost of supporting technologies is included at different variable renewable
electricity generation shares.

Electricity generation technology cost projections | 1

2 Updates and changes to GALLM-E

2.1 GALLM-E overview

After considering a broad range of technology cost projection approaches1 , GALLM-E was chosen
primarily because the cost projection methodology does not require as many subjective choices as
other approaches. The method is highly data driven and the outputs can be traced through and
explained by the assumptions. However, GALLM-E is computationally demanding.
GALLM-E is solved as a mixed integer linear program in which total costs of electricity supply in
each region is minimised to meet annual electricity demand over the projection period. The model
features endogenous technological learning which means that the electricity technology mix and
the cost of electricity generation technologies are solved simultaneously as outputs from the
model. Learning occurs at the global and local level. This means that technology costs may fall in a
region both because of technology deployment elsewhere in the world or because of local
deployment. Each technology is divided into components subject to local learning (e.g. installation
costs), global learning (e.g. globally standardised manufactured elements) and no learning at all
(e.g. mature parts of plant design).
A key feature of GALLM-E is the technology learning curves, also known as experience curves.
Learning curves refer to the observed phenomenon that the costs of new technologies tend to
reduce with the cumulative production of the technology (i.e. ‘learning by doing’). Also, costs tend
to reduce by an approximately constant factor for each doubling of cumulative production. This
observation makes it possible to create cost projections based on projections of the future uptake
of a technology. Projections are created from a mathematical equation as follows:
ICt = IC0 × CCt–b
where IC is the investment cost of a technology at CC cumulative capacity at a given future point in
time, t. IC0 is the investment cost at a given starting period or capacity and b is the learning index.
This index is related to the learning rate as follows:
LR = 100 – 2–b
where LR is the learning rate, represented as a percentage, and which can be interpreted as the
percentage reduction in per unit costs resulting from a doubling of cumulative capacity.
Projections of the global and local uptake of a technology needs to be known in order to project
changes in costs. However, uptake itself depends on projected costs. To resolve this
interdependency, models such as GALLM-E are applied to project cost and uptake simultaneously.

1 Table 1 in Hayward and Graham (2012) compares approaches

2 | Electricity generation technology cost projections

2.1.1 GALLM-T

The learning rate method can be applied to almost any technology where sufficient data is
available and recognising this, CSIRO has extended the approach to the transport sector in a model
called GALLM-T. The development of GALLM-T was necessary in order to project the costs of
battery technologies. While battery storage is becoming increasingly important in the electricity
sector, their cost reduction is still driven mostly by learning in the transport sector, specifically
through electric vehicle deployment. A second motivation is that CSIRO also conducts a range of
transport futures modelling and research and GALLM-T provides a new source of cost data for
such projects (Graham & Reedman, 2015) (Reedman & Graham, 2016).

2.2 GALLM-E and GALLM-T interactions to project battery costs

GALLM-T simultaneously solves for the mix of transport fuel supply and alternative engine
technologies to meet global transport demand and produces the rate of change in their costs
according to the assumed learning rates. In this context we only require the projected deployment
of batteries in electric vehicles and to ensure additional electricity demand from vehicle
electrification is included in the electricity demand to be met in GALLM-E.
The amount of batteries deployed over time via electric vehicle sales is fed into GALLM-E as an
exogenous (fixed) input. GALLM-E is free to choose to deploy additional battery capacity in
stationary electricity storage applications depending on each region’s needs (we have not yet
considered the potential for “vehicle to grid” which would reduce the need for separate stationary
energy battery capacity). Our expectation is that as the cost of solar photovoltaic and wind
generation technologies decreases and their deployment increases, it may be cost effective for
some regions to deploy battery storage as a means of maintaining requirements for reliability.
The projected cost of batteries is a function of the learning rate applied to the sum of both
electricity and transport sector battery capacity deployed. Ideally, we should iterate between
GALLM-E and GALLM-T to find an equilibrium level of deployment and costs since costs and
deployment in one model will impact the other. However, in practice, we find that electric vehicle
deployment saturates in GALLM-T and is not significantly impacted by additional battery
deployment and subsequent cost reductions from electricity sector battery applications.

2.3 Extension to a multi-region model

The original version of GALLM-E had three regions (including Australia), which may be sufficient
for providing cost projections in Australia but it was unable to capture factors such as renewable
resources, trade between regions, energy policies and differences in fuel prices down to a suitable
level of granularity. Based on the approach the OECD takes to regional disaggregation, GALLM-E
now has 13 regions, some of which are individual countries. Where a region and country overlap,
the country has been excluded from that region.
The regions are Africa (AFR), Australia (AUS), China (CHI), Eastern Europe (EUE), Former Soviet
Union (FSU), India (IND), Japan (JPN), Latin America (LAM), Middle East (MEA), North America
(NAM), OECD Pacific (PAO), Rest of Asia (SEA), and Western Europe (EUW).

Electricity generation technology cost projections | 3

GALLM-E now projects the capital cost and uptake of electricity generation technologies in each of
these regions. Each region has its own initial installed capacities of electricity generation
technologies, capital and operating costs, demand for electricity, fuel prices, energy resources
(renewable, fossil fuel and carbon dioxide storage sites), carbon prices and other climate change
and energy policies. Electricity trade between EUE, EUW and FSU has been included.
GALLM-T features the same regions as GALLM-E so uptake of technologies can be shared between
the models. Each region in GALLM-T has its own initial installed capacity of fuel conversion and
new vehicle technologies (battery and fuel cell vehicles), fuel prices, feedstock prices, demand for
transport by mode, carbon prices, transport policies such as ethanol blends and other energy
policies. Trade in fuels is included between all regions. Trade is assumed to occur by shipping.

2.4 Regional constraints on renewable resources

A key feature of recent developments in technology costs in the last decade has been very large
reductions in the cost of solar photovoltaic and wind generation technology. Furthermore, large
reductions in the costs of battery storage could mean the share of these low cost variable
renewable generation technologies is not as limited as previously thought by reliability
constraints.
The lesser influence of both financial and reliability constraints to solar and wind technology
adoption means that we must take greater care in ensuring that we understand at what point
resource constraints, including other renewables such as solar thermal, geothermal and wave
energy may play a role in limiting technology adoption.
While there has for many years been discussion of the potential for new cross-border transmission
schemes for accessing high quality renewable resources (e.g. MENA-EU Solar Energy Super Grid;
Asia Super Grid), we assume no such schemes are developed during our projection period, due to
their great uncertainty. Existing trade via inter-country transmission, such as within the European
region, is accounted for.
Each region’s total supply of renewable generation is therefore limited by the quality of domestic
resources. CSIRO sought information from a variety of sources to determine maximum supply (see
for example, (Government of India, 2016) (Chandler, et al., 2014). Given that supply is assumed to
be fixed, if a region’s electricity demand is rising, the maximum percentage share of renewables
falls over time. Figure 2-1 shows the maximum percentage of demand that can be met by four
renewable technologies for each region in the year 2050.

4 | Electricity generation technology cost projections

 400%

350%

300%
Percentage share of renewable technology

250%

200%

150%

100%

50%

0%
AUS CHI EUE EUW FSU IND JPN LAM MEA NAM PAO SEA
Rooftop PV Large scale PV CSP Wind

Figure 2-1 – Upper limit on percentage of demand that can be met by various renewable technologies in 2050

Data on this topic will likely change over time. Some particular areas for future review are building
integrated solar and offshore wind. The upper limit on rooftop and building integrated solar
depends on the form in which solar panels or other yet to be demonstrated solar structural
materials take and their ease of incorporation into existing and new structures. Business models
for sharing the benefits of solar electricity production across building tenants and owners could
also broaden building integrated solar adoption. Free floating offshore wind technologies could
open up a wider area of ocean resources if proven cost effective and robust.

2.5 Longer learning period for selected technologies

The learning by doing approach allows technologies to achieve rapid cost reductions at the early
stages of their development. However, the requirement that a technology’s capacity must double
before learning is achieved eventually becomes a major limiting factor on achieving ongoing cost
reductions. In contrast, some technologies, due to their modular nature can continue doubling
capacity for longer than others, even after gaining a significant share of the market. To limit
mature modular technology learning, as a standard assumption, we decrease the rate of learning
once the capacity is at significant market share. This approach effectively imposes a soft floor on
the cost reduction that can be achieved within the projection period. This approach is supported
by the existence of engineering and thermodynamic limits such as on the size or strength of a
component, or on the theoretical maximum energy conversion efficiency.
However, we have chosen to modify this approach for two technologies: solar photovoltaic
electricity generation and battery storage. These two technologies are selected because they do
not appear to have a strongly settled set of underlying material components. While silicon panels

Electricity generation technology cost projections | 5

are currently the dominant solar photovoltaic technology, there are a number of alternative
materials being explored where significant progress is being made in achieving efficient energy
conversion and in different systems for installing the solar conversion system (Jacoby, 2016)
(Fraunhofer ISE, 2015). This could mean improvements not only in the panel but in the installation
and balance of system costs. Similarly, alternative battery chemistries and configurations also
continue to be explored with no obvious limit on what might be achieved (Miller, 2017).
Since future science and engineering advances are likely but unknowable in advance, rather than
apply the standard approach which implies innovations peter out with growing maturity, the
alternative approach we take for these two technologies is to apply their high historically observed
learning rate indefinitely. In practice, this means these technologies can achieve steeper cost
reduction curves for longer than other technologies.

6 | Electricity generation technology cost projections

3 Updated technology cost projections

3.1 Global climate policy scenarios

The assumed level of global ambition to reduce greenhouse gas emissions to address the risk of
climate change is a major driver for low emission technology deployment in GALLM-E. In 2015 we
assumed two global climate policy ambition scenarios which we called 550ppm and 450ppm and
which had corresponding carbon price assumptions.
For this 2017 update, we have expressed the scenarios in terms of expected changes in global
average temperature increases above pre-industrial levels of 4-degrees and 2-degrees. This
approach was developed as part of an update to the 2015 Australian National Outlook2. For each
climate ambition scenario, a global carbon price path differentiated by region was constructed
from several sources. For the near term (to 2025 for the 4-degree scenario and to 2020 for the 2-
degree scenario), regional prices were taken from the International Energy Agency (IEA) where a
projected price is provided (IEA, 2016). For countries/regions not covered by the IEA, a zero price
is applied unless the country is shown as having a price in (World Bank, Ecofys and Vivid
Economics, 2016). It is assumed that international policy cooperation results in a global uniform
carbon price in the long term, which is derived from (IPCC, 2014). For the 2-degree global warming
scenario this occurs relatively early, from 2020 onwards. For the 4-degree global warming
scenario, uniform global pricing does not occur until 2040, and there is furthermore a transition
period between 2025 and 2040 where regional prices converge from their near term trajectories
to the long term uniform global price – annual prices are constructed assuming a constant
percentage growth rate over the 15 year transition period.
In the 2-degree scenario, prices increase at a rate of 6% pa from $20 USD (2015 prices) per tCO2e,
in 2020 reaching $86 in 2045 and $115 in 2050. In the 4-degree scenario, uniform prices start at
$40 in 2045, increasing more slowly by 1% pa to only $42 in 2050.
Besides carbon pricing, countries have renewable energy targets, feed-in-tariffs and related
policies to encourage emissions reduction and promote renewable technologies. The IEA’s three
climate change scenarios: ‘Current Policies’, ‘New Policies’ and ‘450ppm’ were used to match non-
carbon price policies to the 2- and 4-degree scenarios.

3.2 Discussion and comparison of cost projections

For selected low emission technologies, we compare the 2017 4-degrees scenario to the 2015
550ppm scenario cost projections (Figure 3-1 to Figure 3-4) and of the 2017 2-degrees scenario to
the 2015 450ppm scenario cost projections (Figure 3-5 to Figure 3-8). We have not included
mature technologies in these figures because they are assumed to be no longer subject to costs
decreases that depend on cumulative deployment. Rather, a small fixed annual decline for general

2 http://www.csiro.au/nationaloutlook/

Electricity generation technology cost projections | 7

productivity improvements is applied across their cost inputs - an assumption which has not
changed compared to previous projections. We have also excluded less prospective technologies
such as enhanced geothermal. We make available the underlying cost projection data for all
technologies in Appendix A. Note that the costs shown are for Australian deployed technologies.
Costs differ in each country due to local learning, particularly on balance of plant (BOP) costs.
Given most assumptions remain similar to the 2015 analysis we expect that most differences in
the 2017 cost projections are due to the new carbon price assumptions, the changes to the model
described in Section 2 and the subsequent different rates of technology deployment projected by
the model.
One theme that emerges from the comparison is that carbon capture and storage (CCS)
technologies (brown coal, black coal and gas) do not achieve as much cost reduction by 2050 in
the 2017 projections as in the 2015 projections. The lack of a strong carbon price signal in the 4-
degrees scenarios means that CCS is not taken up in that scenario and any cost reduction reflects
general productivity improvements rather than learning from deployment.
In the 2-degrees scenario, CCS is deployed, however it is deployed later (the mid rather than early
2020s) and to a lower level of capacity than in the 2015 projections. This likely reflects the impact
of renewables taking up a greater share of generation owing to their significant cost reductions
between 2015 and 2017 that have been incorporated into the projections.
A recurring limitation of GALLM-E’s modelling of CCS is that the model is not able to examine
retrofit projects. It is possible that some CCS retrofit projects could proceed at lower cost than
new build projects and speed up learning for this technology category. This is an opportunity for
consideration in the next decade. However, as existing coal plant ages, the window for life
extension with CCS will also close because new-build CCS plant, with longer achievable life will be
available at lower lifetime cost. The challenge for CCS retrofit plant is that, in the period in which
they are lower cost than new build CCS plant, they lack a strong emissions constraint or carbon
price incentive to support the investment.
The 2017 cost projections for nuclear power remain very similar to 2015. This reflects two
constraints. The first is that many countries choose to limit new or expanded nuclear programs,
particularly in developed countries. As a consequence, while nuclear is expanding in some regions
these are partially offset by retirements elsewhere. Thus, it is difficult for conventional nuclear
power to double capacity and achieve further learning. The second constraint is that the model
does not include new designs of nuclear plants or modular nuclear as a distinct technology type. It
is possible that these technologies have better prospects.

8 | Electricity generation technology cost projections

 10000 9000
9000 8000
8000
7000
7000
6000
6000
5000
5000
$/kW

$/kW
4000
4000
3000
3000
2000 2000

1000 1000
0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050

2017 4 degrees 2015 550ppm 2017 4 degrees 2015 550ppm

Figure 3-1: Brown coal with CCS (left) and black coal with CCS (right)

3500 10000
9000
3000
8000
2500 7000
6000
2000
5000

$/kW
$/kW

1500 4000

1000 3000
2000
500
1000
0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050

2017 4 degrees 2015 550ppm 2017 4 degrees 2015 550ppm

Figure 3-2: Gas with CCS (left) and nuclear (right)

9000 3000

8000
2500
7000

6000 2000

5000
1500
$/kW
$/kW

4000

3000 1000

2000
500
1000

0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050
2017 4 degrees 2015 550ppm 2017 4 degrees 2015 550ppm

Figure 3-3: Solar thermal with 6 hours storage (left) and wind (right)

2500 2500

2000 2000

1500 1500
$/kW

$/kW

1000 1000

500 500

0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050

2017 4 degrees 2015 550ppm 2017 4 degrees 2015 550ppm

Figure 3-4: Rooftop solar photovoltaics (left) and large scale solar photovoltaics (right)

Comparison of 2017 4-degrees and 2015 550ppm scenario costs, 2017 AUS dollars sent-out basis
Electricity generation technology cost projections | 9
 10000 9000
9000 8000
8000 7000
7000 6000
6000
5000
5000

$/kW
$/kW

4000
4000
3000
3000
2000 2000

1000 1000
0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050
2017 2 degrees 2015 450ppm 2017 2 degrees 2015 450ppm

Figure 3-5: Brown coal with CCS (left) and black coal with CCS (right)

3500 10000
9000
3000
8000
2500 7000

2000 6000
5000
$/kW

$/kW
1500 4000
1000 3000
2000
500
1000
0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050
2017 2 degrees 2015 450ppm 2017 2 degrees 2015 450ppm

Figure 3-6: Gas with CCS (left) and nuclear (right)

9000 3000
8000
2500
7000
6000 2000
5000
1500
$/kW

$/kW

4000
3000 1000
2000
500
1000
0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050
2017 2 degrees 2015 450ppm 2017 2 degrees 2015 450ppm

Figure 3-7: Solar thermal with 6 hours storage (left) and wind (right)

2500 2500

2000 2000

1500 1500
$/kW
$/kW

1000 1000

500 500

0 0
2015 2020 2025 2030 2035 2040 2045 2050 2015 2020 2025 2030 2035 2040 2045 2050
2017 2 degrees 2015 450ppm 2017 2 degrees 2015 450ppm

Figure 3-8: Rooftop solar photovoltaics (left) and large scale solar photovoltaics (right)

Comparison of 2017 2-degrees and 2015 450ppm scenario costs, 2017 AUS dollars sent-out basis

10 | Electricity generation technology cost projections

In terms of renewable technologies, wind rooftop solar PV and solar thermal have decreased
faster than expected (in 2015) in both 2017 climate ambition scenarios. Large scale solar PV has
also improved, however, it remains more aligned with the previous 2015 cost in the long term. By
2050, under either scenario the 2017 cost projection for wind and rooftop PV are lower than that
projected in 2015. For wind this reflects greater adoption owing to lower near term costs and
more generous assumptions about regional resource constraints. For rooftop solar PV, the recent
cost estimate reductions have also improved the longer term outlook, together with our new
assumption that learning for both the panel and balance of system can continue at faster rates for
longer. Large scale PV does not achieve as great a cost reduction, and is in fact less than that
projected in either of the 2015 scenarios. This is not due to lack of learning on the panels but the
revised assumption that the balance of system in large scale plants are not as fertile for innovation
as rooftop or building integrated systems.
The projected costs for solar thermal technology have changed mainly due to information being
available from the first commercial plant that is due for completion in South Australia in 2020. This
follows anecdotal evidence that projects in other regions of the world are also achieving lower
costs. Up to 2025, cost reductions and deployment are projected to stall briefly as other
technologies are prioritised. Cost reductions resume at a steady rate to 2050 leading to almost half
the cost level as that projected in both of the 2015 scenario projections.

3.3 Technology deployment by region

The projected share of electricity generation by technology in key world regions is shown in Figure
3-9 for the 4-degree scenario. Europe is a combination of GALLM-E regions EUW and EUE and Asia
includes CHI, IND, JPN, PAO and SEA. “Solar” refers to both photovoltaic and thermal systems.
Brown and black coal are aggregated into “Coal”.
All regions see a large drop in coal-fired electricity generation and an increase in solar between
2030 and 2050 however the shares of other technologies are more variable. This is due to
saturation of available renewable resources which means that the maximum share reduces as
demand increases. For instance, North America sees a reduction in the share of wind generation
but an increase in other renewables, which includes geothermal and ocean energy. Asia has an
increase in both wind and other renewables between 2030 and 2050 and Europe sees a large
increase in wind and solar but a drop in the share of other renewables.
The projected share of electricity generation by technology for the 2-degree scenario is shown in
Figure 3-10. Under the 2-degrees scenario there is also a large reduction in black and brown coal
generation that is somewhat offset by CCS. CCS technologies generate some electricity in all
regions in 2030, and make a large contribution in Asia in particular by 2050. The solar shares are
very similar to the 4-degree scenario however there is some variation across scenarios in wind and
other renewable shares. This means that the uptake of solar is not impacted by the level of the
carbon price but other renewable technologies are. Demand for electricity, which can also impact
the shares of different technologies, is lower under the 2-degree scenario. The higher carbon price
means that in 2030 North America has more gas-fired than coal-fired generation.

Electricity generation technology cost projections | 11

 1.2

0.8
Share of generation by technology

0.6

0.4

0.2

0
2030 2050 2030 2050 2030 2050
Europe North America Asia
Coal Fossil CCS Gas Oil Nuclear Hydro Solar Wind Other renewables

Figure 3-9 – Projected share of generation by broad technology category under the 4-degree scenario for selected
regions for the years 2030 and 2050

1.2

0.8
Share of generation by technology

0.6

0.4

0.2

0
2030 2050 2030 2050 2030 2050
Europe North America Asia
Coal Fossil CCS Gas Oil Nuclear Hydro Solar Wind Other renewables

Figure 3-10 – Projected share of generation by broad technology category under the 2-degree scenario for selected
regions for the years 2030 and 2050

12 | Electricity generation technology cost projections

 6% 7% 3%
19%
20%
18% 17% Coal
Fossil CCS
0%
Gas
Oil
Nuclear
17% Hydro
15% Solar
Wind
22% 23% Other renewables
1% 0%
9% 8%
7% 7%

Figure 3-11 – Projected 2050 global share of generation by technology category under the 4-degree (left figure) and
2-degree (right figure) scenarios.

The projected global shares by generation technology type in the year 2050 under each scenario
are shown in Figure 3-11. The clear differentiating factor between scenarios is the quantity of coal
generation and CCS; CCS essentially replaces coal’s share of generation under the 2-degree
scenario. The renewable share increases slightly from 53% to 54%. The increase is limited because
of a number of factors. The first is that, while carbon prices have increased in the 2-degree
scenario, in some regions renewable generation has already reached the assumed resource limits
in the 4-degree scenario, particularly in the Asian region. The second is that in the 2-degree
scenario, because CCS is deployed, CCS capital costs are significantly lower (around half their level
in the 4-degree scenario) and it is therefore more competitive with renewables. The third factor is
that, given the global scale of the GALLM-E model the information applied to determine the costs
of supporting the reliability of variable renewable generation are conservative compared to more
detailed national level electricity modelling3. Our expectation is that the CCS share remains highly
uncertain and will change over time as our understanding of the cost of supporting the reliability
of variable renewable generation in each region improves.
As a source of comparison, the IEA (2017) World Energy Outlook projects the share of CCS in global
electricity generation to be less than 1% by 2040 in its ‘New Policies’ scenarios and 6% by 2040 in
its ‘Sustainable Development’ scenario. Those two scenarios are most similar to the 4-degree and
2-degree scenarios respectively. Aligning the projection date at 2040, the global shares for CCS in
our 4-degree and 2-degrees scenarios are less than 1% and 10% respectively.
The total amount (in GJ) of fossil fuels consumed is slightly higher under the 2-degree scenario by
2050 owing to the reduced fuel efficiency of CCS plants on an electricity sent-out basis. However,
note that CCS is applied to a mix of coal and gas plant types and therefore this still represents a
reduction in coal-fired electricity, losing market share to gas with CCS. A 90% carbon dioxide
capture rate has been assumed in GALLM-E.

3The analysis in Section 4 shows calculations for Australia of the cost of making variable renewables reliable based on detailed simulations of
different levels of variable renewable electricity generation share in each major state region. Ideally this analysis should be repeated with
country/region specific variable renewable generation profiles but was beyond the current scope of work.

Electricity generation technology cost projections | 13

3.4 Battery storage cost projections
As discussed, this update included changes in our modelling approach in regard to battery storage
cost projections by including transport sector projections. These transport projections, which
include adoption of electric vehicles, are a significant driver for deployment and subsequent cost
reductions in battery storage which can be taken up by the electricity sector. A second change was
to allow batteries to experience a high rate of learning for a longer period than in our previous
projections.
As a result of these two changes in method and/or assumptions the updated battery cost
projections were calculated and are shown in Figure 3-12. For comparison, the figure shows two
previous CSIRO cost projections. Our first projection was provided in an energy storage trends
report commissioned by the Australian Energy Market Commission (Brinsmead, et al., 2015). This
earlier projection provided both a low and high cost case. However, the low cost case has been
more often used in our subsequent work as battery costs continued to fall rapidly following
publication of the report. This low cost case projection was slightly amended for the Electricity
Network Transformation Roadmap to include a 20% lower long term end point for use in scenarios
with 100% renewable shares which would include greater battery adoption and therefore greater
learning.
500

450
"Energy storage trends" report low case
400
Electricity Network Transformation Roadmap
350
2017 update - 2 degrees

300

250
$/kWh

200

150

100

50

0
2015 2020 2025 2030 2035 2040 2045 2050

Figure 3-12: Comparison of battery only cost projections: 2017 update and previous projections, 2017 AUS dollars

We provide separate cost projections for the battery only and the balance of system costs. The
2017 updated battery only cost projection recognises that costs have continued to fall since 2015
and may continue to do so at a similar rate through to 2020 after which we project the rate of
decline to moderate. The level of costs projected in 2050 is around half that of the earlier 2015
low cost projections.
14 | Electricity generation technology cost projections
In utility scale deployments, the balance of plant costs are very significant. We project that
balance of plant costs will almost halve over the projection period from around $520/kWh to
$270/kWh in 2050 (in real 2017 dollars) based on learning rates from (Schmidt, et al., 2017). There
is a significant uncertainty range around balance of system costs which is difficult to define as they
are context dependent. Utility scale battery storage projects have the potential to reduce costs by
co-locating with wind and solar photovoltaic plant. The degree to which that integration is planned
and that resources can be shared will see cost outcomes achieved at the lower end of the range.

3.5 Trends in cost projections

Since the turn of the 21st century the global technology market has undergone a series of marked
trends. These include:
 Initially low gas turbine costs reflecting delayed orders following the Asian financial crisis of the
late 1990s which impacted economic growth and power station building plans.
 Rising prices for coal and wind plants. This reflected strong global economic growth, particularly
in China, that motivated strong power station building programs as well as a widely held
preference for these two mature technologies. There was also a simultaneous increase in power
station raw material costs (such as prices for steel). The preference for coal plant reflected high
oil and gas prices. The preference for wind plant was due to many countries seeking to
demonstrate adoption of low emission technologies with wind being by far the lowest cost large
scale technology.
 Rising prices for solar photovoltaic plant during a period in which there was a temporary
shortage of silicon supply.
Following after these trends, the most recent and dominant trend has been the very rapid fall in
solar photovoltaic and wind technology costs and at the same time a stagnation or increase in the
(projected and actual) costs of alternative low emission technologies such as carbon capture and
storage (CCS) and nuclear.
The previous decade’s rising trend in prices in part explains why, even in 2011, some projections to
2030 still had not anticipated the coming fall in wind and solar photovoltaic technology costs
(Figure 3-13). The greatest change in projected costs has been in solar photovoltaics with the
outlook for costs halving twice over several studies as data is updated to reflect more current cost
data and historical learning rates adjusted.

Electricity generation technology cost projections | 15

 MMA (2008) US DOE (2009) EPRI (2010) CSIRO (2011) SKM MMA (2011)
CSIRO/AETA (2013) CSIRO/APGT (2015) Jacobs (2017) CSIRO (2017)
8000

7000

6000

5000
2017 $/kW

4000

3000

2000

1000

0
Brown coal with CCS Black coal with CCS Large scale solar PV Wind Solar thermal

Figure 3-13: Previous CSIRO and other publicly available electricity generation technology cost projections for 2030

Notes: All CSIRO projections are shown in green with earliest starting on the left where available. The 2015 APGT
was written by EPRI with cost projections provided by CSIRO. Similarly, AETA was written by the Bureau of
Resources and Energy Economics and consultants. MMA, SKM MMA and Jacobs represent a reasonably continuous
projections team with changes to their trading name. Consequently their projections are all shown in blue with
earliest projections starting on the left where available.

The projected cost for CCS has increased by around 50% to 100% across the projections. This is
largely due to slow progress in demonstrating commercial scale CCS plant, due to the high
minimum scale and hence high cost of each demonstration. Earlier cost projections had also
anticipated faster and stronger action to address climate change, including carbon prices high
enough to fund CCS projects. Instead, in most countries, high carbon prices have not emerged as
yet and policy support has been strongest for renewable electricity generation plant through
renewable energy targets, feed in tariffs and other policy mechanisms.
Projections developed across the period 2008 to 2017 for solar thermal costs in 2030 have been
varied. Early expectations in 2011 that solar thermal may achieve similar cost reductions to that
achieved by solar photovoltaics were moderated by 2015. This reflected a concern that solar
thermal has experienced a crowding out of investment, as costs reductions already achieved by
wind and solar photovoltaics snow-balled, soaking up available investment funds. However, recent
costs demonstrated by current projects has restored some confidence. In the long term, as the
share of renewable electricity generation increases in the future and concerns about the variability
of wind and solar photovoltaics begin to be priced in the market through various policies or
market mechanisms, solar thermal will be better positioned to attract a greater share of
investment.

16 | Electricity generation technology cost projections

4 Levelised costs of electricity

CSIRO uses projected capital costs of electricity generation technologies directly in its modelling
rather than deriving the levelised costs of electricity (LCOE) to determine the implications of those
capital costs. However, LCOE is useful for non-technical audiences to make simple comparisons
and projections of the future costs of electricity supply. Even so, the simple comparisons which are
made using LCOE are becoming less useful, as we see the cost of variable renewable electricity
generation technologies fall and their share of generation rise. Variable electricity generation
includes generation from solar photovoltaics and wind power which delivers energy as variable
climate weather conditions make the solar and wind resources available. When the share of
variable electricity generation is low it can rely on the flexibility of other generation capacity to
support system reliability by providing more power when variable supply falls. However, at
sufficiently high shares of variable electricity generation, the power system requires additional
supporting technologies, such as storage or peaking plant, creating additional costs.
Below we provide the conventional LCOE estimates which ignore the differences in variable
generation and only compare the cost of energy delivered, excluding any need for auxiliary
technology to support reliability. Next we provide an alternative estimate which extends the
conventional approach by adding the costs of supporting technologies.

4.1 Conventional levelised cost of electricity estimates

Even without the relatively new complication of needing to make a distinction between variable
versus flexible generation, a major ongoing concern with conventional LCOE estimates is that,
even with a consistent set of capital cost estimates, LCOE calculations include a variety of other
data assumptions for which there is no consistent standard, calculation method or data source.
This means comparing LCOEs between studies can be meaningless. While we cannot ensure
consistency with other studies, to assist with comparison we report our key assumptions in Table
4-1. Note that we use a weighted average real cost of capital of 7% across all technologies
consistent with Australian government guidelines4.
The LCOE estimates are shown in Figure 4-1 for the years 2030 and 2050 for selected low emission
technologies. The estimates are provided as a range. The range considers the differences in capital
costs between the 2- and 4-degrees climate ambition scenarios and the high and low ranges for
the other key assumptions included in Table 4-1. Note that governments may choose to impose
additional costs on nuclear power plants to fund development and operation of centralised
facilities for storage of spent fuel – these costs have not been included. Also, solar thermal
includes 6 hours storage.

4 https://www.pmc.gov.au/sites/default/files/publications/cosst-benefit-analysis.docx

Electricity generation technology cost projections | 17

 200 200

160 160

120 120
2017 $/MWh

80 80

40 40

2030
2050
0 0
Brown coal with Black coal with Nuclear Large scale solar Wind Solar thermal
CCS CCS PV

Figure 4-1: Conventional LCOE estimates for selected technologies

Table 4-1: High and low values for key LCOE assumptions

Capacity factor (%) O&M 2050 ($/MWh) Fuel 2050 ($/MWh)

Low High Low High Low High

Brown coal with CCS 85 85 17 21 15 25

Black coal with CCS 85 85 14 17 25 49

Nuclear 85 85 12 15 11 22

Large scale solar PV 19 32 5 6 0 0

Wind 35 42 6 7 0 0

Solar thermal 40 55 12 17 0 0

18 | Electricity generation technology cost projections

4.2 Extended levelised cost of electricity estimates
The inherent problem with any attempt to take into account the cost of supporting the reliability
of variable electricity generation technologies is that the cost is context dependent. The contexts
that influence the supporting costs include:
 The type and capacity of existing flexible generation or demand management that can be
accessed
 The shape of the demand profile that needs to be met
 The share of variable renewable electricity
 The mix and diversity of variable renewable electricity
 The type of storage technology available (for example, pumped hydro requires a specific
topography that needs to be available in the region being examined)
 The supply reliability standards that must be met.
Each state in Australia has its own unique contexts now and in the future. As such, it is a significant
task to calculate an extended LCOE for variable generation which includes reliability supporting
costs for every current and likely future context.
While such a task is beyond this report’s scope, we were able to draw on previous modelling
analysis by CSIRO which focussed on scenarios where the share of variable renewable generation
was required to increase over time to 100% or slightly less by 2050 (Campey, et al., 2017b)
(Campey, et al., 2017a) (Brinsmead, et al., 2017)5. Under such circumstances, the analysis
examined each state of Australia and found that the requirement for any supporting technology
below 40% variable renewable generation share was nil. However, at this share there is a loss of
capacity factor due to transmission congestion which should be taken into account.
Above 40% renewable share, the modelling found the requirement for supporting technologies
increased at a low to moderate rate depending on each states’ quality of renewable resources and
its contribution to the national renewable share. Up to a renewables share of 60%, the modelling
tended to favour the strategy of building a combination of a small quantity of batteries and
deploying a greater capacity of diverse renewables which does result in greater “spilling” of
energy. However, the increasing losses to the capacity factor does not significantly increase the
total cost of generation because, over time, the cost of capital has also fallen to offset this effect.
As batteries also reduce in cost and the renewable share rises above 60%, the ratio of batteries to
renewable capacity increases. Also, to cover periods of extended low variable renewable output,
the modelling found that adding further batteries alone was not cost effective. Instead we needed
to replace a proportion of retiring flexible plant with gas peaking plant which has low capital costs
and is highly suited to the task of filling in infrequent gaps in supply for relatively extended
periods.

5 Results are reported across several studies but the specific modelling equations are reported in Campey et al. (2017)a

Electricity generation technology cost projections | 19

By the time we reach 90% variable renewable generation in 2050 the following points indicate the
ratios of supporting technologies that were deployed and other requirements:
 Around 8 hours battery storage or 0.75kW of battery storage per kW of variable renewable
capacity
 Around 0.4kW of dispatchable (flexible) capacity (e.g. gas open cycle) per kW variable renewable
capacity
 Synchronous condensers directly proportional to battery capacity
 A 17% and 39% relative loss in capacity factor of wind and solar photovoltaics respectively.
The recommended ratios of storage to renewables are based on cost minimisation and are hence
specific to battery storage. These ratios would change with the different costs and performance
characteristics of other forms of storage such as pumped hydro and solar thermal storage which
have been explored in other high renewable share studies. Our focus on battery storage reflected
the availability of reliable cost and performance data.
The modelling tended to favour a fairly even energy generation mix of wind and solar photovoltaic
to provide diversity, which reduces the capacity required of other supporting technologies.
However, this diversity factor makes it complicated to calculate a unique “extended LCOE for
wind” or “extended LCOE for solar photovoltaics”. Given the apparent importance of the two
technologies working together to create diversity and reducing support costs, it is more
appropriate to calculate an extended LCOE for a combined wind and solar field.
This is the approach that has been taken in Figure 4-2. It shows the extended LCOE for a combined
wind and solar photovoltaic field for a scenario where the national share of those technologies
increases to 90% by 2050. Given some states, particularly South Australia, are already at 40% or
more, some battery storage was required before the nation reached 40% on average across all
states in 2030.

20 | Electricity generation technology cost projections

 140
Open cycle gas
120 Synchronous condenser
Degrading capacity factor
Batteries
100
Combined wind and solar field

80
2016 $/MWh

60

40

20

0
11% VRE share 40% 65% 90%

2015 2030 2040 2050

Figure 4-2: Extended LCOE for combined wind and solar field for a scenario with increasing variable generation
share

The projected extended LCOE uses the combination of technologies that were estimated to be
least cost in previous CSIRO analysis. However, that analysis was not comprehensive. Due to scope
limitations previous modelling did not include pumped hydro and neither the use of or extensions
to state interconnectors. Inclusion of these alternative technologies could reduce the overall cost
of supporting the reliability of variable renewables. Future work will also need to examine the cost
of systems with high solar thermal share which have not been modelled to the same extent as
wind and solar photovoltaics.
With the caveat that more analysis will be required to narrow down the extended LCOE of high
shares of variable renewables, it is interesting to see that, once the cost of supporting
technologies is included in high renewable share electricity system (>80%), the LCOE becomes
similar to that of nuclear and CCS technologies. However, the cost disadvantage for CCS and
nuclear in the short term remains. That is, the first gigawatt of capacity for those technologies will
result in electricity generation costs of more than $120/MWh. However, the first gigawatt of
variable renewable generation, in states where the share is below 40%, results in around half that
cost as it can rely on existing flexible generation capacity.

Electricity generation technology cost projections | 21

Appendix A Projection data tables

A.1 Guide to data tables

Data tables are provided for electricity generation and battery storage capital cost projections that
are shown in Figure 3-1 to Figure 3-8. The technology data here set is slightly larger than that
presented in the main body of the report, which focussed on the most prospective options for
Australia and also excluded mature technologies whose costs are not likely to make significant
further changes. GALLM considers an even larger set of technologies, such as combined heat and
power plant and conventional geothermal, which are important outside of Australia. These are not
reported here due to their lack of relevance for Australia and corresponding difficulty in finding an
initial starting point for an Australian specific development cost.
Apx Table A.1 and Apx Table A.2 show the projections for the 4-degrees and 2-degrees climate
ambition scenarios respectively. The following points provide further information on the table
column titles:
 Rooftop solar panels includes the whole system fully installed
 CCS – carbon dioxide capture and storage
 HELE – high efficiency, low emission plant. This is based on an ultra-super critical steam plant
 BOP – balance of plant
 The battery storage cost is for the batteries only and is applicable to large or small scale
installations. However, the balance of plant is for a large scale plant co-located with a variable
renewable electricity generation field. The battery costs are based on a lithium-ion battery
chemistry.
 The biomass plant is a pulverised fuel steam plant
 Nuclear power refers to conventional reactor designs.
 Solar thermal without storage is not considered because it is less cost effective (on a cost of unit
energy output basis) than including storage.

22 | Electricity generation technology cost projections

Apx Table A.1: Projections data table for 4-degrees climate ambition scenario, 2017 Australian dollars sent-out

Black Black Black Brown Brown Brown Gas Gas Gas Biomass Biomass Large Rooftop Solar Wind Wave Enhanced Nuclear Battery Battery
coal coal coal coal coal coal combined peak with with scale solar thermal geothermal storage storage
HELE with HELE with cycle CCS CCS solar panels (6 hrs) BOP
CCS CCS PV

$/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kWh $/kWh
2017 3074 3312 7000 3510 4140 8250 1501 1025 3161 3784 8456 2100 1600 4815 1950 7157 14475 9313 306 518
2018 3059 3312 6975 3492 4140 8218 1501 1020 3155 3779 8424 1961 1491 4434 1940 7157 14615 9313 232 508
2019 3044 3312 6950 3475 4140 8187 1501 1015 3149 3771 8391 1750 1332 3678 1936 7157 14704 9313 159 498
2020 3028 3312 6926 3457 4140 8155 1501 1009 3143 3758 8359 1553 1190 3299 1930 7157 14836 9313 116 488
2021 3013 3312 6901 3440 4140 8124 1500 1004 3137 3756 8327 1486 1138 3299 1927 7157 14916 9307 105 478
2022 2998 3312 6877 3423 4140 8093 1500 999 3132 3753 8295 1425 1093 3299 1924 7157 15023 9297 103 469
2023 2983 3312 6853 3406 4140 8062 1500 994 3126 3753 8264 1368 1051 3299 1920 7157 15131 9297 101 459
2024 2968 3312 6829 3389 4140 8031 1500 989 3120 3753 8232 1319 1015 3299 1917 7157 15131 9297 98 450
2025 2953 3312 6805 3372 4140 8001 1499 984 3115 3753 8201 1276 984 3299 1891 7157 15131 9297 96 441
2026 2939 3312 6781 3355 4140 7970 1499 980 3109 3753 8170 1233 952 3299 1882 7157 15131 9296 94 433
2027 2924 3312 6757 3338 4140 7940 1499 975 3103 3753 8139 1192 921 3293 1867 7157 15131 9292 92 424
2028 2909 3312 6734 3322 4140 7910 1498 970 3098 3753 8108 1146 885 3273 1856 6078 15131 9292 90 416
2029 2895 3312 6710 3305 4140 7880 1497 965 3092 3753 8077 1095 848 3240 1809 5010 15131 9290 88 407
2030 2880 3312 6687 3288 4140 7850 1496 960 3087 3753 8046 1046 811 3124 1801 4298 15131 9288 85 399
2031 2866 3312 6664 3272 4140 7820 1495 955 3081 3753 8016 1004 779 2953 1784 3923 15131 9244 83 391
2032 2852 3312 6640 3256 4140 7791 1494 951 3076 3753 7986 975 757 2893 1772 3387 15125 9239 82 383
2033 2837 3312 6617 3239 4140 7762 1493 946 3071 3753 7956 938 728 2822 1762 3083 15117 9238 80 376
2034 2823 3312 6594 3223 4140 7732 1492 941 3065 3753 7926 898 697 2753 1756 2939 15098 9237 78 368
2035 2809 3310 6572 3207 4140 7703 1491 936 3060 3753 7896 853 664 2661 1751 2611 15073 9237 76 361
2036 2795 3310 6549 3191 4140 7674 1491 932 3054 3753 7866 834 650 2595 1745 2484 15046 9237 74 354
2037 2781 3310 6526 3175 4140 7645 1490 927 3049 3753 7837 816 636 2545 1740 2413 15044 9237 73 347
2038 2767 3310 6504 3159 4140 7617 1488 922 3044 3753 7807 800 623 2507 1736 2375 15031 9237 73 340
2039 2753 3310 6482 3143 4140 7588 1487 918 3039 3753 7778 784 611 2476 1731 2308 15009 9237 72 333
2040 2739 3310 6459 3128 3969 7560 1485 913 3033 3753 7749 770 600 2450 1722 2224 14942 9235 71 327
2041 2726 3310 6437 3112 3788 7532 1484 909 3028 3753 7720 756 589 2429 1693 2168 14272 9216 71 320
2042 2712 3310 6415 3096 3788 7504 1483 904 3023 3753 7691 744 580 2412 1690 2168 14270 9204 70 314
2043 2699 3310 6393 3081 3788 7476 1482 900 3018 3753 7662 732 571 2397 1688 2168 14270 9194 70 307
2044 2685 3310 6372 3066 3788 7448 1481 895 3013 3753 7634 721 563 2383 1680 2168 14270 9192 69 301
2045 2672 3310 6350 3050 3788 7420 1479 891 3008 3753 7606 709 553 2372 1672 2166 14270 9191 69 295
2046 2658 3310 6328 3035 3788 7393 1478 886 3002 3753 7577 699 545 2331 1663 2163 14216 9190 68 290
2047 2645 3310 6307 3020 3788 7365 1477 882 2997 3753 7549 692 540 2295 1659 2156 14141 9190 68 284
2048 2632 3310 6286 3005 3672 7338 1475 877 2992 3753 7521 684 534 2262 1654 2145 14099 9190 68 278
2049 2619 3310 6264 2990 3574 7311 1473 873 2987 3753 7493 677 528 2233 1651 2133 14051 9189 67 273
2050 2606 3310 6243 2975 3494 7284 1472 869 2982 3753 7466 668 521 2207 1648 2109 14011 9182 67 267

Electricity generation technology cost projections | 23

Apx Table A.2: Projections data table for 2-degrees climate ambition scenario, 2017 Australian dollars sent-out

Black Black Black Brown Brown Brown Gas Gas Gas Biomass Biomass Large Rooftop Solar Wind Wave Enhanced Nuclear Battery Battery
coal coal coal coal coal coal combined peak with with scale solar thermal geothermal storage storage
super HELE with super HELE with cycle CCS CCS solar panels (6 hrs) BOP
critical CCS critical CCS PV

$/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kWh $/kWh
2017 3074 3312 7000 3510 4140 8250 1501 1025 3161 3784 8456 2100 1600 4815 1950 7157 14475 9313 306 518
2018 3059 3268 6975 3492 3988 8218 1501 1020 3155 3779 8424 1952 1484 4434 1945 6297 14534 9313 232 508
2019 3044 3264 6950 3475 3967 8187 1501 1015 3149 3771 8391 1733 1321 3678 1940 5437 14678 9311 159 498
2020 3028 3249 6926 3457 3953 8155 1501 1009 3143 3758 8359 1635 1248 3299 1934 4578 14810 9311 116 488
2021 3013 3234 6901 3440 3944 8124 1500 1004 3137 3756 8327 1492 1143 3299 1925 3718 14916 9307 105 478
2022 2998 3234 6877 3423 3944 8093 1500 999 3132 3753 8295 1421 1091 3299 1921 3249 15016 9296 103 469
2023 2983 3234 6853 3406 3944 8062 1500 994 3126 3753 8264 1363 1048 3299 1909 3022 15106 9296 101 459
2024 2968 3234 6631 3389 3944 7846 1500 989 2927 3753 8042 1317 1013 3299 1901 2939 15078 9296 99 450
2025 2953 3234 6538 3372 3944 7751 1499 984 2854 3753 7944 1270 980 3299 1900 2779 15059 9296 98 441
2026 2939 3234 6385 3355 3944 7598 1499 980 2722 3753 7788 1229 949 3296 1899 2564 14661 9295 96 433
2027 2924 3234 6246 3338 3944 7460 1498 975 2603 3753 7646 1184 914 3137 1899 2431 14562 9289 94 424
2028 2909 3234 6139 3322 3944 7351 1497 970 2516 3753 7535 1148 887 3032 1899 2342 14509 9289 89 416
2029 2895 3234 6072 3305 3944 7281 1496 965 2468 3753 7462 1082 838 2905 1899 2277 14480 9282 86 407
2030 2880 3234 6033 3288 3944 7236 1496 960 2447 3753 7417 1034 802 2836 1876 2228 14455 9259 84 399
2031 2866 3234 5995 3272 3944 7192 1495 955 2427 3753 7372 990 769 2794 1853 2190 14434 9215 83 391
2032 2852 3234 5968 3256 3944 7159 1494 951 2418 3753 7338 954 741 2790 1837 2168 14430 9214 81 383
2033 2837 3234 5943 3239 3944 7128 1493 946 2411 3752 7306 913 710 2693 1825 2168 14426 9214 79 376
2034 2823 3234 5919 3223 3944 7097 1492 941 2404 3752 7275 870 678 2623 1815 2168 14387 9214 77 368
2035 2809 3234 5895 3207 3944 7067 1491 936 2398 3752 7244 851 662 2570 1796 2167 14282 9214 76 361
2036 2795 3234 5871 3191 3944 7037 1491 932 2391 3752 7213 816 637 2528 1766 2167 14227 9214 74 354
2037 2781 3234 5847 3175 3944 7007 1490 927 2384 3752 7182 798 623 2495 1755 2162 14155 9214 73 347
2038 2767 3234 5822 3159 3944 6977 1489 922 2377 3752 7151 782 610 2468 1745 2149 14118 9214 73 340
2039 2753 3234 5799 3143 3944 6947 1487 918 2371 3752 7120 766 598 2445 1736 2125 14086 9214 72 333
2040 2739 3234 5776 3128 3944 6918 1486 913 2365 3752 7091 752 587 2426 1728 2118 14003 9214 71 327
2041 2726 3234 5753 3112 3944 6889 1485 909 2359 3752 7061 738 576 2410 1711 2106 13938 9213 71 320
2042 2712 3234 5730 3096 3944 6860 1483 904 2353 3752 7032 725 567 2396 1707 2096 13938 9192 70 314
2043 2699 3234 5708 3081 3944 6832 1482 900 2347 3752 7002 715 558 2383 1704 2090 13938 9188 70 307
2044 2685 3234 5686 3066 3944 6804 1481 895 2342 3752 6974 703 550 2372 1702 2079 13937 9185 69 301
2045 2672 3234 5664 3050 3944 6776 1480 891 2336 3752 6945 692 541 2334 1700 2071 13933 9185 69 295
2046 2658 3234 5642 3035 3944 6748 1479 886 2331 3752 6916 682 533 2299 1696 2058 13926 9185 68 290
2047 2645 3234 5620 3020 3944 6720 1477 882 2326 3752 6888 674 527 2267 1691 2052 13926 9185 68 284
2048 2632 3234 5599 3005 3944 6693 1476 877 2320 3752 6860 666 521 2239 1685 2052 13926 9184 68 278
2049 2619 3234 5577 2990 3944 6665 1476 873 2315 3752 6832 658 515 2213 1679 2052 13891 9184 67 273
2050 2606 3234 5556 2975 3944 6638 1476 869 2310 3752 6804 652 510 2190 1675 2052 13890 9184 67 267

24 | Electricity generation technology cost projections

Electricity generation technology cost projections | 25
5 References

Brinsmead, T., Graham, P. & Qiu, J., 2017. Economic benefits of the Electricity Transformation Roadmap:
Technical Report, Australia: CSIRO Report for Energy Networks Australia.
Brinsmead, T. S. et al., 2015. Future Energy Storage Trends: An Assessment of the Economic Viability,
Potential Uptake and Impacts of Electrical Energy Storage on the NEM 2015-2035, s.l.: CSIRO for AEMC.
Bureau of Resource and Energy Economics (BREE), 2012. Australian Energy Technology Assessment,
Canberra: Commonwealth of Australia.
Bureau of Resource and Energy Economics (BREE), 2013. Australian Energy Technology Assessment,
Canberra: Commonwealth of Australia.
Campey, T. et al., 2017a. Low Emissions Technology Roadmap, Australia: CSIRO.
Campey, T. et al., 2017b. Low Emissions Technology Roadmap: Technical Report, Australia: CSIRO.
Chandler, W. et al., 2014. China's Future Generation: Assessing the Maximum Potential for Renewable
Power Sources in China to 2050, s.l.: WWF.
CO2CRC, 2015. Australian Power Generation Technology Report, s.l.:
http://www.co2crc.com.au/dls/Reports/LCOE_Report_final_web.pdf.
Fraunhofer ISE, 2015. Current and Future Cost of Photovoltaics. Long-term Scenarios for Market
Development, System Prices and LCOE of Utility-Scale PV Systems, s.l.: Study on behalf of Agora
Energiewende.
Government of India, 2016. A new dawn in renewable energy, India: Ministry of new and renewable energy.
Graham, P. & Reedman, L., 2015. Projecting Road Transport Revenues 2015-2050, Canberra: Report
prepared for the National Transport Commission.
Hayward, J. A. & Graham, P. W., 2017. Electricity generation technology cost projections 2017-2050 (In
Preparation), s.l.: CSIRO.
Hayward, J. & Graham, P., 2012. Australian electricity generation technology cost projections: application of
a global and local learning model, Canberra: CSIRO.
Hayward, J. & Graham, P., 2013. A global and local endogenous experience curve model for projecting
future uptake and cost of electricity generation technologies. Energy Economics, pp. 537-548.
Hayward, J., Graham, P. & Campbell, P., 2011. Projections of the futur ecosts of electricity generation
technologies: an application of CSIRO's Global And Local Learning Model (GALLM): Technical Report,
Canberra: CSIRO.
IEA, 2017. World Energy Outlook, Paris, France: OECD.
IEA, 2016. World Energy Outlook, Paris, France: OECD.
IPCC, 2014. Climate Change 2014: Mitigation of Climate Chang. Contribution of Working Group III to the
Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom
and New York, NY, USA: Cambridge University Press.
Jacoby, M., 2016. The future of low-cost solar cells. Chemical and Engineering News, 2 May, 94(18), pp. 30-
35.
Miller, D., 2017. Battery breakthroughs? Here's how yoru devices could be powered in the future. ABC
News, 22 July.

26 | Electricity generation technology cost projections

Reedman, L. & Graham, P., 2016. Transport Sector Greenhouse Gas Emissions Projections 2016-2035,
Australia: CSIRO.
Schmidt, O., Hawkes, A., Gambhir, A. & Staffell, I., 2017. The future cost of electrical energy storage based
on experience rates. Nature Energy, Volume 2, p. 17110.
World Bank, Ecofys and Vivid Economics, 2016. State and trends of carbon pricing, Washington DC: s.n.

Electricity generation technology cost projections | 27

 CONTACT US FOR FURTHER INFORMATION
t 1300 363 400 Energy
+61 3 9545 2176 Jenny Hayward
e [email protected] t +61 2 6061 6198
w www.csiro.au e [email protected]
w www.csiro.au/energy
AT CSIRO, WE DO THE Energy
EXTRAORDINARY EVERY DAY Paul Graham
We innovate for tomorrow and help t +61 2 4960 6061
improve today – for our customers, all e [email protected]
Australians and the world. w www.csiro.au/businessunit
Our innovations contribute billions of
dollars to the Australian economy
every year. As the largest patent holder
in the nation, our vast wealth of
intellectual property has led to more
than 150 spin-off companies.
With more than 5,000 experts and a
burning desire to get things done, we are
Australia’s catalyst for innovation.
CSIRO. WE IMAGINE. WE COLLABORATE.
WE INNOVATE.

28 | Electricity generation technology cost projections