WORLD

ENERGY
MODEL
DOCUMENTATION

2020 VERSION
2
Contents
1 Background ................................................................................................................................................ 5
1.1 WEO scenarios ................................................................................................................................ 6
1.2 New features in the World Energy Outlook 2020 .......................................................................... 8
1.3 World Energy Model structure ..................................................................................................... 10

2 Technical aspects and key assumptions .................................................................................................. 11

2.1 Population assumptions ............................................................................................................... 12
2.2 Macroeconomic assumptions ....................................................................................................... 13
2.3 Prices ............................................................................................................................................ 13

3 Energy demand ........................................................................................................................................ 18

3.1 Industry sector .............................................................................................................................. 19
3.2 Transport sector ........................................................................................................................... 25
3.3 Buildings sectors ........................................................................................................................... 31
3.4 Demand-side response ................................................................................................................. 34

4 Power generation and heat plants .......................................................................................................... 36

4.1 Electricity generation .................................................................................................................... 36
4.2 Value-adjusted Levelized Cost of Electricity ................................................................................. 40
4.3 Electricity transmission and distribution networks ...................................................................... 44
4.4 Hourly model ................................................................................................................................ 45
4.5 Mini- and off-grid power systems ................................................................................................ 46
4.6 Renewables, combined heat and power and distributed generation modules ........................... 47

5 Other energy transformation .................................................................................................................. 50

5.1 Oil refining and trade .................................................................................................................... 50
5.2 Coal-to-liquids, Gas-to-liquids, Coal-to-gas .................................................................................. 51
5.3 Hydrogen transformation ............................................................................................................. 52

6 Energy supply ........................................................................................................................................... 52

6.1 Oil ................................................................................................................................................. 52
6.2 Natural gas.................................................................................................................................... 57
6.3 Coal ............................................................................................................................................... 57
6.4 Bioenergy ...................................................................................................................................... 59

7 Emissions ................................................................................................................................................. 62
7.1 CO2 emissions ............................................................................................................................... 62
7.2 Non-CO2 greenhouse gases and CO2 process emissions .............................................................. 62
7.3 Air pollution .................................................................................................................................. 63
7.4 Oil and gas methane emissions model ......................................................................................... 63

8 Investment ............................................................................................................................................... 69
8.1 Investment in the energy supply chain......................................................................................... 69
8.2 Demand-side investments ............................................................................................................ 71

3
 8.3 Financing for investments ............................................................................................................ 71

9 Energy access ........................................................................................................................................... 72

9.1 Defining modern energy access.................................................................................................... 72
9.2 Outlook for modern energy access .............................................................................................. 73
9.3 Affordability of basic electricity services ...................................................................................... 74

Annex 1: WEM regional definitions .................................................................................................................... 75

A1.1 World Energy Outlook (WEO) publication .................................................................................... 75
A1.2 Oil and natural gas supply modules .............................................................................................. 77
A1.3 Coal supply module ..................................................................................................................... 78

Annex 2: References ........................................................................................................................................... 78

4
1 Background
Since 1993, the International Energy Agency (IEA) has provided medium- to long-term energy projections using
the World Energy Model (WEM). The model is a large-scale simulation model designed to replicate how energy
markets function and is the principal tool used to generate detailed sector-by-sector and region-by-region
projections for the World Energy Outlook (WEO) scenarios. Updated every year and developed over many years,
the model consists of three main modules: final energy consumption (covering residential, services, agriculture,
industry, transport and non-energy use); energy transformation including power generation and heat, refinery
and other transformation – such as Coal to Liquids or hydrogen production; and energy supply. Outputs from the
model include energy flows by fuel, investment needs and costs, CO2 emissions and end-user prices.

The WEM is a very data-intensive model covering the whole global energy system. Much of the data on energy
supply, transformation and demand, as well as energy prices is obtained from the IEA’s own databases of energy
and economic statistics (http://www.iea.org/statistics). Additional data from a wide range of external sources is
also used. These sources are indicated in the relevant sections of this document.

The WEM is constantly reviewed and updated to ensure its completeness and relevancy. The development of
the WEM benefits from expert review within the IEA and beyond and the IEA works closely with colleagues in
the modelling community, for example, by participating in the annual International Energy Workshop
(http://internationalenergyworkshop.org) and hosting the 2019 edition.
The current version of WEM covers energy developments up to 2050 in 26 regions. Depending on the specific
module of the WEM, individual countries are also modelled: 12 in demand; 101 in oil and gas supply; and 19 in
coal supply (see Annex 1). The WEM is designed to analyse:
 Global and regional energy prospects: These include trends in demand, supply availability and constraints,
international trade and energy balances by sector and by fuel in the projection horizon.
 Environmental impact of energy use: CO2 emissions from fuel combustion are derived from the projections
of energy consumption. CO2 process emissions have been estimated based on the production of industrial
materials while non-CO2 emissions originating from non-energy sectors rely on the scenarios from the IPCC
5th Assessment Report scenario database. Methane from oil and gas emissions are assessed through bottom-
up estimates and direct emissions measurements (see Methane Tracker 2020). Local pollutants are also
estimated linking WEM with the GAINS model of the International Institute for Applied Systems Analysis
(IIASA).
 Effects of COVID, policy actions and technological changes: Alternative scenarios analyse the impact of
COVID and delayed recovery, strengthened policy actions and technological developments on energy
demand, supply, trade, investments and emissions.
 Investment in the energy sector: The model evaluates investment requirements in the fuel supply chain
needed to satisfy projected energy demand in the projection horizon. It also evaluates demand-side
investment requirements, including energy efficiency, electric vehicles and industrial carbon capture and
storage.

 Modern energy access prospects: These include trends in access to electricity and clean cooking facilities. It
also evaluates additional energy demand, investments and CO 2 emissions due to increased energy access.

5
 1.1 WEO scenarios
The World Energy Outlook makes use of a scenario approach to examine future energy trends relying on the
WEM. For the World Energy Outlook 2020 (WEO-2020), detailed projections for three scenarios were modelled
and presented: the Stated Policies Scenario, the Delayed Recovery Scenario and the Sustainable
Development Scenario. The scenarios differ with respect to what will be the impact of Covid-19 on the global
economy and what is assumed about future government policies related to the energy sector. There is much
uncertainty about what governments will actually do over the coming decades, but it is highly likely that they will
continue to intervene in energy markets. Indeed, many countries have announced formal objectives; but it is
very hard to predict with any degree of certainty what policies and measures will actually be introduced or how
successful they will be. The commitments and targets will undoubtedly change in the course of the years to come.

The Stated Policies Scenario (STEPS) – our central scenario – takes into account the policies and implementing
measures affecting energy markets that had been adopted as of mid-2020, together with relevant policy
proposals, even though specific measures needed to put them into effect have yet to be fully developed (Table 1).
The Stated Policies Scenario assumes only cautious implementation of current commitments and plans. This is
done in view of the many institutional, political and economic obstacles which exist, as well as, in some cases, a
lack of detail in announced intentions and about how they will be implemented. For example, the GHG- and
energy-related components of the Nationally Determined Contributions (NDCs) pledged under the Paris
Agreement are incorporated. Where the energy policy landscape has continued to evolve since the NDCs were
announced the STEPS has been updated, becoming more ambitious in terms of GHG emissions reductions in
some countries and less ambitious in others. But we take a generally cautious view in the Stated Policies Scenario
of the extent and timing of which policy proposals will be implemented.

Table 1 Definitions and objectives of the WEO-2020 scenarios

Stated Policies Sustainable Development Delayed Recovery Scenario

Scenario Scenario

Definitions Existing policies and recently An integrated scenario specifying a A scenario where the pandemic
announced commitments and pathway aiming at: ensuring lasts longer and the economic
plans, including those yet to be universal access to affordable, recovery is consequently weaker
formally adopted, are reliable, sustainable and modern than in STEPS.
implemented in a cautious energy services by 2030 (SDG 7);
manner. substantially reducing air pollution
(SDG 3.9); and taking effective
action to combat climate change
(SDG 13).

Objectives To provide a benchmark to assess To demonstrate a plausible path to To explore the wide uncertainties
the potential achievements (and concurrently achieve universal raised by the global pandemic of
limitations) of recent energy access, set a path towards Covid-19.
developments in energy and meeting the objectives of the Paris
climate policy. Agreement on climate change and
significantly reduce air pollution.

The Delayed Recovery Scenario (DRS) is introduced this year to reflect on the uncertainties linked to the
implications of the pandemic on the global economy. In this scenario, the pandemic last longer and the economic
recovery is weaker than assumed in the STEPS. Lockdowns in various forms are much more prolonged; periodic
confinements, social distancing and other restrictive measures become part of everyday life; and the public
health crisis strains the ability of many governments to provide financial lifelines to households and companies,
exacerbating the slump. The key variable that changes in the DRS is the assumed rate of economic growth, which

6
affects a wide range of other modelling parameters, such as industrial output, freight shipments, construction
activity, and sales and ownership levels for appliances and other energy-using equipment. The initial energy
policy assumptions in the DRS are the same as those in the STEPS.

The Sustainable Development Scenario (SDS), introduced for the first time in WEO-2017, provides an energy
sector pathway that combines the fundamentals of sectoral energy policy with three closely associated but
distinct policy objectives, all of which are crucial pillars of the UN Sustainable Development Goals (SDGs). First,
it describes a pathway to the achievement of universal access to modern energy services by 2030, including not
only access to electricity but also clean cooking. Second, it paints a picture that is consistent with the direction
needed to achieve the objectives of international climate goals, including those of the Paris Agreement. Third, it
posits a large reduction in other energy-related pollutants, consistent with a dramatic improvement in global air
quality and a consequent reduction in premature deaths from household air pollution. Fourth, this year, it
integrates the stimulus packages required for a global sustainable recovery from Covid-19. Investments in the
2021-2023 period are therefore aligned with the Sustainable Recovery depicted in the World Energy Outlook
Special Report (IEA, 2020a). The objective is to lay out an integrated least-cost strategy for the achievement of
these important policy objectives, alongside energy security, in order to show how the respective objectives can
be reconciled, dealing with potentially conflicting priorities, so as to realise mutually-supportive benefits (see
Box 1).

The Net Zero Emissions by 2050 case (NZE2050) supplements the SDS analysis. The SDS sees many advanced
economies reaching net-zero emissions by 2050 at the latest, and puts the world on track for net-zero emissions
by 2070. The NZE2050 includes the first detailed IEA modelling of what would be needed over the next ten years
to put CO2 emissions on a pathway to net-zero emissions globally by 2050.

The WEM scenarios allow us to evaluate the impact of specific policies and measures on energy demand,
production, trade, investment needs, supply costs and emissions. A policies and measures database, detailing
policies addressing renewable energy, energy efficiency, and climate change, supports the analysis. This database
is available at: http://www.iea.org/policies/.

Box 1: An integrated approach to energy and sustainable development

The Sustainable Development Scenario, introduced for the first time in WEO-2017, integrates three key
objectives of the UN 2030 Agenda for Sustainable Development: universal access to modern energy services
by 2030 (embodied in SDG 7), reducing health impacts of air pollution (SDG 3.9), and action to tackle climate
change (SDG 13). As a first step, we use the WEM to assess how the energy sector would need to change to
deliver universal access to modern energy services by 2030. To analyse electricity access, we combine cost-
optimisation with new geospatial analysis that takes into account current and planned transmission lines,
population density, resource availability and fuel costs.
Second, we consider outdoor air pollution and climate goals. The policies necessary to achieve the multiple
SDGs covered in the Sustainable Development Scenario are often complementary. For example, energy
efficiency and renewable energy significantly reduce local air pollution, particularly in cities, while access to
clean cooking facilitated by liquefied petroleum gas also reduces overall greenhouse gas emissions by
reducing methane emissions from incomplete combustion of biomass as well as by reducing deforestation.
Trade-offs can also exist, for example between electric vehicles reducing local air pollution from traffic, but
at the same time increasing overall CO2 emissions if there is not a parallel effort to decarbonise the power
sector. Ultimately, the balance of potential synergies or trade-offs depends on the route chosen to achieve
the energy transition, making an integrated, whole-system approach to scenario building essential. The

7
 multiple objectives of the Sustainable Development Scenario mean that technology choices differ from
other scenarios solely driven by climate considerations. The emphasis of the Sustainable Development
Scenario is on technologies with short project lead times in the power sector in particular, such as
renewables, while the longer-term nature of climate change allows for other technology choices. Modern
uses of biomass as a decarbonisation option is also less relevant in a Sustainable Development Scenario than
in a single-objective climate scenario. This is because biomass is a combustible fuel, requiring post-
combustion control to limit air pollutant emissions and – depending on the region in question - making it
more costly than alternatives.

Since WEO-2018 the Sustainable Development Scenario also looks at the implications for the energy sector
for achieving several of the targets under United Nations Sustainable Development Goal 6 (clean water and
sanitation for all) and what policymakers need to do to hit multiple goals with an integrated and coherent
policy approach. Since WEO-2019, the time horizon is 2050, instead of 2040 as used in previous versions, in
order to reflect in our modelling the announcements made by several countries to achieve carbon neutrality
by 2050 and also allows us to model the potential for new technologies (such as hydrogen and renewable
gases) to be deployed at scale. The interpretation of the climate target embodied in the Sustainable
Development Scenario also changes over time, as a consequence of both ongoing emissions of CO 2 as well
as developments in climate science. WEO-2020 uses the IPCC’s Special Report on Global Warming of 1.5°C
(IPCC, 2018) to interpret the temperature consequences of emissions, and as a guide to emissions occurring
outside the Energy sector from Land Use, Land-Use Change and Forestry.

1.2 New features in the World Energy Outlook 2020

The following changes were made to WEM for the purposes of the WEO-2020:

Regional scope

 The United Kingdom has been removed from the EU grouping and included to Other OECD Europe group
of countries.
Final energy consumption

Behavioural analysis

 Energy future depends on political and industrial decisions but also importantly on people behaviour.
Their decisions – to purchase an electric car, to insulate a home, to cook using LPG rather than wood –
can be guided by policy and regulations, and influenced by markets, but are in many cases ultimately a
matter of personal choice. The pandemic of COVID 19 has also impacted behaviour and persistence aof
these changes are still uncertain. Some could be relatively long lasting, such as increased teleworking,
but others may prove to be short-lived. Analysis on the impact of behaviour changes on oil demand in
the Stated Policies Scenario have been performed, in particular due to the reduction of business travel,
the shift from public transport to private cars or the delay in replacing old cars while the Delayed
Recovery Scenario also takes into account behaviour changes linked to the slower economic revovery
and a persistence of the pandemic, such as a slower urbnanisation in some developing economies, larger
average home size, more packaging materials or more reliance on personal vehicles.
To better understand the role of behaviour change in scenarios compatible with the Paris Agreement
(the Sustainable Development Scenario and the Net Zero Emissions by 2050 case), a side-analysis has
been performed. The analysis looks at practical measures of everyday life habits that can be changed to

8
 reduce energy and emissions footprint. Examples of measures are wider development of teleworking
for compatible jobs, decrease of average speed of 7 kilometres per hour, reduction of average laundry
temperature by 10°C, or reconsideration of aviation travel.

Buildings module

 Historical energy demand for space heating/space cooling and historical Heating Degree Day
(HDD)/Cooling Degree Day (CDD) data is combined to normalise projections of space heating/space
cooling energy demand, removing the impact of year on year volatility in energy services needs.
Projections of space heating/space cooling energy demand remain linked to projected changes in
HDDs/CDDs by region under each scenario’s temperature pathway.

Industry module

 Two additional direct uses of hydrogen – methanol production via CO2 hydrogenation and direct
reduction of iron – are included. These two technologies are still at an early stage of development but
are promising to decarbonise the industry sector. Technical and economic parameters are aligned with
the Energy Technology Perspectives model.

Transport module
 The cost curves assumptions across all vehicle types were updated based on JRC work (Krause et al,
2017; Krause and Donati, 2018). Regional characteristics and economic factors have been taken into
account in order to expand cost curves coverage for all WEM regions.
 The representation of alternative heavy trucks has been improved. Modelling results have been
validated via using real data (EV Volumes).
 An evaluation of car market has been conducted and provided an estimation of total car sales and
electric car sales for 2020. WEM model fully captures this up-to-date analysis.

Power generation

 An analysis done on cost of capital for utility-scale solar PV projects has been performed, based on a
hybrid approach, combining data from the financial markets, academic literature and auction results
enhanced thanks to interviews with market experts.
 Update of the electricity network database: line lengths including the split between transmission and
distribution, and smart grids.

Investment and financing

 A more detailed historical estimate was made of investments associated with carbon capture, utilisation
and storage (CCUS), including for natural gas processing and in power and industry applications, and for
projects under advanced development stages.
 An estimation of the share of digital electricity grids investment by scenario was made for the first time.
 More detailed financing cost analysis was performed for shale companies, coal mining, and for
comparisons of advanced with emerging market and developing economies.
 An analysis of the ownership of clean-energy related investment in the Sustainable Development
Scenario was carried out based on estimates of the share carried out by state-owned enterprises (SOEs)
and government sponsors (public sources) compared with that led by private actors.
 A detailed analysis of the capital structure of energy investments was performed by analysing the mix
of debt and equity that developers, households and project companies use in the primary financing of
projects and assets.

9
Emissions

 Methane emissions assessment was improved. The Methane tracker analysis now integrates direct
emissions measurements in addition to the bottom-up estimates. These measures are made by
stationary monitors, ground vehicles, or aerial instruments such as satellites, drones, and planes and
have been scrutinised to only keep credible sources.

Energy access

 A new analysis was conducted on the impact of the Covid-19 pandemic on the affordability of basic
electricity services for households in Africa and Developing Asia. Using poverty data from Lakner et al.
(2020), as well as country electricity prices, we analysed the extent to which Covid-19 induced poverty
could bring about energy poverty if households become unable to afford basic electricity services. We
considered two bundles of electricity services: an essential bundle (including mobile phone charging,
four lightbulbs, and moderated use of a fan and television), and an extended bundle (including the
essential bundle plus one refrigerator, and double use of the fan and the television). The number of
people at risk of losing basic electricity services was estimated by combining data on the costs of these
bundles in different countries with data on the number of additional households pushed across different
poverty lines ($1.90/day, $3.20/day or $5.50/day) as a result of the crisis.

1.3 World Energy Model structure

The WEM is a simulation model covering energy supply, energy transformation and energy demand. The majority
of the end-use sectors use stock models to characterise the energy infrastructure. In addition, energy-related
CO2 emissions and investments related to energy developments are specified. Though the general model is built
up as a simulation model, specific costs play an important role in determining the share of technologies in
satisfying an energy service demand. In different parts of the model, Logit and Weibull functions are used to
determine the share of technologies based upon their specific costs. This includes investment costs, operating
and maintenance costs, fuel costs and in some cases costs for emitting CO2. (Figure 1)

The main exogenous assumptions concern economic growth, demographics and technological developments.
Electricity consumption and electricity prices dynamically link the final energy demand and transformation
sector. Consumption of the main oil products is modelled individually in each end-use sector and the refinery
model links the demand for individual products to the different types of oil. Demand for primary energy serves
as input for the supply modules. Complete energy balances are compiled at a regional level and the CO 2 emissions
of each region are then calculated using derived CO2 factors. The time resolution of the model is in annual steps
over the whole projection horizon. The model is each year recalibrated to the latest available data. The formal
base year is 2018, as this is the last year for which a complete picture of energy demand and production is in
place. However, we have used more recent data wherever available, and we include 2019 and 2020 estimates
for energy production and demand. Estimates are based on updates of the Global Energy Review reports which
relies on a number of sources, including the latest monthly data submissions to the IEA’s Energy Data Centre,
other statistical releases from national administrations, and recent market data from the IEA Market Report
Series that cover coal, oil, natural gas, renewables and power.

10
Figure 1: World Energy Model Overview

Note: CTL = coal-to-liquids, GTL = gas-to-liquids, CTG = coal-to-liquids.

2 Technical aspects and key assumptions

Demand side drivers, such as steel production in industry or household size in dwellings, are estimated
econometrically based on historical data and on socioeconomic drivers. All end-use sector modules base their
projections on the existing stock of energy infrastructure. This includes the number of vehicles in transport,
production capacity in industry, and floor space area in buildings. The various energy service demands are
specifically modelled, in the residential sector e.g. into space heating, water heating, cooking, lighting,
appliances, space cooling. To take into account expected changes in structure, policy or technology, a wide range
of technologies are integrated in the model that can satisfy each specific energy service. Respecting the efficiency
level of all end-use technologies gives the final energy demand for each sector and sub-sector (Figure 2).
Simulations are carried out on an annual basis. The WEM is implemented in the simulation software Vensim
(www.vensim.com), but makes use of a wider range of software tools.

11
Figure 2: General structure of demand modules

Energy service
Econometric demand Least-cost Technology/ Efficiency Final energy
Drivers analysis approach levels
(demand for fuel allocation demand
useful energy)

The same macroeconomic and demographic assumptions are used in all the scenarios, unless otherwise
specified. The projections are based on the average retail prices of each fuel used in final uses, power generation
and other transformation sectors. These end-use prices are derived from projected international prices of fossil
fuels and subsidy/tax levels.

2.1 Population assumptions

Rates of population growth for each WEM region are based on the medium-fertility variant projections contained
in the United Nations Population Division report (UNPD, 2019). In WEO-2020, world population is projected to
grow by 0.9% per year on average, from 7.7 billion in 2019 to 9.2 billion in 2040. Population growth slows over
the projection period, in line with past trends: from 1.2% per year in 2000-2019 to 0.9% in 2019-2030 (Table 2).

Estimates of the rural/urban split for each WEM region have been taken from UNPD (2019). This database
provides the percentage of population residing in urban areas by country in 5-yearly intervals over the projection
horizon. By combining this data1 with the UN population projections an estimate of the rural/urban split may be
calculated. In 2019, about 56% of the world population is estimated to be living in urban areas. This is expected
to rise to 64% by 2040.

Table 2: Population assumptions by region

Compound average annual growth rate Population (million) Urbanisation share
2000-19 2019-25 2019-40 2019 2040 2019 2040
North America 0.9% 0.7% 0.6% 493 558 82% 87%
United States 0.8% 0.6% 0.5% 329 367 82% 87%
C&S America 1.1% 0.8% 0.6% 520 592 81% 86%
Brazil 1.0% 0.6% 0.4% 211 229 87% 91%
Europe 0.3% 0.1% 0.0% 695 697 75% 81%
European Union 0.2% 0.0% -0.1% 448 437 75% 80%
Africa 2.6% 2.4% 2.2% 1 309 2 077 43% 54%
South Africa 1.4% 1.2% 0.9% 59 71 67% 76%
Middle East 2.2% 1.7% 1.3% 243 321 72% 79%
Eurasia 0.4% 0.4% 0.3% 235 248 65% 70%
Russia -0.1% -0.1% -0.2% 145 138 75% 80%
Asia Pacific 1.0% 0.7% 0.5% 4 177 4 661 49% 60%
China 0.5% 0.3% 0.1% 1 406 1 422 61% 77%
India 1.4% 0.9% 0.7% 1 366 1 593 34% 46%
Japan 0.0% -0.4% -0.5% 126 113 92% 94%
Southeast Asia 1.2% 0.9% 0.7% 661 767 49% 61%
World 1.2% 1.0% 0.8% 7 672 9 154 56% 64%

Source: IEA WEO-2020.

1 Rural/Urban percentage split is linearly interpolated between the 5-yearly intervals.

12
 2.2 Macroeconomic assumptions
Economic growth assumptions for the short to medium term are are broadly consistent with the latest assessments
from the IMF. Over the long term, growth in each WEM region is assumed to converge to an annual long-term
rate. This is dependent on demographic and productivity trends, macroeconomic conditions and the pace of
technological change.
In WEO-2020, the pandemic triggers a sharp recession in 2020, with a global gross domestic product (GDP) decline
of 4.6%. Where feasible, governments and central banks respond with large-scale fiscal stimulus programmes and
monetary expansion so as to maintain financial stability and limit negative spillovers. Social distancing measures and
eventually a vaccine enable a gradual recovery of the services sector. Over the period 2019-2040, GDP is expected to
grow on average by 3% per year in STEPS and SDS but only 2.6% in DRS (Table 3).

The way that economic growth plays through into energy demand depends heavily on the structure of any given
economy, the balance between different types of industry and services, and on policies in areas such as pricing
and energy efficiency.

Table 3: Real GDP average growth assumptions by region and scenario

STEPS/SDS DRS
2010-19 2019-25 2025-40 2019-40 2019-40
North America 2.3% 1.4% 2.0% 1.9% 1.4%
United States 2.3% 1.3% 1.9% 1.7% 1.4%
Central and South America 1.0% 1.8% 3.1% 2.7% 2.2%
Brazil 0.7% 1.2% 3.1% 2.6% 2.0%
Europe 1.9% 1.4% 1.5% 1.5% 1.1%
European Union 1.6% 1.2% 1.3% 1.3% 0.9%
Africa 3.1% 2.6% 4.4% 3.9% 3.5%
South Africa 1.5% 1.0% 2.8% 2.3% 1.9%
Middle East 2.2% 1.1% 3.1% 2.5% 2.1%
Eurasia 2.2% 1.6% 2.1% 2.0% 1.6%
Russia 1.6% 1.2% 1.6% 1.5% 1.1%
Asia Pacific 5.5% 4.2% 3.9% 4.0% 3.5%
China 7.2% 4.9% 3.6% 4.0% 3.6%
India 6.6% 4.5% 5.7% 5.4% 4.9%
Japan 1.0% 0.7% 0.9% 0.8% 0.6%
Southeast Asia 5.1% 4.2% 4.1% 4.2% 3.6%
World 3.4% 2.7% 3.1% 3.0% 2.6%
Notes: STEPS = Stated Policies Scenario SDS = Sustainable Development Scenario; DRS = Delayed Recovery Scenario. Calculated based on
GDP expressed in year-2019 US dollars in purchasing power parity terms. The same GDP assumptions are used for both STEPS and SDS.
Source: IEA WEO-2020.

2.3 Prices

2.3.1 International fossil fuel prices

International prices for coal, natural gas and oil in the WEM reflect the price levels that would be needed to
stimulate sufficient investment in supply to meet projected demand. They are one of the fundamental drivers
for determining fossil-fuel demand projections in all sectors and are derived through iterative modelling. The

13
supply modules calculate the output of coal, gas and oil that is stimulated under the given price trajectory taking
account of the costs of various supply options and the constraints on production rates. In the case that the price
is not sufficient to cover global demand, a price feedback is provided into the previous price level and the energy
demand is recalculated. The new demand arising from this iterative process is again fed back into the supply
modules until the balance between demand and supply is reached in each year of projections. The resulting fossil
fuel price trajectories appear smooth, but in reality prices are likely to be more volatile and cyclic.

Fossil fuel price paths vary across the scenarios. For example, in the Stated Policies Scenario, although policies
are adopted to reduce the use of fossil fuels, demand is still high. That leads to higher prices than in the
Sustainable Development Scenario, where the lower energy demand means that limitations on the production
of various types of resources are less significant and there is less need to produce fossil fuels from resources
higher up the supply cost curve.
The equilibrium prices for fuels have been revised down from those in the WEO-2019 because of the dampening
effect of the crisis on demand, and because of changes to strategies and cost structures on the supply side.
However, although prices are lower, the possibility of price volatility and of new price cycles has risen.

The oil price follows a smooth trajectory over the projection horizon. We do not try to anticipate any of the
fluctuations that characterise commodity markets in practice, although near-term demand for oil remains robust
in the Stated Policies Scenario.

Table 4: Fossil fuel prices by scenario

Sustainable Delayed
Stated Policies
Development Recovery

Real terms ($2019) 2010 2019 2025 2030 2035 2040 2025 2040 2025 2040

IEA crude oil ($/barrel) 91 63 71 76 81 85 57 53 59 72

Natural gas ($/MBtu)

United States 5.1 2.6 3.5 3.5 3.8 4.2 2.1 2.0 3.2 3.7
European Union 8.7 6.7 6.7 7.5 7.9 8.3 4.8 4.9 6.3 7.6
China 7.8 8.2 8.4 8.3 8.5 8.8 6.0 6.4 7.9 8.2
Japan 12.9 10.1 9.2 8.9 8.9 9.0 5.4 5.7 8.4 8.5

Steam coal ($/tonne)

United States 60 46 53 44 47 50 37 32 48 44
European Union 108 61 66 71 70 69 57 55 60 64
Japan 125 84 77 79 78 77 68 61 71 71
Coastal China 135 92 83 83 82 79 73 67 76 73
Notes: MBtu= million British thermal units. The IEA crude oil price is a weighted average import price among IEA member countries. Natural
gas prices are weighted averages expressed on a gross calorific-value basis. The US natural gas price reflects the wholesale price prevailing
on the domestic market. The European Union and China gas prices reflect a balance of pipeline and LNG imports, while the Japan gas price
is solely LNG imports; the LNG prices used are those at the customs border, prior to regasification. Steam coal prices are weighted averages
adjusted to 6 000 kilocalories per kilogramme. The US steam coal price reflects mine-mouth prices plus transport and handling cost. Coastal
China steam coal price reflects a balance of imports and domestic sales, while the European Union and Japanese steam coal price is solely
for imports.
Source: IEA WEO-2020.

2.3.2 CO2 prices

CO2 price assumptions are one of the inputs into WEM as the pricing of CO2 emissions affects demand for energy
by altering the relative costs of using different fuels. Several countries have already today introduced emissions
trading schemes in order to price carbon, while many others have schemes under development. Other countries

14
have introduced carbon taxes – taxes on fuels according to their related emissions when combusted – or are
considering to do so.

The Stated Policies Scenario takes into consideration all existing or announced carbon pricing schemes, at
national and sub-national level. In the Sustainable Development Scenario, it is assumed that CO2 pricing is
established in all advanced economies and that CO2 prices in these markets start to converge from 2025, reaching
$140/tonne CO2 in most advanced economies in 2040. In addition, several developing economies are assumed
to put in place schemes to limit CO2 emissions. All regional markets have access to offsets, which is expected to
lead to a convergence of prices (Table 5).

Table 5: CO2 prices in selected regions by scenario ($2019 per tonne)

Region Sector 2025 2040
Stated Policies
Canada Power, industry, aviation, others* 34 38
Chile Power 8 20
China Power, industry, aviation 17 35
European Union Power, industry, aviation 34 52
Korea Power, industry 34 52
South Africa Power, industry 10 24
Sustainable Development
Advanced economies Power, industry, aviation** 63 140
Selected developing economies Power, industry, aviation** 43 125
*In Canada's benchmark/backstop policies, a carbon price is applied to fuel consumed in additional sectors. **Coverage of aviation is limited
to the same regions as in the Stated Policies Scenario.
Note: Carbon prices in the DRS are close to those of the STEPS.
Source: IEA WEO-2020.

2.3.3 End-user prices

2.3.3.1 Fuel end-use prices

For each sector and WEM region, a representative price (usually a weighted average) is derived taking into
account the product mix in final consumption and differences between countries. International price
assumptions are then applied to derive average pre-tax prices for coal, oil, and gas over the projection period.
Excise taxes, value added tax rates and subsidies are taken into account in calculating average post-tax prices for
all fuels. In all cases, the excise taxes and value added tax rates on fuels are assumed to remain unchanged over
the projection period. We assume that energy-related consumption subsidies are gradually reduced over the
projection period, though at varying rates across the WEM regions and the scenarios. In the Sustainable
Development Scenario, the oil price drops in comparison to the Stated Policies Scenario due to lower demand
for oil products. In order to counteract a rebound effect in the transport sector from lower gasoline and diesel
prices, a CO2 tax is introduced in the form of an increase of fuel duty to keep end-user prices at the same level
as in the Stated Policies Scenario. All prices are expressed in US dollars per tonne of oil equivalent and assume
no change in exchange rates.

2.3.3.2 Electricity end-use prices

The model calculates electricity end-use prices as a sum of the wholesale electricity price, system operation cost,
transmission & distribution costs, supply costs, and taxes and subsidies (Figure 3). Wholesale prices are
calculated based on the costs of generation in each region, under the assumption that all plants recover their
variable costs and that new additions recover their full costs of generation, including their capital costs. System
operation costs are taken from external studies and are increased in the presence of variable renewables in line

15
with the results of these studies. Transmission and distribution tariffs are estimated based on a regulated rate of
return on assets, asset depreciation and operating costs. Supply costs are estimated from historic data, and taxes
and subsidies are also taken from the most recent historic data, with subsidy phase-out assumptions
incorporated over the Outlook period in line with the relevant assumptions for each scenario.

Figure 3: Components of electricity prices

There is no single definition of wholesale electricity prices, but in the World Energy Model the wholesale price
refers to the average price (across time segments) paid to generators for their output. They reflect the region-
specific costs of generating electricity for the marginal power plants in each time segment, plus any capital costs
that are not recovered. The key factors affecting wholesale prices are therefore:

 The capital cost of electricity generation plants;

 The operation and maintenance costs of electricity generation plants; and
 The variable fuel and, if applicable, CO2 cost of generation plants’ output.

Wholesale electricity price

The derivation of the wholesale price for any region makes two fundamental assumptions:

 Electricity prices must be high enough to cover the variable costs of all the plants operating in a region in a
given year.
 If there are new capacity additions, then prices must be high enough to cover the full costs – fixed costs as
well as variable costs – of these new entrants.

Derivation of a simplified merit order for thermal power plants

For each region, WEM breaks the annual electricity demand volume down into four segments:
 baseload demand, representing demand with a duration of more than 5944 hours per year;
 low-midload demand, representing demand with a duration of 3128 to 5944 hours per year;
 high-midload demand, representing demand with a duration of 782 to 3128 hours per year; and
 peakload demand, representing demand with a duration of less than 782 hours per year.
This results in a simplified four-segment load-duration curve for demand (Figure 4). This demand must be met
by the electricity generation capacity of each region, which consists of variable renewables – technologies like
wind and solar PV without storage whose output is driven by weather – and dispatchable plants (generation
technologies that can be made to generate at any time except in cases of technical malfunction). In order to

16
account for the effect of variable renewables on wholesale prices, the model calculates the probable contribution
of variable renewables in each segment of the simplified load-duration curve. Subtracting the contribution of
renewables from each segment in the merit order leaves a residual load-duration curve that must be met by
dispatchable generators.

Figure 4: Load-duration curve showing the four demand segments

Load

(GW)

Peak
Mid 1 Mid 2
Base

t t t t =8760
peak mid1 mid2 base

Time (sorted)

Calculation of average marginal cost in each merit order segment

Given the variable costs of all the plants in operation in each region, the WEM calculates a merit order of
dispatchable plants in each region. This ranks all the plants in order from those with the lowest variable costs to
the highest.

It then calculates which types of generator are used during each segment of the residual load-duration curve
based on the merit order; i.e. plants with the lowest variable costs are given priority, and plants with the highest
variable costs are used only in peak periods.
Once the generation from each plant has been allocated to the four segments of the merit order, the model
calculates the marginal variable cost of generation in each segment by looking at average variable cost of the
additional plants operating in each segment. For example, for the low-midload segment of the merit order, the
model excludes plants that are also operating in the baseload period and calculates the average variable cost of
the remainder. This gives a price for each merit order segment based on the average marginal variable cost of
generators operating in that segment.
Given that the model assumes that new entrants must recover their full generation costs in addition to ensuring
that all plants recover their variable costs, the model then calculates total revenues to all plants based on the
segments in which they operate and the price in each segment. For example, a baseload plant would receive the
peakload price for 782 hours of its operation, the high-midload price for 3128 - 782 = 2346 hours of its operation,
the low-midload price for 5944 - 3128 = 2816 hours and the baseload price for the rest of its operating hours. If
there are new entrants, and if the price in any segment is too low to cover their costs, then the price in those
segments is increased to the level required to justify new entry.

17
Calculation of wholesale price based on average marginal cost
Once a price has been calculated in each segment that satisfies the twin requirements of meeting all generators’
variable costs and new entrants’ full costs, the wholesale price level is then calculated as follows:
∑4𝑠=1(𝑝𝑠 ∙ 𝑑𝑠 ∙ ℎ𝑠 )
𝑊ℎ𝑜𝑙𝑒𝑠𝑎𝑙𝑒 𝑝𝑟𝑖𝑐𝑒 =
∑4𝑠=1(𝑑𝑠 ∙ ℎ𝑠 )

where s represents the four periods, ps is the price in each segment (in $/MWh) , ds is the demand level in each
segment (in MW), and hs is the number of hours in the period (in h). (Note that this results in a volume-weighted
wholesale price, rather than a time-weighted price).

2.3.4 Subsidies to fossil fuels

The IEA measures fossil fuel consumption subsidies using a price-gap approach. This compares final end-user
prices with reference prices, which correspond to the full cost of supply, or, where appropriate, the international
market price, adjusted for the costs of transportation and distribution. The estimates cover subsidies to fossil
fuels consumed by end-users and subsidies to fossil-fuel inputs to electricity generation.

The price-gap approach is designed to capture the net effect of all subsidies that reduce final prices below those
that would prevail in a competitive market. However, estimates produced using the price-gap approach do not
capture all types of interventions known to exist. They, therefore, tend to be understated as a basis for assessing
the impact of subsidies on economic efficiency and trade. Despite these limitations, the price-gap approach is a
valuable tool for estimating subsidies and for undertaking comparative analysis of subsidy levels across countries
to support policy development (Koplow, 2009).

3 Energy demand
All 26 model regions are modelled in considerable sectoral and end-use detail. Specifically:

 Industry is composed of six sub-sectors;

 Buildings energy demand is separated into six end-uses;
 Transport demand is separated into nine modes with considerable detail for road transport.
Total final energy demand is the sum of energy consumption in each final demand sector. In each sub-sector or
end-use, at least seven types of energy are shown: coal, oil, gas, electricity, heat, hydrogen and renewables. The
main oil products – liquefied petroleum gas (LPG), naphtha, gasoline, kerosene, diesel, heavy fuel oil (HFO) and
ethane – are modelled separately for each final sectors.
In most of the equations, energy demand is a function of activity variables, which again are driven by:

 Socio-economic variables: In all end-use sectors GDP and population are important drivers of sectoral
activity variables.
 End-user prices: Historical time-series data for coal, oil, gas, electricity, heat and biomass prices are
compiled based on the IEA Energy Prices & Taxes database and several external sources. Average end-
user prices are then used as a further explanatory variable ― directly or as a lag.

18
 3.1 Industry sector
The industrial sector in the WEM is split into six sub-sectors: aluminium, iron and steel, chemical and
petrochemical, cement, pulp and paper, and other industry.2 The iron and steel sub-sector is modelled together
with the sub-sectors of blast furnaces, coke ovens and own use of those two in the industry sector. However, in
accordance with the IEA energy balances, in Annex A of WEO-2020 energy demand from coke ovens and blast
furnaces is not listed under industry, but under ‘other energy sector’. Similarly, petrochemical feedstocks are
modelled as part of the chemicals and petrochemicals industry, but they are not included in industry in Annex A,
but under final energy consumption in the category ‘other’.
Energy consumption in the industry sector is driven by the demand for specific products in the energy-intensive
sectors – aluminium, iron and steel, chemicals and petrochemicals, cement, and pulp and paper – and by value
added in industry for the non-specified industry sectors (Figure 5). Production of energy-intensive goods is
econometrically projected for a specific year with the help of the following variables: population, end-use energy
prices, value added in industry, per capita consumption of the previous year and a time constant. Historic
production data is collected from a range of sources, including International Aluminium Institute (2019)
(aluminium), World Steel Association (2019) (steel), METI (2018) (ethylene, propylene and aromatics), USGS
(2018a) (ammonia), USGS (2018b) (cement), RISI (2019) and FAO (2019) (paper).

Since WEO-2019, activity projection calculations are performed with a tool shared with the Energy Technogy
Perspectives industry team, industrial production and value added are therefore perfectly consistent and aligned
with the Energy Technoloy Perspectives scenarios. This tool builds one macroeconomic drivers, such as GDP,
population, value added in industry as well as industry parameters such as historical production and production
capacities which are under construction, or planned.

Steel projections of demand and production were overhauled in preparation for the IEA Steel Roadmap 2020.
The new methodology explicitly models steel demand based on a per-capita approach reflecting saturation levels
of demand. It then derives production of steel on a per-country basis. In WEO-2020, a new steelmaking pathway
using hydrogen for direct reduced iron production is added to the technology basket. It is an alternative to use
of coal and leads to a drastic reduction of CO2 emissions as long as hydrogen is produced from a low-carbon fuel.
Based on the projected production numbers it is possible to calculate the capacity necessary to satisfy the
demand. Furthermore, we estimate current capacity and capacity vintage in each model region, which allows the
calculation of retired capacity given our assumptions on average lifetime. This allows us to determine the
required capacity additions as the sum of replacing retired capacity and meeting demand increases in a specific
year. Major energy efficiency improvements are generally limited in scope for existing industrial infrastructure.
This is reflected in our modelling by restricting the adoption of energy-efficient equipment to newly installed
capacity. However, we allow for early retirement of existing infrastructure in order to adopt more efficient
infrastructure.

2 Otherindustry is an aggregate of the following (mainly non-energy intensive) sub-sectors: non-ferrous metals, non-metallic
minerals (excluding cement), transport equipment, machinery, mining and quarrying, food and tobacco, wood and wood
products, construction, textile and leather, and non-specified.

19
Figure 5: Structure of the industry sector

Note: ETP = Energy Technology Perspectives

Final energy consumption in each sub-sector is calculated as the product of production projections and energy
intensity of the manufacturing process. While the energy consumption per unit of output is fairly stable for
existing infrastructure, the energy intensity of new capacity depends on the adoption of energy-efficient
equipment and the level of energy prices. 3
Technological efficiency opportunities are detailed by each industrial process for aluminium, iron and steel, five
major product groups in chemicals and petrochemicals, cement, pulp and paper, and cross-cutting technologies
in non-energy intensive sectors (Figure 6). Energy-efficient technologies are adopted as a function of their
payback period and their potential penetration rate, which varies by scenario. Next to single equipment
efficiency, systems optimisation and process changes represent further efficiency options integrated in the
industry sector model. Process changes take the form of an increased use of scrap metal in the aluminium
industry, increased use of scrap metal, direct reduced iron and electric arc furnaces in the iron and steel industry,
a decreased clinker-to-cement ratio in the cement industry, and an increased use of recycled paper in the pulp
and paper industry.
The data on energy-saving technologies is compiled from industry associations, individual companies and range
of pertinent literature sources.

3 For more details on modelling energy efficiency potentials in industry in WEO, see Kesicki and Yanagisawa (2014).

20
Figure 6: Major categories of technologies by end-use sub-sector in industry

Accounting for physical and technological constraints, the share of each energy source is projected on an
econometric basis relying on the previous year’s share, the fuel price change, the price change in the previous
year and a time constant. In this context, electricity is separately modelled from fossil fuels, heat, biomass and
waste because there are very limited possibilities to substitute electricity for another fuel or vice versa. However,
a potential electrification of the industry sector is taken into account via wider process changes (e.g. increasing
the share of electric arc furnaces in steel production). Fuel switches, for example from oil-based products to
natural gas, are possible and modelled via a multiple logit model. First, a utility function is defined for each fuel:
𝑝𝑟𝑖𝑐𝑒𝑖,𝑡
𝑉𝑖,𝑡 = 𝛼𝑖 ∗ + 𝛽𝑡𝑖𝑚𝑒 ∗ 𝑡 + 𝛾𝑎𝑑𝑗
𝑝𝑟𝑖𝑐𝑒𝑓𝑢𝑒𝑙 𝑎𝑣𝑒𝑟𝑎𝑔𝑒,𝑡

𝑤𝑖𝑡ℎ 𝑖 = 𝑐𝑜𝑎𝑙, 𝑜𝑖𝑙, 𝑛𝑎𝑡𝑢𝑟𝑎𝑙 𝑔𝑎𝑠, 𝑑𝑖𝑠𝑡𝑟𝑖𝑐𝑡 ℎ𝑒𝑎𝑡 𝑎𝑛𝑑 𝑏𝑖𝑜𝑚𝑎𝑠𝑠

where Vi,t is the utility function of fuel i at year t, αi is a regression coefficient for fuel i, pricei,t is the fuel price of
fuel i at year t and pricefuel average,t is the weighted average price of all fuels at time t. βtime is a time constant (in
general, this is set to zero) and γadj is an adjustment factor that represents non-price influences, such as fuel-
specific policies.

In a next step, the choice probability is determined based on the utility function of each fuel:

21
 𝑒𝑥𝑝(𝑉𝑖,𝑡 )
𝜋𝑖,𝑡 =
∑𝑖 𝑒𝑥𝑝(𝑉𝑖,𝑡 )

where πi,t is the choice probability of fuel i at time t.

The fuel share is eventually calculated taking into account the fuel share in the previous year and the choice
probability:
𝑠ℎ𝑎𝑟𝑒𝑖,𝑡 = 𝑠ℎ𝑎𝑟𝑒𝑖,𝑡−1 + 𝛿 ∗ (𝜋𝑖,𝑡 − 𝑠ℎ𝑎𝑟𝑒𝑖,𝑡−1 )

where sharei,t stands for the share of fuel i in year t, and δ is between 0 and 1 and represents the adjustment
speed.
Since WEO-2017, heat supply capacities and production costs within industry are explicit and new and
renewables technologies deployment modelling was integrated into a single framework. This has been done
together with adding one temperature dimension to the modelling, in the form five temperature levels (0-60°C,
60-100°C, 100-200°C, 200-400°C and above 400°C), defining potentials in which the different technologies can
deploy, depending on their specific costs and performances at each temperature level. Deployment of these
technologies is assessed against a counterfactual technology representing the average fossil-fuel-based
technology that would otherwise be used that given year, through Weibull functions using the average levelized
production costs of the different options and allowing for the calibration of inertia, policies and existing/lack of
infrastructure.

3.1.1 Chemicals and petrochemicals sector

The chemicals and petrochemicals sector is characterised by a variety of products that can be produced via
different pathways. Furthermore, in this sector, energy is used not only as a fuel but also as a feedstock. In the
WEM, we have separately modelled the following intermediate products, which are the most energy-intensive
ones to make:

 Organic chemicals
o Petrochemicals:
 Ethylene
 Propylene
 Aromatics (benzene, toluene and xylenes)
o Methanol
 Inorganic chemicals
o Ammonia

These product groups account for around half of total fuel consumption and for the vast majority of feedstock
consumption. Products that make up the rest of petrochemical and chemical feedstock consumption are
butadiene, butylene and carbon black. The distinction between fuel use and feedstock use is important as energy
used as feedstock cannot be reduced through efficiency measures.
In order to analyse the energy consumption for these five major intermediate products, the following principal
production routes have been implemented in the model:

 Steam cracking (for the production of ethylene, propylene and aromatics)

 Refinery streams (for the production of propylene from fluid catalytic cracking and aromatics from
catalytic reforming)
 Propane dehydrogenation (for the production of propylene)
 Methanol-to-olefins (for the production of ethylene and propylene)
 Coal/biomass gasification and natural gas steam reforming (for the production of synthesis gas)

22
  Methanol synthesis (for the production of methanol from synthesis gas)
 Ammonia synthesis (for the production of ammonia from synthesis gas)

Since the specific energy consumption in steam cracking depends on the oil product being used, inputs into this
process are divided into: ethane, liquid petroleum gas, naphtha and heavy fuel oil. As for the other sub-sectors,
future production volumes are based on an econometric projection, but in addition the projection takes account
of feedbacks from oil supply and the refinery modules to account for the availability of feedstock.

For WEO-2018, the updated WEM Industry module includes the possibility to use hydrogen as direct fuel or as
feedstock for ammonia production. Intensities and efficiencies of related processes are aligned with assumptions
used for the Energy Technology Perspectives report (IEA, 2017c). For ammonia the used hydrogen does not
appear in the balance, as it is only an intermediate product. Only the electricity consumed in the electrolysis,
synthesis and in the air-separation processes appear in the balance. For WEO-2020, this feature has been
extended to methanol production.

Chemical sector models have been updated for WEO-2018 as a result of work feeding into the Future of
Petrochemicals report. Data sources for historic production of chemical products have been updated, along with
the feedstock intensities, in both cases with the goal of better aligning with the Energy Technology Perspectives
report (IEA, 2017c). The sectoral boundaries for propylene and aromatics have been updated, moving refinery
production out of the chemicals sector energy/feedstock demand and including better representation of energy
demand for steam crackers.

3.1.2 Motor efficiency

A sub-model for industrial electric motors (illustrated in Figure 7) was is part of WEM since WEO-2016. The driver
for the model is the motor system service demand. This is driven by the projected value-added in the industry
sector and can be thought of as the demand for motor system service. The model then has several steps to
control efficiencies of three separate modules (being the end-use, the motor and a variable speed drive [VSD]).
The efficiencies of these three steps multiplied with each other give the motor system efficiency, Eff0. Dividing
the motor system service by Eff0 gives the electricity consumed in industrial electric motor systems.

Figure 7: Structure of the electric motor model

Motor Motor Motor

electricity service system
service
Electricity Motor End-use
VSD
Eff1 Eff2 Eff3 =
Eff4*Eff5...

Motor system efficiency, Eff0 = Eff1 * Eff2 * Eff3

The end-use efficiency is calculated as a weighted average of efficiencies in four end-uses: pumps, fans,
mechanical movement and compressors. For each of these four end-uses there is a stock model (with starting
year 1980) with efficiencies of each vintage of normal and “efficient” devices. The share of sales of efficient
devices is determined by the payback periods of such an investment, controlled by a Gompertz function. The
end-use module includes the efficiency of a throttle, which represents the fact that in many motor systems, the
outflow is controlled by with a throttle or a damper, thereby bringing the system efficiency down.

23
3.1.3 Material efficiency

Beyond energy efficiency technologies and measures, the WEM industry model can also represent material
efficiency strategies, enabling further energy savings. Given the large share of energy costs in production costs
in energy-intensive sectors, the potentials to increase energy efficiency are in general more limited compared
with less energy-intensive industries. Possible material efficiency strategies to limit the growth of these sectors’
energy consumption are:

 Re-use materials: Use of post-consumer scrap directly (i.e. without re-melting) for the same or other
applications.
 More efficient production: Reduce the losses in the product process by increasing manufacturing and
semi-manufacturing yield rates.
 Light-weighting products: Produce the same product with a lower average mass per product.
 Increase recycling: Increase collection rates of post-consumer scrap.
 Divert fabrication scrap: Instead of remelting fabrication scrap, it can be used for other applications.
 More intensive use of products: Use material-intensive goods more intensively, e.g. by sharing a car or
using a building for a higher share of the day.
 Longer life times: Extend the life time of material components in products.

For this purpose, material flow models for aluminium and steel demand enable realistic assumptions for these
material efficiency strategies to be incorporated into the model, helping the assessment of future demand and
the amount of scrap metal (semi-manufacturing, manufacturing and post-consumer scrap) available for metals
production. This modelling work builds mainly on the literature and previous IEA publications relating to material
efficiency (IEA 2019a). The scope of this approach is limited to materials and energy demand within the
respective industry sectors. It does not analyse the implications on energy consumption upstream, in mining or
the transportation of materials, nor the consequences for downstream energy consumption, e.g. from more
efficient, lighter cars. Nor does the study analyse the potential for energy savings from substituting materials,
e.g. using plastics for metals. The coverage of material efficiency was enhanced for key sectors aluminium,
cement and steel in the Sustainable Development Scenario of WEO-2019.

24
 3.2 Transport sector
The transportation module of the WEM consists of several sub-models covering road transport, aviation, rail and
shipping navigation (Figure 8). The WEM fully incorporates a detailed bottom-up approach for the transport
sector in all WEM regions.

Figure 8: Structure of the transport sector

Sub-sectors
Activity variables
End-use Road transport
energy prices
Passenger-
kilometres
Historical Aviation
trends

Rail
GDP

Tonne-kilometres
Navigation
Population

Other

Note: ‘Other’ includes pipeline and non-specified transport.

For each region, activity levels such as passenger-kilometres and tonne-kilometres are estimated econometrically
for each mode of transport as a function of population, GDP and end-user price. Transport activity is linked to
price through elasticity of fuel cost per kilometre, which is estimated for all modes except passenger buses and
trains and inland navigation. This elasticity variable accounts for the “rebound” effect of increased car use that
follows improved fuel efficiency. Energy intensity is projected by transport mode, taking into account changes in
energy efficiency and fuel prices. The road module is calibrated to historical fuel use, i.e. the gasoline, diesel,
natural gas and electricity, which is updated every year and ensure that the model reflects closely on recent
developments in terms of vehicle stocks, vehicle mileages and vehicle efficiencies. A gap factor is used to account
for differences between test cycle fuel consumption and on-road fuel use.

3.2.1 Road transport

Road transport energy demand is broken down among passenger light duty vehicles (PLDVs), light commercial
vehicles (LCVs), buses, medium trucks, heavy trucks and two- and three-wheelers. The model allows fuel
substitution and alternative powertrains across all sub-sectors of road transport. The gap between test and on-
road fuel efficiency, i.e. the difference between test cycle and real-life conditions, is also estimated and
projected.

As the largest share of energy demand in transport comes from oil use for road transport, the WEM contains
technology-detailed sub-models of the total vehicle stock and the passenger car fleet. In its origin, the stock
projection model is based on an S-shaped Gompertz function, proposed in Dargay et al. (2006). This model gives
the vehicle ownership based on income (derived from GDP assumptions) and 2 variables: the saturation level

25
(assumed to be the maximum vehicle ownership of a country/region) and the speed at which the saturation level
is reached. The equation used is:
𝑏𝐺𝐷𝑃𝑡
𝑉𝑡 = 𝑦 𝑒 𝑎𝑒
where V is the vehicle ownership (expressed as number of vehicles per 1,000 people), y is the saturation level
(expressed as number of vehicles per 1,000 people), a and b are negative parameters defining the shape of the
function (i.e. the speed of reaching saturation). The saturation level is based on several country/region specific
factors such as population density, urbanisation and infrastructure development. Passenger car ownership is
then calculated based on the detailed vehicle fleet data in the IEA Mobility Model (MoMo) plus other regional
statistics. Using the equation above, changes in passenger car ownership over time are modelled, based on the
average current global passenger car ownership. Both total vehicle stock and passenger vehicle stock projections
are then derived based on our population assumptions. Projected vehicle stocks and corresponding vehicle sales
are then benchmarked against actual annual vehicle sales and projected road infrastructure developments. The
resulting vehicle stock projections can therefore differ from those that would be derived by the use of the
Gompertz function alone.

The analysis of passenger light-duty vehicle (PLDV) uses a cost tool that guides the choice of drivetrain
technologies and fuels as a result of their cost-competitiveness. The tool acts on new passenger-LDV sales as
depicted in Figure 9, and determines the share of each individual technology in new passenger LDVs sold in any
given year.

Figure 9: The role of passenger-LDV cost model

The purpose of the cost tool is to guide the analysis of long-term technology choices using their cost-
competitiveness as one important criterion. The tool uses a logit function for estimating future drivetrain choices
in passenger LDV.4 The share of each PLDV type j allocated to the passenger light duty vehicle market is given by
𝑝𝑟
𝑏𝑗 𝑃𝑃𝐿𝐷𝑉
𝑗
𝑆ℎ𝑎𝑟𝑒𝑗 = 𝑟𝑝
∑𝑗 (𝑏𝑗 𝑃𝑃𝐿𝐷𝑉 )
𝑗

where

4Originally developed to describe the growth of populations and autocatalytic chemical reactions, logit functions can be
applied to analyse the stock turnover in different sectors of the energy system. Here, it uses the cost-competitiveness of
technology options as an indicator for the pace of growth.

26
  PPLDVj is the annual cost of a vehicle, including annualised investment and operation and maintenance
costs as well as fuel use
 rp is the cost exponent that determines the rate at which a PLDV will enter the market
 bj is the base year share or weight of PLDVj

The cost database in the cost tool builds on an analysis of the current and future technology costs of different
drivetrains and fuel options, comprising the following technology options:

 conventional internal combustion engine (ICE) vehicles (spark and compression ignition)
 hybrid vehicles (spark and compression ignition)
 plug-in hybrids (spark and compression ignition)
 electric cars with different drive ranges
 hydrogen fuel cell vehicles

The model takes into account the costs of short- and long-term efficiency improvements in personal transport
distinguishing numerous options for engine (e.g. reduced engine friction, the starter/alternator, or transmission
improvements) and non-engine measures (e.g. tyres, aerodynamics, downsizing, light-weighting or lighting). In
addition, it uses projections for the costs of key technologies such as batteries (NiMH and Li-ion) and fuel cells.
The pace of technology cost reductions is then calculated using learning curves at technology-specific learning
rates.

The cost analysis builds on a comprehensive and detailed review of technology options for reducing fuel
consumption. The database was reviewed by a panel of selected peer-reviewers, and feeds into the cost tool.
The cost database is constantly reviewed and takes account of recent research. This year, the cost curves
assumptions across all vehicle types were updated based on JRC work (Krause et al, 2017; Krause and Donati,
2018). Regional characteristics and economic factors have been taken into account in order to expand cost curves
coverage for all WEM regions.

For the purpose of electric cars projections, a thorough review on electromobility targets for the top 20 global
automakers has been conducted for WEO-2019. This analysis permits us to assess if the automakers
commitments for launching new electrified car models are falling behind the necessary EVs rollout for meeting
fuel economy goals and Zero Emission Vehicles mandates.

Regarding hydrogen vehicles projections, they take into account the recent car market developments, policy
announcements and the key outcomes from IEA’s hydrogen report (IEA, 2019c).
Road freight transport vehicles can be broadly classified into light-commercial vehicles (<3.5t), trucks (3.5t – 16t)
and heavy trucks (>16t). For the latter two categories, WEM comprises two detailed sub-models to guide the
development of average fuel economy improvements on the one hand, and technology choices on the other
hand. For the former, the model endogenises the decision of investments in energy efficiency by taking the view
of rationale economic agents on the basis that minimising costs is a key criterion for any investment decision in
this sector. Using the efficiency cost curves of NRC, the model calculates the undiscounted payback period of an
investment into more fuel-efficient trucks and heavy trucks. The model then allows for investments where the
calculated payback period is shorter than an assumed minimum payback period that is required by fleet
operators (generally assumed between 1 and 3 years, depending on the region). The problem is solved in an
iterative manner as the model seeks to deploy the next efficiency step on the efficiency cost curve as determined
by literature, but may use efficiency improvement levels in between individual steps on the efficiency cost curve.
An example of an efficiency cost curve is depicted in Figure 10.

27
Figure 10: Illustration of an Efficiency Cost Curve for Road Freight

USD/vehicle Short-term
Long-term

Fuel consumption improvement rate (litres/100km)

In a second step, the model simulates the cost effectiveness of a conventional internal combustion engine vehicle
against other competing options such as hybrid vehicles and natural gas vehicles. The simulation is guided by the
use of a Weibull function. Alternative powertrains for medium- and heavy-duty trucks have been implemented
in the WEM: fuel cell, battery electric and plug-in hybrid electric.

In order to assess the problems created due to chicken-and-egg-type of situations when it comes to the
deployment of those alternative fuels in transport that require a dedicated refuelling infrastructure, and to better
reflect potential spill-over effects of the use of such alternative fuels in other sectors of the energy system, the
WEM has two dedicated sub-models, one covering natural gas infrastructure and the other electricity-related
refuelling infrastructure. In principle, both modules seek to quantify the costs and benefits of increased
infrastructure availability for transmission and distribution of these alternative fuels. Thus, in the case of natural
gas, an enhanced share of natural gas use in primary energy demand (for example in the power generation
sector) would lead to the development of a transmission grid in the economy; and similarly, an increased share
of natural gas in final energy consumption by end use sectors (for example, in industry or buildings) will lead to
an expanded distribution grid close to the consumption centres, thereby impacting the overall availability of
natural gas in the economy and simultaneously driving down the transmission and distribution costs for all
consumers, including the transportation sector. Moreover, an increased share of natural gas vehicles (NGVs) in
total vehicle sales in the region would gradually improve the development and utilization of a refuelling network
due to increased density of vehicles served per station, thereby reducing the average cost of refuelling. This
relationship is thus implemented as a positively reinforcing loop, wherein increased penetration of natural gas
(in all other sectors, not just transport) and natural gas vehicles helps driving down the overall refuelling
infrastructure costs. In essence, the relationship of these spill-over benefits can be illustrated as in Figure 11.
For the case of electric vehicles, availability of transmission and distribution grid is less of an issue, especially in
OECD countries, thanks to the already existing widespread use of electricity in different end use sectors
(especially buildings). However, the availability of electric recharging infrastructure is one of the important
constraints in this case, and hence it is important to determine how a reduction in refuelling costs could influence
the possibility for oil substitution in road transport. Therefore, the electric vehicle (EV) sub-module assesses the
cascading effect of an increased share of electric vehicles in overall vehicle sales on bringing down the refuelling
costs. Detailed cost curves were prepared outlining the reduction of refuelling costs with the increase in overall
vehicle stock of electric vehicles. These cost curves were provided as an exogenous input to the model, so as to
continuously adjust the refuelling costs as the share of EV sales rises in the future.

28
Finally, based on projections of the average fuel consumption of new vehicles by vehicle type, the road transport
model calculates average sales and stock consumption levels (on-road and testcycle) and average emission levels
(in grammes of CO2 per kilometre) over the projection period. It further determines incremental investment costs
relative to other scenarios and calculates implicit CO 2 prices that guide optimal allocation of abatement in
transport.

Figure 11: Refuelling infrastructure cost curve (illustrative)

Refuelling Cost = F (NGV share in Vehicle Sales)

Distribution Cost = F (gas share in final consumption)
Transmission Cost = F (gas share in primary energy)
Infrastructure costs

Driving factor for spill over benefits

3.2.2 Aviation
Aviation is among the fastest growing transport sectors. The aviation model is updated in collaboration with the
IEA’s Energy Technology Perspectives team which maintains the Mobility Model (MoMo). The model aims at
assessing air traffic measured in revenue passenger kilometres (RPKs) and for passenger travel and revenue
tonne-kilometres (RTK) for cargo. RPKs refer to the number of passengers which generate revenue multiplied by
the kilometres they fly. RTKs refer to the number of tonnes carried which generate revenue multiplied by the
kilometres they are flown. A detailed review of publically available historical data for the aviation sector was
conducted to update the historical database of WEM and MoMo. It includes data from the International Civil
Aviation Organization, aircraft manufacturers such as Airbus, Boeing and Embraer, and the Japan Aircraft
Development Corporation. As a result, RPKs were estimated at over 8 trillion in 2018, while RTKs were at around
1 trillion (ICAO, 2019).

Future RPK and RTK growth is guided by projections of various factors, using growth in per capita income by
region as a main driver:
 the number of flights per year and capita by model region, which grows as a function of population and
income growth ;
 the average flight distances by model region, which gradually declines on a global level in line with
recent trends (although with differences by model region); and
 the average flight occupancy, which is assumed to remain constant at current levels.

To assess future fuel consumption as a result of RPK and RTK growth, the model projects the resulting fuel
intensity that global fuel consumption growth complies with an annual average fuel efficiency improvement of
RPK growth of 2% per year from 2020 to 2050, as expressed by the Assembly of the International Civil Aviation
Organization (ICAO). A further sub-model calculates investment costs and marginal abatement costs split by the
types of abatement measure.

29
3.2.3 Maritime
In collaboration with the IEA’s Energy Technology Perspectives team, which maintains the IEA Mobility Model, a
specific approach is implemented in the WEM since WEO-2016. The aim of this overhaul was to have a better
understanding of maritime freight demand from a bottom-up perspective, which is driven by projections of
maritime trade. In the previous approach, we regarded energy demand for international maritime transport from
a top-down perspective driven by growth in GDP PPP. The new bottom-up structure is based on the ASIF (Activity,
Structure, Intensity and Fuel use) framework (Schipper, 2010) to assess energy demand and CO 2 emissions by
region and ship type.

The activity variable represents the maritime trade demand in tonne-kilometre, i.e. tonnes carried multiplied by
number of kilometres they are shipped. It covers global physical flows of maritime trade of 19 commodity types,
by origin-destination points between 26 regions. Physical and monetary trade numbers and projections (2010-
2050) were derived from the International Transport Forum freight model (Martinez et al., 2014) and revised to
reflect changes in value to weight ratios of energy products. The data were aggregated to account for five ship
types (oil tankers, bulk carriers, general cargo, container ships and others). The structure variable is interpreted
as the load factor, i.e. the average capacity utilization per ship per trip, which allows deriving the vehicle-
kilometres projected for each region and for each ship type. Load factor projections are based on historically
observed growth rates of the average size of the different ship types, which are published by UNCTAD. The
capacity utilization factor is kept constant. Furthermore, the base year energy intensity values are derived from
the IMO 3rd GHG study (IMO, 2014). In the Stated Policies Scenario, projections of the energy intensity variable
take into account the effect of Energy Efficiency Design Index (EEDI), introduced by the International Maritime
Organisation (IMO). The EEDI mandates a minimum 10% improvement in the energy efficiency per tonne-km of
new ship designs from 2015, 20% from 2020 and 30% from 2025 to 2030. These improvements are benchmarked
against the average efficiency of ships built between 1999 and 2009. In the Sustainable Development Scenario,
energy efficiency improvements are assumed to converge towards the maximum efficiency improvement
potentials, which were assessed for each type of ship. Lastly, combining the activity, structure and energy
intensity variable determines the final energy consumption by region and by ship type. Multiplying this number
with the CO2 emission factors of the different fuels modelled (heavy fuel oil, marine diesel oil, LNG and biodiesel)
gives the total CO2 emissions.

3.2.4 Behaviour change analysis

Severaly analysis regarding behaviour change in transport have been carried out:

 Ex-post analysis for the impact of behaviour change on aviation sector has been developed. Historical
data (OAG, AIM from UCL) has been used to disaggregate aviation activity per person and distance.
Changes in occupancy factors have been assumed to assess the impact of behaviour change in oil
demand.
 In WEO 2019, an in-depth analysis was performed on the rise of SUVs at a country level. Both
commercial (IHS Markit, Jato Dynamics) and in-house (GFEI, MoMo Database) dataset have been used.
These analysis have been updated and extended to 2019, using the Marklines database. The moderate
growth of SUVs is anticipated in the STEPS, while a steeper rise in SUVs is assumed in the DRS.
 The car market in 2020 was analysed using multiple sources (Marklines, EV Volumes etc.), estimating a
drop in car sales in 2020. We estimated how much of this drop comes from delayed new purchases and
how much from delayed replacement, based on WEM results. Econometric functions have been applied
to project the future trend, assuming that the car market will return to normal by 2030. The impact was
captured in both the STEPS and the DRS.

30
  Recent data show a shift from public transport to private vehicles due to health concerns. Publicly
available reports (i.e. survey by Ipsos) were used to estimate the mobility needs that have to be covered
either by bicycles or private cars. Different assumptions have been made for different WEM regions,
depending on the accessibility to bikes (i.e. low accessibility in the United States, high accessibility in the
Netherlands), and the impact on oil demand due to this modal shift was estimated.
 Regarding the analysis on the impact of teleworking, a literature review on the distance of a commute
by mode for key WEM regions has been carried out. These data have been expanded to all regions and
the oil demand for each mode have been estimated. After assuming the maximum potential of
teleworking of the workforce (i.e. 20% by 2030), the impact of teleworking on oil demand was assessed.

3.3 Buildings sectors

The buildings sector module of the WEM is subdivided into the residential and services sectors, both having a
similar structure (Figure 12). Population, GDP and dwelling occupancy drive the activity variables, such as floor
space, appliance ownership, number of households (residential sector) and value added (in the services sector).

Figure 12: Structure of the buildings sector

Sub-sectors

Space heating

Activity variables
Space cooling

Urbanisation Floor space

Water heating

Population Number of
households Cooking

Dwelling
occupancy Appliances Lighting
ownership

GDP Appliances
Services
Refrigeration
value added
Cleaning

Brown goods

Other

In the residential and services sectors, energy demand is further subdivided into six standard end uses in
buildings, namely space and water heating, appliances (divided into four different categories: refrigeration –
fridge and freezer; cleaning – washing, drying machines and dish washers; brown goods – TVs and computers;
and other appliances), lighting, cooking and space cooling. These sub-modules project final energy consumption
from the base year over the projection horizon in three steps.
In a first step, the demand for an energy service, i.e. the useful energy demand, is determined, based on the
activity variables.

𝐸𝑛𝑑 𝑢𝑠𝑒 𝑠𝑒𝑟𝑣𝑖𝑐𝑒 𝑑𝑒𝑚𝑎𝑛𝑑 = 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 ∗ 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦

31
Here, activity refers to the main driver of the energy service demand – for the residential sector it is floor space
area, people per household, and appliances ownership; and for services, it is valued added by the service sector.
Intensity refers to the amount of energy service (e.g. space heating) needed per unit of activity variable (e.g.
floor space). The activity variables are projected econometrically, based on historical data and linking to socio-
economic drivers such as GDP and population. For each end use, the intensity variable is projected using the
historical intensity and adjusting, for each projection year, to the change in average end-user fuel prices (using
price elasticity) and change in average per capita income (using income elasticity). In the specific case of space
heating and cooling, the intensity projections are also adjusted for historical variations in temperature. The
impact of climate change on space heating and cooling demand can be included as well. Based on the anticipated
change in heating and cooling degree days due to climate change, the increase in heating and cooling demand is
quantified (McNeil and Letschert, 2007). Existing policies related to retrofit and insulation on existing buildings
are also taken into account, as a higher level of retrofit and insulation results in a lower level of energy service
demand.

Multiplying the activity by the intensity gives the end-use service demand (useful energy consumed). Thus, the
incremental energy service demand from year to year could be an outcome of increased demand for service
(largely in the case of developing economies, where demand is still evolving), or the retirement of old units
according to equipment retirement rates (as is the case in a majority of advanced economies). Both result in a
need for new equipment.

In a second step, the technologies to supply the end-use service demand are chosen. For each end use, there is
a detailed set of technologies available to the model (Figure 13). Within each technology option, for example a
gas boiler, there are several types, representing the varying levels of efficiency and the associated investment
cost. Additionally, there is a possibility to switch technologies, whereby heat-pumps could be used for space
heating, instead of gas boilers. Within the residential sector, additional detail regarding bioenergy allows for
more accurate modelling of the historical and projected use of biogas digesters to meet home energy needs, as
well as the use of bioethanol and other liquids in cooking stoves and household heating equipment.

Figure 13: Major categories of technologies by end-use subsector in buildings

Insulation and Space and Lighting

Appliances Cooling Cooking
Energy water heating
Management
Incandescent Fossil fuel-
Varying Air based
Fossil fuel-based
Insulation levels of conditioner (LPG/gas/coal)
(coal/oil/gas)
(walls, double efficiency
glazed windows, - Conventional Fluorescent
Electric
floor, roof) - Condensing Heat pump

Fluorescent Solid
Electric boiler Bioenergy
Retrofit rapid start
Heat radiator
LED Biofuels
Active control Heat pump (air
or ground
source)
Biogas

Renewables

The technology choice is made based on relative costs, efficiencies of the technologies and policy constraints, if
any. The share of technologies is allocated by a Weibull function based on their specific costs per unit of service

32
demand supplied, which includes investment costs, operating and maintenance costs, and fuel costs. The routine
allocates the different technologies to satisfy the new service demand for every year over the Outlook period.
This allocation is subject to upper and lower boundaries, reflecting real-world constraints such as technology
availability and adoption, policies, and market barriers. Other measures to reduce end-use service demand, such
as insulation and active control, are included in the technology allocation routine: they are deployed where they
are economically viable.

To assess equipment and appliance efficiency, and related costs, we consulted with a large number of companies,
experts and research institutions at the national and international levels. We also conducted an extensive
literature review to catalogue technologies that are now used in different parts of the world and to judge their
probable evolution (Anandarajah, et al., 2011; Econoler, et al., 2011; IEA, 2010; IEA, 2011; IEA, 2012b; Kannan,
et al., 2007; Waide, 2011; IEA, 2013a; IEA2014b). The efficiency potential for electrical appliances has been
determined using the BUENAS (Bottom-Up Energy Analysis System) model, an international appliance policy
model developed by Lawrence Berkeley National Laboratory (LBNL). BUENAS covers thirteen economies that
together account for 77% of global energy consumption, and twelve different end-uses, including air
conditioning, lighting, refrigerators and industrial motors (LBNL, 2012). The assessment of efficiency potential in
the services sector buildings also benefitted from preliminary estimates available from GBPN (Global Buildings
Performance Network) and CEU (Central European University) study on buildings (GBPN and CEU, 2012).

In a third step, total final energy consumption in the residential and service sector is obtained based on the
efficiencies of existing and new building equipment. Efficiency represents the amount of energy needed to meet
a unit of service demand, and thus represents the technical performance of the equipment or appliances. Final
energy consumption in the buildings sector is a summation of the sub-sectoral energy consumed by the total
technology stock, which includes the historical (declining) stock of appliances and equipment, and the new
technologies added every year over the Outlook period by the technology allocation routine.
1
𝐹𝑖𝑛𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦 𝑐𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 = ∗ 𝐸𝑛𝑑 𝑢𝑠𝑒 𝑠𝑒𝑟𝑣𝑖𝑐𝑒 𝑑𝑒𝑚𝑎𝑛𝑑
𝜂
At the same time, investments in all technology additions, as well as insulation, retrofit, and automation and
control measures, are calculated. Carbon emissions related to the buildings sector are also calculated.

The buildings module is directly linked to the access (electricity access and clean cooking access) module to take
into account the growth of electricity and of alternative fuels or stoves for cooking (see Section 9).

Regarding behavioural changes in the residential sector such as lower indoor air remperature settings, lower us
of air conditioning, use of line-drying and cool washing, a literature review was carried out to assess the impact
on energy consumption. Assumptions regarding the potential allowed us to assess the total impact and the
resulting decrease in CO2 emissions.

Regarding the analysis on the impact of teleworking, a literature review on the impact of working for home on
the increase in residential consumption for key WEM regions has been carried out. These data have been
expanded to all regions and the total increase for each fuel has been estimated. After assuming the maximum
potential of teleworking of the workforce (i.e. 20% by 2030), the impact of teleworking on residential
consumption was assessed.

For DRS, different assumptions regarding behavioural change were made to assess the level of energy services.
Weaker consumer purchain power lead to lower activity levels in commerciall buildings. Lower household
incomes lead to a reduced demand for new buildings, as well as lower ownership levels and slower turnover of
appliances.

33
 3.4 Demand-side response
While demand-side integration measures such as energy efficiency and electrification are long-standing
components of the WEM, to assess demand-side response a new tool has been developed, since measurement
of demand-side response requires a higher temporal resolution. To assess the potential amount of flexibility in
end-use electricity demand that might be used to facilitate higher penetration of variable renewables, a three-
step methodology was used. The first was to assess temporarily the load profile for each sector and subsector or
end-use (residential and services (e.g. space heating, water heating.), industry (e.g. steel, chemicals industry),
transport (e.g. road and rail) and agriculture) for every 24 hours of 36 typical days (weekday, Saturday and Sunday
of each month). The aggregate electricity demand of each end-use or subsector temporally was matched to the
total load profile of a given country.5 An example of the load aggregation is displayed in Figure 14: .

The second step was to assess the share of demand that is flexible in each end-use. This share is the product of
three flexibility factors, sheddability, controllability and acceptability (Ookie Ma, 2013):
 Sheddability: Share of the load of each end-use that can be shed, shifted or increased by a typical DSR
strategy.

 Controllability: Share of the load of each end-use which is associated with equipment that has the
necessary communications and controls in place to trigger and achieve load sheds/shifts.

 Acceptability: Share of the load for a given end-use which is associated with equipment or services
where the user is willing to accept the reduced level of service in a demand-response event in exchange
for financial incentives.

This framework enables scenarios to consider demand flexibility from various technologies and at varying levels
of social acceptability.

The third step was to integrate the DSR profiles in the hourly model to determine the load that can be shifted,
given market conditions in the region analysed.

5Data from ENTSO-E, PJM, ERCOT, MISO, NEISO, NYISO were used to replicate respectively the overall load curves of
European Union, United States and India.

34
Figure 14: Illustrative load curves by sector for a weekday in February in the European Union compared
to the observed load curve by ENTSO-E for 2014
Industry Agriculture and transport
GW

0h 8h 16h 24h 0h 8h 16h 24h

Iron and steel Chemicals Cement Agriculture Road
Paper Aluminium Other industry Rail Other transport

Residential Services
GW

0h 8h 16h 24h 0h 8h 16h 24h

Water heating Other Appliances Refrigeration Cleaning Brown goods
Cooking Lighting Space heating Cooling
Total
GW

0h 4h 8h 12h 16h 20h 24h

Industry Agriculture Transport Services Residential ENTSO-E

The load profile is an aggregation of the sectoral load profiles

for typical days of the year

Note: ENTSO-E represents the aggregated load curve for the 28 European Union countries. Sources: (ENTSO-E, 2016); IEA analysis.

35
4 Power generation and heat plants
Based on electricity demand, which is computed in all demand sectors (described in section 3), the power
generation module calculates the following:

 Amount of new generating capacity needed to meet demand growth and cover retirements and
maintain security of supply.
 Type of new plants to be built by technology.
 Amount of electricity generated by each type of plant to meet electricity demand, cover transmission
and distribution losses and own use.
 Fuel consumption of the power generation sector.
 Transmission and distribution network infrastructure needed to meet new demand and replace retiring
network assets.
 Wholesale and end-use electricity prices.
 Investment associated with new generation assets and network infrastructure.

4.1 Electricity generation

The structure of the power generation module is outlined in Figure 15. The purpose of the module is to ensure
that enough electrical energy is generated to meet the annual volume of demand in each region, and that there
is enough generating capacity in each region to meet the peak electrical demand, while ensuring security of
supply to cover unforeseen outages.

Figure 15: Structure of the power generation module

Electricity demand
+ losses + own use

Load curve
Plant Capacity Factors Investment assumptions

Retirements of existing Additions of new plants by

plants technology

Merit order
Existing capacity by plant Fuel and CO2 prices

Refurbishments/ Retrofitting
Efficiencies
/ Mothballing of existing
plants

End-user prices Wholesale price Generation by plant Fuel consumption by plant

CO2 emissions by plant type

Investment needs
Key results

The model begins with existing capacity in each region, which is based on a database of all world power plants.
The technical lifetimes of power plants are assumed to range between 45 and 60 years for existing fossil-fuel
plants and nuclear plants (unless otherwise specified by government policies). The lifetimes of wind and solar PV

36
installations are assumed to have a distribution centred around 25 years, ranging from 20 to 30 years;
hydropower projects 50 years; and bioenergy power plants 25 years.

4.1.1 Capacity additions

The model determines how much new generation capacity is required annually in each region by considering the
change in peak demand compared to the previous year, retirements of generation capacity during the year, and
any increase in renewable capacity built as the result of government policy. Installed generating capacity must
exceed peak demand by a security-of-supply margin; if this margin is not respected after changes in demand,
retirements, and renewables additions, then the model adds new capacity in the region. In making this
calculation, the model takes into account losses in transmission and distribution networks and electricity used
by generation plants themselves.

Because of the stochastic nature of the output of variable renewables such as wind and solar PV, only a
proportion of the installed capacity of these technologies can be considered to contribute to the available
generation margin. This is reflected in the modelling by the use of a capacity credit for variable renewables. This
capacity credit is estimated from historical data on hourly demand and hourly generation from variable
renewables in a number of electricity markets, and it reflects the proportion of their installed capacity that can
reliably be expected to be generating at the time of peak demand.
When new plants are needed, the model makes its choice between different technology options on the basis of
their regional value-adjusted levelised cost of electricity (VALCOE), which are based on the levelised cost of
electricity (LCOE), also referred to as the long-run marginal cost (LRMC). The LRMC of each technology is the
average cost of each unit of electricity produced over the lifetime of a plant, and is calculated as a sum of levelised
capital costs, fixed operation and maintenance (O&M) costs, and variable operating costs. Variable operating
costs are in turn calculated from the fuel cost (including a CO2 price where relevant) and plant efficiency. Our
regional assumptions for capital costs are taken from our own survey of industry views and project costs,
together with estimates from NEA/IEA (2010). The weighted average cost of capital (pre-tax in real terms) is
assumed to be 8% in the OECD and 7% in non-OECD countries.

The LRMC calculated for any plant is partly determined by their utilisation rates. The model takes into account
the fact that plants will have different utilisation rates because of the variation in demand over time, and that
different types of plants are competitive at different utilisation rates. (For example, coal and nuclear tend to be
most competitive at high utilisation rates, while gas and oil plants are most competitive at lower utilisation rates).
The specific numerical assumptions made on capital costs, fixed O&M costs, and efficiency can be found on the
WEO website: https://www.iea.org/reports/world-energy-model/techno-economic-inputs.

The levelised cost module computes LRMCs (or LCOEs) for the following types of plant:
 Coal, oil and gas steam boilers with and without CCUS (carbon capture, utilisation and storage);
 Combined-cycle gas turbine (CCGT) with and without CCUS;
 Open-cycle gas turbine (OCGT);
 Integrated gasification combined cycle (IGCC);
 Oil and gas internal combustion;
 Fuel cells;
 Bioenergy;
 Geothermal;
 Wind onshore;
 Wind offshore;

37
 Hydropower (conventional);
 Solar photovoltaics;
 Concentrating solar power; and
 Marine

Regional LRMCs are also calculated for nuclear power but additions of nuclear power capacity are subject to
government policies.

4.1.2 Generation volumes

For each region, the model determines the generation from each plant based on the capacity installed and the
level of electricity demand. Demand is represented as four segments:

 baseload demand, representing demand with a duration of more than 5944 hours per year;
 low-midload demand, representing demand with a duration of 3128 to 5944 hours per year;
 high-midload demand, representing demand with a duration of 782 to 3128 hours per year; and
 peakload demand, representing demand with a duration of less than 782 hours per year.

The model subtracts from the demand in each segment any generation coming from plants that must run – such
as some CHP plants and desalination plants – and also generation from renewables. For generation from variable
renewables, the amount of generation in each demand segment is estimated based on the historical correlation
between generation and demand. The remainder of the demand in each segment must be met by production
from dispatchable plants. The model determines the mix of dispatchable generation by constructing a merit
order of the plants installed – the cumulative installed generation capacity arranged in order of their variable
generation costs – and finding the point in the merit order that corresponds to the level of demand in each
segment (Figure 16). As a result, plants with low variable generation costs – such as nuclear and lignite-burning
plants in the Figure 16 example – will tend to operate for a high number of hours each year because even
baseload demand is higher than their position in the merit order. On the other hand, some plants with high
variable costs, such as oil-fired plants, will operate only during the peak demand segment.

Figure 16: Example merit order and its intersection with demand in the power generation module

200
$/MWh

150

100
Oil
Steam
Gas and
50 GT
GT
Gas CCGT
CHP Lignite and steam coal
0 Nuclear
Baseload Low-midload High-midload Peak MWh
demand * demand * demand * demand *
*
Demand here means demand net of generation by “must run” plants such as desalination and some CHP plants, and net of generation by
renewables.

38
4.1.3 Calculation of the capacity credit and capacity factor of variable renewables
Power generation from weather-dependent renewables such as wind and solar power varies over time and the
characteristics of the power supply from variable renewables have to be taken into account for the decisions on
dispatch and capacity additions of the remaining, mostly dispatchable power plants. The effect of all variable
renewables (solar PV, solar CSP without storage and wind on- and offshore) is taken into account via the capacity
credit and the capacity factor in each load segment.

The capacity credit of variable renewables reflects the proportion of their installed capacity that can reliably be
expected to be generating at the time of high demand in each segment. It determines by how much non-variable
capacity is needed in each load segment. The capacity factor gives the amount of energy produced by variable
renewables in each load segment and determines how much non-variable generation is needed in each segment.

Both, capacity credit and capacity factor are calculated based the comparison between the hourly load profile
and the wind and solar supply time-series, derived from meteorological data. To quantify the effects of variable
renewables, the hourly load profile is compare to the hourly residual load, being the electricity load after
accounting for power generation from variable renewables (see Figure 17a). By sorting the residual load, the
levels of average and maximal demand per load segment can be determined. The difference between the load
levels of the normal load and the residual load gives the impact of variable renewables on the power generation
and capacity needs (see Figure 17b).

Figure 17: Example electricity demand and residual load

a) Load and residual load for selected days b) Load and residual load duration curve for one year
100 100
GW

Renewable
GW

90 90
energy
80 80 generation
70 70
60 Load
60
50 50
40 40
30 30 Residual
20 20 load
10 10
0 0
1
1001
2001
3001
4001
5001
6001
7001
8001
1
25
49
73
97
121
145
169
193
217
241
265
289
313
337

hour hours (sorted)

The capacity factor of variable renewables (varRE) per load segment can be calculated generation per load
segment s of the residual load.
𝑅𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝐺𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑁𝑒𝑒𝑑𝑠𝑛𝑜𝑛−𝑣𝑎𝑟,𝑠 Generation varREs
𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑓𝑎𝑐𝑡𝑜𝑟𝑠 = =
𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑣𝑎𝑟𝑅𝐸 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑣𝑎𝑟𝑅𝐸
For capacity additions, the peak load segment is relevant. The capacity credit is estimated based on the difference
between maximal load and maximal residual load:
𝑅𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑁𝑒𝑒𝑑𝑠𝑛𝑜𝑛−𝑣𝑎𝑟 maxt (𝐿𝑜𝑎𝑑(𝑡)) − maxt (𝑅𝑒𝑠𝑖𝑑𝑢𝑎𝑙 𝐿𝑜𝑎𝑑(𝑡))
𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑐𝑟𝑒𝑑𝑖𝑡𝑝𝑒𝑎𝑘 = =
𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑣𝑎𝑟𝑅𝐸 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑣𝑎𝑟𝑅𝐸
Meteorological data (wind speed and solar irradiation) for several years was used for the capacity credit
calculation. In aggregating the results of capacity credit obtained from different years of meteorological data, as
first order approach it was assumed that the annual peak residual demand is normally-distributed and calculated
the capacity credit based on the difference between peak demand and the point one standard deviation above
the residual peak demand (Figure 18).

39
Figure 18: Exemplary electricity demand and residual load

p
Capacity credit calculation based
on this difference

Probability density
function of residual
demand- σ

Expected peak Peak GW

residual demand demand

The meteorological data wind and solar data stems from the following re-analysis datasets

 World Wind Atlas (Sander + Partner GmbH): Global dataset of hourly wind speeds at 10 m height, 1979-
2009, derived from reanalysis data based on climate modelling (Suraniana, 2010)
 Wind supply time-series for West and Eastern US as derived by WWITS (2010) and EWITS (2011).
 Wind and solar supply time-series for Europe-27 as provided by Siemens AG (Heide, 2010) for each
major Region in Europe. Original meteorological wind speed stems from Reanalysis data (WEBROG,
2008).
 Hourly solar irradiation data from satellite observations for the US (NREL,2010)
 Estimation of solar irradiation based on solar height (Aboumahboub, 2010)

4.2 Value-adjusted Levelized Cost of Electricity

Major contributors to the Levelized Cost of Electricity (LCOE) include overnight capital costs; capacity factor that
describes the average output over the year relative to the maximum rated capacity (typical values provided); the
cost of fuel inputs; plus operation and maintenance. Economic lifetime assumptions are 25 years for solar PV,
onshore and offshore wind. For all technologies, a standard weighted average cost of capital was assumed (7-8%
based on the stage of economic development, in real terms). Details are in Table 6.
The value-adjusted LCOE (VALCOE) is a metric for competitiveness for power generation technologies and was
presented for the first time in WEO-2018, building on the capabilities of the WEM hourly power supply model. It
is intended to complement the LCOE, which only captures relevant information on costs and does not reflect the
differing value propositions of technologies. While LCOE has the advantage of compressing all the direct
technology costs into a single metric which is easy to understand, it nevertheless has significant shortcomings: it
lacks representation of value or indirect costs to the system and it is particularly poor for comparing technologies
that operate differently (e.g. variable renewables and dispatchable technologies). VALCOE enables comparisons
that take account of both cost and value to be made between variable renewables and dispatchable thermal
technologies.

40
Table 6: Technology costs by selected region in the Stated Policies Scenario
Capital costs Capacity factor Fuel and O&M LCOE VALCOE
($/kW) (%) ($/MWh) ($/MWh) ($/MWh)
2019 2040 2019 2040 2019 2040 2019 2040 2019 2040
United Nuclear 5000 4500 90 90 30 30 105 100 105 100
States Coal 2100 2100 60 60 30 30 75 75 75 75
Gas CCGT 1000 1000 50 50 30 40 50 65 45 65
Solar PV 1220 680 21 23 10 10 75 40 80 55
Wind onshore 1560 1440 42 44 10 10 50 45 55 55
Wind offshore 4260 2160 41 48 35 20 155 70 150 75
European Nuclear 6600 4500 75 75 35 35 150 110 145 115
Union Coal 2000 2000 40 40 40 40 120 150 110 125
Gas CCGT 1000 1000 40 40 45 65 65 110 50 75
Solar PV 840 490 13 14 10 10 85 50 85 80
Wind onshore 1560 1420 28 31 15 15 75 65 80 80
Wind offshore 3800 2040 49 59 15 10 110 50 115 60
China Nuclear 2600 2500 80 80 25 25 65 60 65 60
Coal 800 800 60 60 30 30 50 70 50 60
Gas CCGT 560 560 50 50 60 75 75 100 70 95
Solar PV 790 450 17 19 10 5 55 30 55 55
Wind onshore 1220 1140 25 27 15 10 65 55 65 60
Wind offshore 3000 1640 32 44 25 15 130 55 130 55
India Nuclear 2800 2800 80 80 30 30 70 70 70 70
Coal 1200 1200 60 60 30 35 55 55 55 50
Gas CCGT 700 700 50 50 50 70 60 85 60 65
Solar PV 610 350 20 21 5 5 35 20 40 50
Wind onshore 1060 1020 26 29 10 10 55 50 60 55
Wind offshore 3140 1700 29 38 25 15 135 60 135 70
Note: O&M = operation and maintenance; LCOE = levelised cost of electricity; VALCOE = value-adjusted LCOE; kW = kilowatt; MWh =
megawatt-hour; CCGT = combined-cycle gas turbine. LCOE and VALCOEs figures are rounded. Lower figures for VALCOE indicate improved
competitiveness.
Source: IEA WEO-2020

The VALCOE builds on the foundation of the average LCOE (or LRMC) by technology, adding three elements of
value: energy, capacity and flexibility. For each technology, the estimated value elements are compared against
the system average in order to calculate the adjustment (either up or down) to the LCOE. After adjustments are
applied to all technologies, the VALCOE then provides a basis for evaluating competitiveness, with the technology
that has the lowest number being the most competitive (Figure 19). The VALCOE is applicable in all systems, as
energy, capacity and flexibility services are provided and necessary in all systems, even though they may not be
remunerated individually. In this way, it takes the perspective of policy makers and planners. It does not
necessarily represent the perspective of investors, who would consider only available revenue streams, which
may also include subsidies and other support measures, such as special tax provisions, that are not included in
the VALCOE.

41
Figure 19: Moving beyond the LCOE, to the value-adjusted LCOE

The impact of the value adjustment varies by technology depending on operating patterns and system-specific
conditions. Dispatchable technologies that operate only during peak times have high costs per MWh, but also
relatively high value per MWh. For baseload technologies, value tends to be close to the system average and
therefore they have a small value adjustment. For variable renewables, the value adjustment depends mainly on
the resource and production profile, the alignment with the shape of electricity demand and the share of variable
renewables already in the system. Different operational patterns can be accounted for in the VALCOE, improving
comparisons across dispatchable technologies.

The VALCOE is composed of LCOE and energy, capacity as well as flexibility value. Its calculation goes as follows:
𝑉𝑎𝑙𝑢𝑒 𝑎𝑑𝑗𝑢𝑠𝑡𝑚𝑒𝑛𝑡𝑠
⏞
𝐸𝑛𝑒𝑟𝑔𝑦 𝑣𝑎𝑙𝑢𝑒 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑣𝑎𝑙𝑢𝑒 𝐹𝑙𝑒𝑥𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝑣𝑎𝑙𝑢𝑒

𝑉𝐴𝐿𝐶𝑂𝐸𝑥 = 𝐿𝐶𝑂𝐸𝑥 + [𝐸 − 𝐸𝑥 ] + ⏞
⏞̅ [𝐶̅ − 𝐶𝑥 ] +⏞
[𝐹̅ − 𝐹𝑥 ]

The adjustment for energy value [𝐸𝑥 ] of a technology x (or generation unit) is the difference between the
̅ ]. [𝐸𝑥 ] is calculated as follows:
individual unit to the system average unit [𝐸
8760 $
$ ∑ℎ [𝑊ℎ𝑜𝑙𝑒𝑠𝑎𝑙𝑒𝑃𝑟𝑖𝑐𝑒ℎ ( ) 𝑥 𝑂𝑢𝑡𝑝𝑢𝑡𝑥,ℎ (𝑀𝑊)]
𝑀𝑊ℎ
𝐸𝑛𝑒𝑟𝑔𝑦 𝑣𝑎𝑙𝑢𝑒 𝑥 ( )=
𝑀𝑊ℎ ∑8760
ℎ 𝑂𝑢𝑡𝑝𝑢𝑡𝑥,ℎ (𝑀𝑊)

Wholesale electricity prices and output volumes for each technology x in each hour h of the year are simulated.
Wholesale prices are based on the marginal cost of generation only and do not include any scarcity pricing or
other cost adders, such as operating reserves demand curves present in US markets. Hourly models are applied
for the United States, European Union, China and India. For other regions, wholesale prices and output volumes
are simulated for the four segments of the year presented in section 4.1.2.

The adjustment for capacity value [𝐶𝑥 ] of a generation unit is calculated as follows:
$ 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑐𝑟𝑒𝑑𝑖𝑡𝑥 𝑥 𝐵𝑎𝑠𝑖𝑠 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑣𝑎𝑙𝑢𝑒 ($/𝑘𝑊)
𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑣𝑎𝑙𝑢𝑒𝑥 ( )=
𝑀𝑊ℎ (𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑓𝑎𝑐𝑡𝑜𝑟𝑥 𝑥 ℎ𝑜𝑢𝑟𝑠 𝑖𝑛 𝑦𝑒𝑎𝑟/1000)

42
The Capacity credit reflects the contribution to system adequacy and it is differentiated for dispatchable versus
renewable technologies:

 Dispatchable power plants = (1-unplanned outage rate by technology)

 Renewables = analysis of technology-specific values by region with hourly modelling

The Basis capacity value is determined based on simulation of capacity market, set by the highest “bid” for
capacity payment. Positive bids reflect the payment needed to fill the gap between total generation costs
(including capital recovery) and available revenue.

The Capacity factor is differentiated by technology:

 Dispatchable power plants = modelled as simulated operations in previous year

 Wind and solar PV = aligned with latest performance data from IRENA and other sources, improving
over time due to technology improvements
 Hydropower and other renewables = aligned with latest performance data by region and long-term
regional averages

The flexibility value [𝐹𝑥 ] of a generation unit is calculated as follows:

$
$ 𝐹𝑙𝑒𝑥𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝑣𝑎𝑙𝑢𝑒 𝑚𝑢𝑙𝑡𝑖𝑝𝑙𝑖𝑒𝑟𝑥 𝑥 𝐵𝑎𝑠𝑒 𝑓𝑙𝑒𝑥𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝑣𝑎𝑙𝑢𝑒 ( )
𝑘𝑊
𝐹𝑙𝑒𝑥𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝑣𝑎𝑙𝑢𝑒𝑥 ( )=
𝑀𝑊ℎ (𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑓𝑎𝑐𝑡𝑜𝑟𝑥 𝑥 ℎ𝑜𝑢𝑟𝑠 𝑖𝑛 𝑦𝑒𝑎𝑟/1000)

 The Flexibility value multiplier by technology is based on available market data and held constant over
time. Targeted changes in the operations of power plants to increase flexibility value are not
represented.
 The Base flexibility value is a function of the annual share of variable renewables in generation,
informed by available market data in the EU and US. The flexibility value is assumed to increase with
rising VRE shares, up to a maximum equal to the full fixed capital recovery costs of a peaking plant.

4.2.1 Advantages and limitations of the VALCOE

VALCOE has several advantages over the LCOE alone:

 It provides a more sophisticated metric of competitiveness incorporating technology-specific

information and system-specific characteristics
 It reflects information/estimations of value provided to the system by each technology (energy,
capacity/adequacy and flexibility)
 It provides a robust metric of competitiveness across technologies with different operational
characteristics (e.g. baseload to peaking, or dispatchable to variable)
 It provides a robust metric of competitiveness with rising shares of wind and solar PV

However, network integration costs are not included, nor are environmental externalities unless explicitly priced
in the markets. Fuel diversity concerns, a critical element of electricity security, are also not reflected in the
VALCOE.
The VALCOE approach has some parallels elsewhere, in other approaches used for long-term energy analysis, as
well some real-world applications. Optimisation models implicitly represent the cost and value of technologies,
but may be limited by the scope of costs included, such as those related to ancillary services. Other long-term
energy modelling frameworks, such as the NEMS model used by the US Department of Energy, have incorporated
cost and value in capacity expansion decisions. In policy applications, in the auction schemes in Mexico, average

43
energy values for prospective projects have been simulated and used to adjust the bid prices, seeking to identify
the most cost-effective projects.

4.2.2 Financing costs for utility-scale solar PV

The declining costs of solar PV have been impressive, with innovation driving down construction costs by 80%
from 2010 to 2019 (IRENA, 2020). Cost reductions have been complemented by improved performance resulting
from higher efficiency panels and greater use of tracking equipment. Financing costs, however, have received
little attention despite their importance. The weighted average cost of capital (WACC) can account for until half
of the levelised cost of electricity (LCOE) of utility-scale solar PV projects.

WEO-2020 focused on financing cost through an extensive work based on data from financial markets and
academic literature, and on the analysis of auction results and power purchase agreements (PPAs),
complemented by a large number of confidential interviews with experts and practitioners around the world.
The analysis of business models draws on the key revenue risk components – price, volume and off-taker risk –
and their implications for the cost of capital. It focuses on models where prices paid for solar generation are
defined largely by policy mechanisms, which support the vast majority of deployment worldwide. The findings
of this analysis on the prevailing average costs of capital in major solar PV markets underpin the projections in
the IEA World Energy Model. Full merchant projects (without any form of price guarantee external to markets)
were considered as a point of comparison and an indicative WACC provided, though to date this model remains
somewhat theoretical for solar PV. In the longer term, this type of investment may become more common.

4.3 Electricity transmission and distribution networks

The model calculates investment in transmission and distribution networks. Transmission networks transport
large volumes of electricity over long distances at high voltage. Most large generators and some large-scale
industrial users of electricity are connected directly to transmission networks. Distribution networks transform
high-voltage electricity from the transmission network into lower voltages, for use by light-industrial,
commercial, and domestic end-users.
Investment in grid infrastructure are driven by three factors: investment in new grid infrastructure to
accommodate growing demand, investment to replace or refurbish assets that reach the end of their operational
lifetime and investments required to integrate renewables in the power sector.

4.3.1 Investment due to new growth

New investment due to growth in electricity demand is assumed to scale with increase in electricity demand.
This calculation is performed for distribution and transmission networks separately and for each region, as the
increase in line length per region depends on a number of region-specific factors (e.g. population density).

The investment per type (transmission and distribution) is calculated as follows:

𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 𝑛𝑒𝑤 = (𝛽 ∙ 𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝑖𝑛 𝑝𝑜𝑤𝑒𝑟 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛) ∙ 𝐿𝑖𝑛𝑒 𝑐𝑜𝑠𝑡𝑠

The term 𝛽 , which reflects the additional amount of network length needed for each additional unit of
generation, is estimated for each region using data on network length and generation for the period 1970-2014.6
The unit costs of addition transmission or distribution lines are derived from observed capital expenditure data.

6 The data on historic network size and capital expenditure stems from ABS Energy Research (2010), Global
Transmission & Distribution Report Ed. 9-2010, and NRG Expert (2013), Electricity Transmission and Distribution Report
and Database Edition 3, 2012-2017.

44
For future years we assume that the real unit cost of networks increases as labour costs increase, taking into
account the differences in labour costs between countries.

4.3.2 Investment due to ageing infrastructure

Assuming an average lifetime of 40 years, the amount of grid infrastructure in need for refurbishment is
determined and the corresponding investment is calculated
𝑎𝑔𝑒
𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 = 𝐿𝑖𝑛𝑒 𝑙𝑒𝑛𝑔𝑡ℎ 𝑟𝑒𝑎𝑐ℎ𝑖𝑛𝑔 40 𝑦𝑒𝑎𝑟𝑠 ∗ (𝐿𝑖𝑛𝑒 𝑐𝑜𝑠𝑡𝑠 ∗ 𝑟𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑓𝑎𝑐𝑡𝑜𝑟)

Because building new assets entails additional costs to those entailed in refurbishing them, a region-specific cost
reduction factor is introduced.

4.3.3 Additional investment due to renewables

A considerable amount of the capacity additions projected over the WEO period is from renewables. The
geographical location of these technologies is often strongly influenced by the location of the underlying resource
(e.g. areas where the wind is strong or insolation is high), which may not be close to existing centres of demand.
In addition, some of these technologies, mainly solar PV, are connected at the end-user side of the grid
infrastructure. This modular deployment of generation capacity can lead to increase distribution capacity needs.

Because the introduction of large quantities of remote or variable renewables was not a marked feature of the
historic development of electricity networks (with the exception of regions where remote hydroelectricity
represents a large proportion of the generation mix), the addition of more renewables is likely to increase the
average length of network additions and the cost of transmission and distribution per unit of energy.

Additional transmission network costs are derived based on specific renewable grid integration costs, derived
from a literature review. For example for wind, the typically range between $100 and $250 per kW of installed
wind capacity. Regional differences due to geography and labour costs are taken into account.

The estimation of costs of distribution grid extensions for renewables contains a lot more uncertainties than the
transmission grid costs, as less data or studies are available on the technically complex distribution network is
available and own use of distributed generation can in turn lead to a reduced need for distribution grid
infrastructure. Therefore, we assume, that additional network investment is required only if the electricity
generated from distributed generation, such as solar PV in buildings and bioenergy in industry, exceeds local
demand and is fed back to the system.

4.3.4 Investment for digitalisation of grid operations

In WEO-2020 an estimation of the share of digital electricity grids investment by scenario was made for the first
time. This estimation builds on historical data collected from TSOs, electricity network service providers and
smart grid actors.

4.4 Hourly model

To quantify the scale of the challenge arising from the integration of high shares of VRE and to assess which
measures could be used to minimise curtailment, a new hourly model has been developed for WEO-2016, to
provide further insights into the operations of power systems. The model builds upon the annual projections
generated in the WEM and makes it possible to explore emerging issues in power systems, such as those that
arise as the share of VRE continues to rise. The model then feeds the main WEM model with information about
additional constraints on the operations of different power plants. The model is a classical hourly dispatch model,
representing all hours in the year, setting the objective of meeting electricity demand in each hour of the day for

45
each day of the year at the lowest possible cost, while respecting operational constraints. 7 All 106 power plant
types recorded in the WEM and their installed capacities are represented in the hourly model, including existing
and new fossil-fuelled power plants, nuclear plants and 16 different renewable energy technologies. The fleet of
power plants that is available in each year is determined in WEM and differs by scenario, depending on the
prevalent policy framework. These plants are then made available to the hourly model and are dispatched (or
chosen to operate) on the basis of the short-run marginal operating costs of each plant (which are mainly
determined by fuel costs as projected in WEM) to the extent required to meet demand. The dispatch operates
under constraints: there are minimum generation levels to ensure the flexibility and stability of the power system
and to meet other needs (such as combined heat and power); the variability of renewable resources (such as
wind and solar) determines the availability of variable renewables and, hence, the maximum output at any point
in time; and ramping constraints apply, derived from the level of output in the preceding hour and the
characteristics of different types of power plants. The hourly dispatch model does not represent the transmission
and distribution system, nor grid bottlenecks, cross-border flows or the flow of power through the grid. It
therefore simulates systems that are able to achieve full integration across balancing areas in each WEM region
(e.g. United States, European Union, China and India).

Key inputs to the model include detailed aggregate hourly production profiles for wind power and solar PV for
each region, which were generated for the WEO by combining simulated production profiles for hundreds of
individual wind parks and solar PV installations, distributed across the relevant region 8 The individual sites were
chosen to represent a broad distribution within a region, allowing the model to represent the smoothing effect
achieved by expanding balancing areas. On the demand side, the model uses a detailed analysis, with hourly
demand profiles for each specific end-use (such as for lighting or water heating in the residential sector), coupled
with the annual evolution of electricity demand by specific end-use over the Outlook period from the main WEM
model (see Section 3.4).

The hourly model accounts for grid, flexible generation and system-friendly development of VRE, in three steps:
first, it assesses the amount of curtailment of variable renewables that would occur without demand-side
response and storage. Second, it deploys demand-side response measures, based on the available potential in
each hour for each electricity end-use. And third, it uses existing and new storage facilities to determine the
economic operations of storage based on the price differential across hours and charge/discharge periods. It
thereby enables the integration needs arising from growing shares of renewables to be assessed.
Among the other important model outputs is the resulting hourly market price, which can drop to zero in the
hours when generation from zero marginal cost generators (such as variable renewables) is sufficient to meet
demand. By multiplying the market price by generation output in each hour, the model calculates the revenues
received for the output in each hour by each type of plant, creating a basis for calculating the value of VRE.
Naturally, the model also includes hourly operation information for each plant type, including fuel costs and
associated greenhouse-gas and pollutant emissions.

4.5 Mini- and off-grid power systems

Since the Africa Energy Outlook in 2014, the representation of mini- and off-grid systems, related to those gaining
access to electricity, has been improved and better integrated into the WEM. In line with the approach for on-
grid power systems, to meet additional electricity demand, the model chooses between available technologies

7 The model works on an hourly granularity, and therefore all intra-hour values of different devices (e.g. of storage
technologies) are not captured.
8 Wind and solar PV data are from Renewables.ninja (https://beta.renewables.ninja/) and Ueckerdt, F., et. al. (2016).

46
for mini- and off-grid systems based on their regional long-run marginal costs, and using detailed geospatial
modelling to take into account several determining factors. For the Africa Energy Outlook 2019, the IEA refined
its analysis using up-to-date technology costs, demand projections, and the latest version of the Open Source
Spatial Electrification Tool (OnSSET)9 developed by KTH, to cover in detail 44 countries in sub-Saharan Africa. The
technologies are restricted by the available resources in each region, including renewable energy resources such
as river systems, biomass feedstocks (e.g. forests and agricultural residues), wind and the strength of solar
insolation. Back-up power generation for those with access to the grid, typically gasoline or diesel fuelled, was
also represented to the model, with its projected use tied to the quality of the on-grid power supply.

4.6 Renewables, combined heat and power and distributed generation

modules
The projections for renewable electricity generation, combined heat and power (CHP), and distributed
generation (DG) are derived in separate sub-modules. The future deployment of these technologies and the
investment needed for such deployment were assessed on the basis of potentials in each country/region.

4.6.1 Combined heat and power and distributed generation

The CHP option is considered for fossil fuel and bioenergy-based power plants. The CHP sub-module uses the
potential for heat production in industry and buildings together with heat demand projections, which are
estimated econometrically in the demand modules. The distributed generation sub-module is based on
assumptions about market penetration of DG technologies.

4.6.2 Renewable energy

The projections of renewable electricity generation are derived in the renewables sub-module, which projects
the future deployment of renewable sources for electricity generation and the investment needed. The
deployment of renewables is based on an assessment of the potential and costs for each source (bioenergy,
hydropower, photovoltaics, concentrating solar power, geothermal electricity, wind, and marine) in each of the
25 WEM regions.10 Our modelling, in all scenarios, incorporates a process of learning-by-doing that affects the
costs. The evolution of the cost components are ultimately expressed through the the value-adjusted LCOE
metric for evaluating competitiveness under market conditions (see section 4.2). However, while technology
learning is integral to the approach, the Outlook does not try to anticipate technology breakthroughs. By
including financial or other incentives for the use of renewables and non-financial barriers in each market, as
well as technical and social constraints, the model calculates deployment as well as the resulting investment
needs on a yearly basis for each renewable source in each region. The methodology is illustrated in Figure 20.11

9 For more details on the Open Source Spatial Electrification Tool, see www.onsset.org; for the latest
OnSSET methodology update refer to Korkovelos, A. et al. (2019).
10A number of sub-types of these technologies are modelled individually, as follows. Biomass: small CHP, medium CHP,
electricity only power plants, biogas-fired, waste-to-energy fired and co-fired plants. Hydro: large (≥10MW) and small
(<10MW). Wind: onshore and offshore. Solar photovoltaics: large-scale and buildings. Geothermal: electricity only and CHP.
Marine: tidal and wave technologies.
11 For a detailed description of the approach used in this model – which was originally developed by Energy Economics Group
(EEG) at Vienna University of Technology in cooperation with Wiener Zentrum für Energie, Umwelt und Klima – see Resch
et.al. (2004).

47
Figure 20: Approach used for the renewables module

Database on costs
Database on potentials

Long -term realisable potentials in Learining rates

terms of primary use and/or electricity Regional differentiation
(global and regional)
(generation and capacity)

Specific investment
GDP per capita
costs
Restrictions
(market, technical and
societal constraints)
Operating and maintenance
Fuel costs
costs

Dynamic cost-resource curves Financial incentives

Calculation of renewables deployment and investment needs

The model uses dynamic cost-resource curves.12 The approach consists of two parts:

1) For each renewable source within each region, static cost-resource curves13 are developed. For new
plants, we determine long-term marginal generation costs. Realisable long-term potentials (see Box 2)
have been assessed for each type of renewable in each region.
2) Next, the model develops for each year a dynamic assessment of the previously described static cost-
resource curves, consisting of:
Dynamic cost assessment: The dynamic adaptation of costs (in particular the investment and the
operation and maintenance components) is based on the approach known as “technological
learning”. Learning rates are assumed by decade for specific technologies.
Dynamic restrictions: To derive realisable potentials for each year of the simulation, dynamic
restrictions are applied to the predefined overall long-term potentials. Default figures are derived
from an assessment of the historical development of renewables and the barriers they must
overcome, which include:

12 Theconcept of dynamic cost-resource curves in the field of energy policy modelling was originally devised for the research
project Green-X, a joint European research project funded by the European Union’s fifth Research and Technological
Development Framework Programme – for details see www.green-x.at.
13 Renewable energy sources are characterised by limited resources. Costs rise with increased utilisation, as in the case of
wind power. One tool to describe both costs and potentials is the (static) cost-resource curve. It describes the relationship
between (categories of) available potentials (wind energy, biomass, and hydropower) and the corresponding (full) costs of
utilisation of this potential at a given point-of-time.

48
  Market constraints: The penetration of renewables follows an S-curve pattern, which is
typical of any new commodity.14 Within the model, a polynomial function has been chosen
to describe this impact – representing the market and administrative constraints by region.
 Technical barriers: Grid constraints are implemented as annual restrictions which limit the
penetration to a certain percentage of the overall realisable potential.

Box 2: Long-term potential of renewables

The starting point for deriving future deployment of renewables is the assessment of long-term realisable
potentials for each type of renewable and for each world region. The assessment is based on a review of the
existing literature and on the refinement of available data. It includes the following steps:
1. The theoretical potentials for each region are derived. General physical parameters are taken into
account to determine the theoretical upper limit of what can be produced from a particular energy,
based on current scientific knowledge.
2. The technical potential can be derived from an observation of such boundary conditions as the
efficiency of conversion technologies and the available land area to install wind turbines. For most
resources, technical potential is a changing factor. With increased research and development,
conversion technologies might be improved and the technical potential increased.

Long-term realisable potential is the fraction of the overall technical potential that can be actually realised
in the long term. To estimate it, overall constraints like technical feasibility, social acceptance, planning
requirements and industrial growth are taken into consideration.

4.6.3 Wind offshore technical potential

In collaboration with Imperial College London, a detailed geospatial analysis was undertaken for WEO-2019 to
assess the technical potential for offshore wind worldwide. The study was among the first to use the “ERA-5”
reanalysis, which provides four decades of historic global weather data. “Renewables.ninja” extrapolates wind
speeds to the desired hub height and converts them to output using manufacturers’ power curves for turbine
models. Results can be found on the IEA website.

Data
The availability of high-resolution satellite data and computing gains has significantly improved the granularity
and accuracy of wind resource assessments in recent years. Emerging wind turbine designs are also cause to
update potential assessments, as they increase performance in well-established areas and make lower quality
resources more suitable for energy production.

Exclusions
Commercially available offshore wind turbines are currently designed for wind speeds of more than 6 m/s. Some
companies are also looking into turbine designs for lower wind speeds.

Following the International Union for Conservation of Nature’s (IUCN) classification of maritime protection areas,
those categorised as Ia, Ib, II and III were excluded from the study (IUCN, 2013). However, at each project level
other environmental considerations must also be taken into account and a full environmental impact assessment

14 AnS-curve shows relatively modest growth in the early stage of deployment, as the costs of technologies are gradually
reduced. As this is achieved, there will be accelerating deployment. This will finally be followed by a slowing-down,
corresponding to near saturation of the market.

49
conducted as mandated by public authorities. Buffer zones were also excluded for existing submarine cables
(within 1 kilometre [km]), major shipping lanes (20 km), earthquake fault lines (20 km) and competing uses such
as existing offshore oil and gas installations and fisheries.

Turbine designs
In order to assess the global technical potential, best-in-class turbines were chosen with specific power of 250,
300 and 350 watt per square metre (W/m2) that corresponds to low-medium, medium and high wind speeds.
The power curves of these turbines were used in conjunction with the global capacity factors of each
5 km by 5 km cell selected for the analysis to derive the technical potential of offshore wind in terms of capacity
and generation. New power curves were synthesised for next-generation turbines with rated capacity of up to
20 MW, for which data are not yet available (Saint-Drenan et al., 2019).

Further to this, the analysis takes into account further considerations such as offshore wind farm designs,
distance from shore and water depth, offshore wind cost devlopments and the technical potential.

5 Other energy transformation

5.1 Oil refining and trade
The refinery and trade module links oil supply and demand. It is a simulation model, with capacity development
and utilisation modelled for 134 individual countries, with the remaining countries grouped into 11 regions. This
module has several auxiliaries that stretch into supply and demand domains to better link both:

- Natural gas liquids module to determine yields of various products as well as condensate.
- Extra-heavy oil and bitumen module to model synthetic crude oil output and diluent requirements for
bitumen.
- Split of oil demand into different production categories for all sectors except road transport and
aviation. The latter are provided by WEM’s transport demand model.

Crude distillation (CDU) capacity is based on 2019 data from the IEA. Capacity expansion projects that are
currently announced are assessed individually to identify only the projects that are very likely to go ahead. Some
of these are delayed from their announced start-up dates to allow for a more realistic timeline. The model also
takes into account refinery closures that have been announced. Beyond 2025, new capacity expansion is
projected based on crude availability and product demand prospects for each of the regions specified below.
Projections for refining sector activity are based primarily on CDU capacity and utilisation rates. Secondary unit
capacities (such as fluid catalytic cracking, hydrocracking) and run rates, are defined by the required output mix
to match product demand. Among oil-importing regions, priority call on international supply of crude oil is given
to those where demand is growing: robust domestic demand is effectively a proxy for refinery margins that are
not explicitly calculated or used by the model.

50
Figure 21: Schematic of refining and international trade module

Region A Region B
regional regional
demand demand
refining refining
capacity
products capacity
crude oil crude oil
output output
products

products
Crude oil
market
regional regional
demand demand
refining refining
capacity products capacity
crude oil crude oil
output output
Region C Region D

Oil output and demand projections are provided by WEM’s fossil-fuel supply and final energy consumption
modules. Refineries do not provide for 100% of oil product demand. For the purposes of this analysis, we show
the net call on refineries after the removal of biofuels, liquefied petroleum gas (LPG), ethane and light naphtha
from natural gas liquids (NGL), synthetic liquids from coal-to-liquids (CTL) and gas-to-liquids (GTL) and additives.

The supply-side nomenclature for the refining model is slightly different from the oil supply model. The term
“crude oil” used in the model describes all crude oils that have conventional-type quality for processing purposes.
This includes conventional crude oil from the supply model, some extra heavy oils that are not diluted or
upgraded, tight oil and synthetic crude from bitumen upgrading processes. Diluted bitumen and condensate are
represented as separate streams for intake and trade modelling purposes.
Yields, output and trade are defined for the following product categories: LPG, naphtha, gasoline, kerosene,
diesel, heavy fuel oil and other products (which include petroleum coke, refinery gas, asphalt, solvents, wax, etc).
Vacuum gas oil is added to the middle distillates pool for trade flow purposes. Crude oil trade position refined
products balances follow WEO’s demand model granularity of 26 individual countries or regions (Figure 21).

5.2 Coal-to-liquids, Gas-to-liquids, Coal-to-gas

Coal and natural gas are hydrocarbons that can be transformed to other hydrocarbons through thermal and/or
catalytic processes. The most common example is the Fischer-Tropsch reaction, which transforms syngas – a
mixture of carbon monoxide and hydrogen mainly – into carbon chains of 5 to 20 carbons. The syngas is produced
by a high temperature and relatively high pressure process called gasification. The syngas can also be used to
produce methane via Sabatier reaction – in that case the transformation is called coal-to-gas.
These technologies are relatively costly, especially in CAPEX, and therefore require very cheap coal or gas to
compete with international markets. They are also developed to increase energy sufficiency and sovereignety in
large consumer countries. Best example is China. For these two reasons, there is a currently limited number of

51
projects in the pipeline in a limited number of countries. Projections are consistent with the status of the projects
– under construction or planned – and are updated every year on a project by project basis.

In WEO-2020, accounting of energy-related CO2 emissions, also know as “process own use”, have been aligned
with the other energy transformation processes. CCUS assumptions have also been refined to be consistent with
other large scale energy models, such as Energy Technology Perspectives model (IEA, 2020b).

5.3 Hydrogen transformation

There are different pathways to produce hydrogen, which is then used in end-use, refinery and the power and
heat sectors. In the WEM, three of them are included:
 Steam methane reforming: methane reacts with steam under relatively high pressure in the presence
of a catalyst to produce hydrogen, carbon monoxide, and a relatively small amount of carbon dioxide.
Steam reforming is endothermic: heat must be supplied to the process for the reaction to proceed. This
process accounts for the lion’s share of hydrogen production today, especially for ammonia production
and refineries. The hydrogen yield can be improved with water gas shift (WGS) reaction to produce
carbon dioxide and additional hydrogen from carbon monoxide and water.
 Steam methane reforming equipped with CCUS unit: low-carbon hydrogen can be produced from
natural gas with the addition of a carbon capture process (direct capture of CO2 flow from the WGS and
amine scrubbing or metal organic adsorption for natural gas combustion) to the steam methane
reformer.
 Water electrolysis: in an electrolyser, water is split into hydrogen and oxygen. There are several
technologies under development (low temperature, high temperature, with or without membranes).
Main energy applications use to be to provide reliable energy supply in very remote areas but
electrolysers are more and more deployed, coupled with solar PV and wind plants, to avoid curtailement
and produce when electricity value converges to zero.

The market shares of these three pathways are assessed on a region-by-region basis, depending on natural gas
price and availability and solar PV and wind potential. In addition, policies are considered: for example, the EU
hydrogen strategy is part of the Sustainable Development Scenario.

6 Energy supply
6.1 Oil
The purpose of this module is to project the level of oil production in each country through a partial bottom-up
approach15 building on:

 the historical series of production by countries;

 standard production profiles and estimates of decline rates at field and country levels derived from the
detailed field-by-field analysis first undertaken in WEO-2008 and updated since;
 an extensive survey of upstream projects sanctioned, planned and announced over the short term in both
OPEC and Non-OPEC countries, including conventional and non-conventional reserves, as performed by the
IEA Oil Market Report team; this is used to drive production in the first 5 years of the projection period (a
summary of the differences in methodology between WEO and the Medium-Term Oil Market Report is
included as Box 3);

15 “Bottom-up” in this context means “based on field-by-field analysis”.

52
 a methodology, which aims to replicate as much as possible the decision mode of the industry in developing
new reserves by using the criteria of net present value of future cash flows;
 a set of economic assumptions discussed with and validated by the industry including the discount rate used
in the economic analysis of potential projects, finding and development costs, and lifting costs;
 an extensive survey of fiscal regimes translating into an estimate of each government’s take in the cash flows
generated by projects; and
 values of remaining technically recoverable resources (Table 7) calculated based on information from the
United States Geological Survey (USGS), BGR and other sources.

The paragraphs below describe how the USGS data are used in the WEM. USGS publishes its World Petroleum
Assessment, a thorough review of worldwide conventional oil (and gas) resources. In it, USGS divided the
resources into three parts:

 Known oil, which contains both cumulative production and reserves in known reservoirs.
 Undiscovered oil, a basin-by-basin estimate of how much more oil there may be to be found, based on
knowledge of petroleum geology.
 Reserves growth, an estimate of how much oil may be produced from known reservoirs on top of the known
reserves. As the name indicates, this is based on the observation that estimates of reserves (including
cumulative production) in known reservoirs tend to grow with time as knowledge of the reservoir and
technology improves. For the 2000 assessment, reserve growth as a function of time after discovery was
calibrated from observation in US fields, and this calibration applied to the known worldwide reserves to
obtain an estimate of worldwide reserves growth potential.

Since the 2000 assessment, USGS has regularly published updates on undiscovered oil in various basins, and
these were considered in the WEM. In 2012, USGS published an updated summary of worldwide undiscovered
oil, as well as a revised estimate for reserves growth based on a new field-by-field method focused on the large
fields in the world. Previously the known oil estimates used by the USGS when generating its reserve growth
estimates had not been released publicly. However, a recent report provides its assumptions, albeit aggregated
at a global level (USGS, 2015). The USGS estimate of cumulative production and reserves outside the United
States is 2 060 billion barrels, which is in close alignment with the IEA equivalent estimate of 2 050 billion barrels.
For conventional oil, the USGS estimates of undiscovered oil and reserves growth published in 2012 provide the
key foundation for the values used in WEM. The WEM estimates of remaining technically recoverable resources
combine USGS undiscovered, USGS reserves growth and IEA estimates for known. A similar analysis, based on
the same USGS publications, feeds into the IEA NGLs and natural gas resources database, which allows looking
at total conventional liquid hydrocarbons resources and conventional gas resources.

Box 3: WEO differences in methodology compared with the Medium-Term Oil Market Report

The IEA publishes annually projections of oil supply and demand for the next five years in the Medium Term
Oil Market Report (MTOMR), and for the next two and half decades in the WEO. Those two sets of
projections use different methodologies that evolve every year. This makes comparisons not
straightforward for some readers. This box summarizes the key differences.

A very important difference between MTOMR and WEO is the oil price assumption. MTOMR assumes that
the oil price follows the futures market curve at the time of publication; this is then used for the demand
projection, and supply is assumed to follow, with OPEC filling the gap between field-by-field projections of
non-OPEC supply and demand. WEO determines the equilibrium price that brings supply and demand in
balance. However, to avoid generating investment/price cycles which would obscure policy effects and long
term trends, this equilibrium is performed as a trend and not year-by-year.

53
 WEO relies on the field-by-field analysis of MTOMR to guide production by country in the first five years of
the projection period. The country by country methodology is also extended to OPEC countries, so OPEC is
not treated as the swing producer, though constraints thought to represent possible OPEC policies are
incorporated in the WEM oil supply module.

Results are also often presented slightly differently in the two reports. Conventional and unconventional oil
may be grouped differently with WEO including all of Canadian oil sands and Venezuelan Orinoco production
in unconventional, while MTOMR generally counts only upgraded bitumen or extra-heavy oil as
unconventional.

In analysing and projecting oil demand, WEO and MTOMR have methodological differences. Since WEO is
concerned with projections of supply and demand of all energy sources and projects a world energy balance
in the future, it incorporates all demand components. Due to the nature of these components, they can be
with a plus or a minus sign (i.e. increasing or decreasing the demand figure). Therefore, while WEO
incorporates statistical differences and refinery transformation losses into historical demand values and
projects those into the future, MTOMR’s demand definition does not include these two categories in its
historical values and projections.

WEO also splits biofuels from historical oil demand and projects oil demand and biofuels demand separately.
OMR does not separate biofuels from the historical oil demand, and the oil demand is projected with a mix
of biofuels. As a result, one barrel of oil from MTOMR projections has lower energy content than that of
WEO if biofuels are projected to grow. A direct comparison of WEO and OMR results is thus only possible if
biofuels are stripped off MTOMR values of oil demand.

The differences in refining mainly concern the interpretation of installed capacity. WEO discounts most of
idled capacity of Chinese teapot and smaller refineries that run below 30% utilization rates. It also discards
the mothballed capacity in entirety, even if the owner of the refinery has announced that it is a temporary
economic shutdown. MTOMR and WEO may also differ in their projection of firm capacity additions within
the same timeframe.

Each country’s projected oil production profile is made of six components. Conventional fields also distinguished
by water depth (onshore, shallow offshore [water depth less than 400 metres], deepwater fields [between 400-
2 000 metres] and ultra-deepwater fields [greater than 2 000 metres]). For unconventional oil, extra-heavy oil
and bitumen is also distinguished by mining or in situ technologies and tight oil by play productivity.
 Production from currently producing fields as of an estimated end-2020: the projected decline rates in
each country are derived from the analysis summarised in Box 4;
 Production from discovered fields with sanctioned, planned and announced developments;
 Production from discovered fields awaiting development;
 Production from fields yet to be discovered;
 Production of natural gas liquids; and
 Production of unconventional oil.

Trends in oil production are modelled using a bottom-up methodology, making extensive use of our database of
worldwide ultimately technically recoverable resources. The methodology aims to replicate investment decisions
in the oil industry by analysing the profitability of developing reserves at the project level (Figure 22).

54
Figure 22: Structure of the oil supply module

Global
Deman Production
Projected
production brought
required to be on stream per
Existing brought on country, within
Producti stream logistical and URR
constraints

Decline
Rates
NPV for
developing Ranking
Costs production in of NPVs
country X

In the WEM oil supply module, production in each country or group of countries is separately derived, according
to the type of asset in which investments are made: existing fields, new fields and non-conventional projects.
Standard production profiles are applied to derive the production trend for existing fields and for those new
fields (by country and type of field) which are brought into production over the projection period.

The profitability of each type of project is based on assumptions about the capital and operating costs of different
types of projects, and the discount rate, representing the cost of capital. The net present value of the cash flows
of each type of project is derived from a standard production profile. Projects are prioritised by their net present
value and the most potentially profitable projects are developed. Constraints on how fast projects can be
developed and how fast production can grow in a given country are also applied. These are derived from
historical data and industry inputs. When demand cannot be met without relaxing the constraints, this signals
that oil prices need to be increased.

6.1.1 US tight oil model

A tight oil module is part of WEM since WEO-2016 and it explores the sensitivity of production of tight oil in the
United States to changes in price and resource availability. The module projects possible future production across
23 shale plays taking into account the estimated ultimate recovery (EUR), initial production, rate of decline and
drilling costs of wells drilled and completed across different areas of each play. Existing production is modelled
by estimating decline parameters of wells based on latest production information available, and the time when
these wells were completed.

Price dynamics affect the number of rigs that are available to drill new wells, with a lag between increases in
prices and increases in the number of rigs operating (as observed empirically). Technology increases both the
speed at which new wells can be drilled and completed (the number of wells per rig) and the amount of
production from each well (the EUR/well). Conversely, the EUR/well of a given area in a given play is assumed to
degrade as that area is depleted over time.

55
Box 4: Methodology to account for production decline in oil and gas fields

The World Energy Outlook has previously presented analyses of decline rates in oil fields on a number of
occasions, based on looking at actual production data time series for a large number of fields. The outcome
of this work is a value for observed decline rates by type of field, geographical location and phase of decline,
as well as an estimate for the difference between observed decline rates and natural decline rates (the
decline rate that would be observed in the absence of further investment in producing fields).
In principle this provides the elements to project the future production of all fields in decline among the set
of fields used. The methodology could be as follows:

 For each field in the database, assign a type (super-giant, giant… onshore, offshore, deepwater)
and determine the current decline phase.
 Project future production for each field as per corresponding decline rate provided in WEO-2013,
updating decline rates as the field changes phase.
But this does not allow the projection of world production from all currently producing fields, as one also
needs to project production from fields currently ramping up (i.e. one needs to know their future peak year
and peak production) and from declining fields not in the database. This is done using a proprietary
commercial database that contains a representation of possible future production for all fields in the world.
Based on this more complete data set, the WEM oil supply module uses a country-by-country
parameterisation of natural decline rates (for each resources type) and a production profile for resources
developed in each country during the projection period (i.e. resources developed in a given year then
provide a ramping-up of production, followed by peak and decline). As shown in Figure 23, this
parameterization gives a good match with the results of the proprietary database (as the two databases
have slightly different base productions, both are normalized to allow a clearer comparison of decline) for
the long term decline; in the short term, the IEA field-by-field analysis (coming from the Medium Term Oil
Market Report) is more conservative that the commercial database, as it accounts for expected field
maintenance and weather disruptions.

Figure 23: Evolution of production of currently producing conventional oil fields from a field-by-
field database and from the WEM
120%

100%

80%

60% Field Database

WEO
40%

20%

0%
2000 2005 2010 2015 2020 2025 2030 2035
Source: Rystad Energy AS, IEA analysis and databases.

56
Rigs are distributed across plays based on current activity, and the expected cost effectiveness of new wells that
are drilled. It is assumed that while operators would aim to drill only in their most productive areas, some wells
will inevitably be located in regions with lower EUR/well or higher decline rates. The product of numbers rigs,
wells/rig, and production/well then gives the new production that comes online in each play in each month
starting in January 2019. Results from this module are directly fed into WEM for each of the scenarios
implemented.

A similar model was developed for shale gas production in the United States.

6.2 Natural gas

Gas production and trade projections are derived from a hybrid WEM gas supply module involving bottom-up
and top-down approaches. The module has similar inputs, logic and functionality as the oil supply module
described above. However, contrary to oil which is assumed to be freely traded globally, gas is assumed to be
primarily regionally traded, with inter-regional trade constrained by existing or planned pipelines, LNG plants
and long-term contracts. So the module is first run for 20 regions (see Annex 1), for which indigenous production
is modelled on the basis of remaining technically recoverable resources (Table 7) and depletion rates, taking
account of production costs and prices in the region. Subtracting domestic production from demand, in aggregate
for each importing regional block, yields gas import requirements. For each gas net-exporting regional block,
aggregate production is determined by the level of domestic demand and the call on that region’s exportable
production (which is determined by the import needs of the net importing regions and supply costs). Long term
contracts (current, or assumed for the future) are served first, then exporting regions compete on the basis of
marginal production costs plus transport costs, within the current and assumed future LNG and pipeline
capacities. This provides an inter-block gas trade matrix. The effects of pricing policies (current or assumed for
the future) of exporting regions can also be taken into account.

Production within each region is allocated to individual countries according to remaining technically recoverable
resources, depletion rates and relative supply costs, with a logic similar to that of the oil supply module, but with
“demand” being provided by the respective regional production derived in the previous step.

6.3 Coal
The coal module is a combination of a resources approach (Table 7) and an assessment of the development of
domestic and international markets, based on the international coal price. Production, imports and exports are
based on coal demand projections and historical data, on a country basis. Four markets are considered: coking
coal, steam coal, lignite and peat. World coal trade, principally constituted of coking coal and steam coal, is
separately modelled for the two markets and balanced on an annual basis.

57
Table 7: Remaining technically recoverable fossil fuel resources, end-2019

Oil Proven Resources Conventional Tight NGLs EHOB Kerogen

(billion barrels) reserves crude oil oil oil

North America 238 2 422 241 218 163 800 1 000

Central and South America 850 245

293 60 50 494 3

Europe 15 115 59 19 28 3 6

Africa 126 449 308 54 85 2 -

Middle East 834 1 136 903 29 161 14 30

Eurasia 146 951 237 85 59 552 18

Asia Pacific 51 284 126 72 67 3 16

World 1 702 6 208 2 118 536 613 1 868 1 073

Natural gas Proven Resources Conventional Tight Shale Coalbed

(trillion cubic metres) reserves gas gas gas methane

North America 16 149 50 10 81 7

Central and South America 28

8 84 15 41 -

Europe 5 47 19 5 18 5

Africa 19 101 51 10 40 0

Middle East 81 121 102 9 11 -

Eurasia 77 169 132 10 10 17

Asia Pacific 22 139 45 21 53 20

World 229 810 426 80 254 49

Coal Proven Resources Coking Steam Lignite

(billion tonnes) reserves coal coal

North America 257 8 389 1 032 5 839 1 519

Central and South America 14 60 3 32 25

Europe 135 979 165 411 403

Africa 15 343 45 297 0

Middle East 1 41 19 23 -

Eurasia 191 2 015 343 1 041 632

Asia Pacific 457 8 954 1 508 6 032 1 413

World 1 070 20 781 3 114 13 675 3 992

Notes: NGLs = natural gas liquids; EHOB = extra-heavy oil and bitumen. The breakdown of coal resources by type is an IEA estimate. Coal
world resources exclude Antarctica.
Source: IEA WEO-2020.

58
 6.4 Bioenergy
Bioenergy is an important renewable energy option in all of its forms: solid (biomass), liquid (biofuels) and gas
(biogas and biomethane). Bioenergy provides a significant portion of renewables-based electricity and transport
fuels in all scenarios in the WEO and as gas can also contribute to decarbonise the gas network. Many regions or
countries have or are considering policies that will increase the demand for bioenergy in the power and transport
sectors further in the future.

The Bioenergy supply module, part of WEM since WEO-2012, is designed to assess the ability of WEO regions to
meet their demand for bioenergy for power generation and biofuels with domestic resources.16 Where they are
not able to do so, the module also simulates the international trade of solid biomass and biofuels. The availability
of bioenergy is restricted to renewable sources of biomass feedstock that is not in competition with food.

The Biogas and biomethane supply module, added to the WEM for WEO-2019, is designed to assess the
sustainable technical potential and costs of biogas and biomethane for all the WEO regions. This analysis includes
feedstocks that can be processed with existing technologies, that do not compete with food for agricultural land,
and that do not have any other adverse sustainability impacts (e.g. reducing biodiversity). Feedstocks grown
specifically to produce biogas, such as energy crops, are also excluded. This module excludes international trades
of biogas and biomethane.

6.4.1 Bioenergy supply module

Biomass supply potentials by region

The feedstock supply potentials are built on a wide range of data related to land, crops and food demand,
originating largely from the database of the Food and Agriculture Organization of the United Nations (FAO), as
well as academic literature and the Global Agro-Ecological Zones (GAEZ) system, a collaborative project involving
FAO and the Institute for Applied Systems Analysis (IIASA).

Total supply potentials by region in the bioenergy supply module are the sum of the potential supply for four
categories of feedstocks: forestry products, forestry residues, agricultural residues and energy crops (Figure 24).
Starting from current activity levels, ramping up collection and delivery of these often diffuse feedstocks requires
significant lead times before maximum potential supply levels can be reached. The potential supply of forestry
and agricultural residues is reduced by industrial and residential use to produce heat, as well as demand for
traditional uses.

Figure 24: Schematic of biomass supply potentials

Total biomass feedstock

supply potential

Forestry Forestry Agricultural Energy

products residues residues crops

Additional
Forestry Agricultural Available Crop
forest Usable share Usable share
activities yields arable land choice/yield
growth

16 The module does not assess demand or supply related to biogas or waste.

59
Forestry products include only forestry activities, such as harvesting trees and complementary fellings, for the
primary purpose of producing power or transport biofuels. The maximum potential availability of forestry
products is limited to the expected growth in total forest area per year, after other forestry demands are met, in
each region, thereby avoiding direct deforestation.

Forestry residues are those materials, or secondary products, produced from forestry activities where the
primary motivation is something other than to produce bioenergy. These include forestry scraps, bark leftover
from the timber industry, industrial by-products and waste wood. The maximum potential availability is limited
by the level of the related activities and the usable share of the leftover materials.

Agricultural residues are the leftover materials after harvesting crops, such as corn stover, straw and bagasse
from sugarcane processing. Data for harvests by region include the following crops: barley, maize (corn), oats,
rice, sorghum, wheat, other cereals, rapeseed, soybeans, sunflower seed, and sugarcane. The maximum
potential availability is limited by the amount of crops harvested and by the recoverable share of the residues. It
is important for a portion of the residues to remain in fields to replenish soil nutrients and maintain yields for
future harvests, by helping reduce soil erosion and maintaining water and temperature in the soils. The
percentage of these residues that can be made available for energy production in a sustainable manner is region-
and crop-specific, and is still being investigated actively.
Energy crops are those grown specifically for energy purposes, including sugar and starch feedstock for ethanol
(e.g. corn, sugarcane, and sugar beet), vegetable-oil feedstock for biodiesel (e.g. rapeseed, soybean and oil palm
fruit) and lignocellulosic material (e.g. switchgrass, poplar and miscanthus) for advanced biofuels. The maximum
potential availability is determined by the available arable land, after taking into account food-related demand
for land, crop choice and rising yields over time.

The potential supply from energy crops (million tonnes) is calculated as follows:

𝑃𝑡,𝑟 = ∑(𝑥𝑡,𝑟,𝑙,𝑔,𝑐 ∗ 𝑦𝑡,𝑟,𝑙,𝑔,𝑐 ∗ 𝑠𝑡,𝑟,𝑐 )

𝑙,𝑔,𝑐

where, for a given year t and region r,

- 𝑃𝑡,𝑟 is the potential biomass feedstock supply from energy crops;
- 𝑥𝑡,𝑟,𝑙,𝑔,𝑐 is the available land by type l, grade g, and crop c;
- 𝑦𝑡,𝑟,𝑙,𝑔,𝑐 is the crop yield; and
- 𝑠𝑡,𝑟,𝑐 is the share of available land for each crop.

Available land is divided into three grades of land quality (prime, good and marginal) and three types of land
(cultivated, unprotected grassland and unprotected forest land). 17 Lower quality grades of land provide lower
crop yields. In this assessment, unprotected forest land is not allowed to be converted to crop lands and so is
unavailable for bioenergy purposes. Crop yields are defined by region, reflecting the average growing conditions
in a region, and are assumed to continue to improve moderately through 2035. Crop choice is influenced by
currently favoured crops for bioenergy, the changing economics of feedstock (through increased yields and
relative attractiveness compared to the fossil fuel alternative), and policy development. For example, policy goals
for advanced biofuels will increase demand for lignocellulosic energy crops, decreasing the share of land devoted
to conventional feedstock.

17
Classifications and data from Fischer et al. (2011), Scarcity and abundance of land resources: competing uses and the
shrinking land resource base. SOLAW Background Thematic Report - TR02.

60
Supply to meet demand
Demand for biomass feedstock is based on demand projections in the WEO for both the power and transport
sectors (demand for other sectors is assumed to be met from domestic resources). To meet demand, domestic
supplies are given priority; the remainder is covered through international markets. The model is calibrated to
meet existing trade flows reported in a range of industry reports, including the F.O. Licht series “World Ethanol
& Biofuel Report”, and government reports, such as regional Global Agricultural Information Network (GAIN)
reports on biofuels by the US Department of Agriculture.

Domestic supply
Biomass feedstock competes to meet demand on the basis of conversion costs, including feedstock prices and
the energy contents of feedstock. Several biomass feedstock types can be used for both power generation and
the production of biofuels. These include forestry products, forestry residues and agricultural residues. Where
this is the case, the net present values for both uses are compared and ranked, based on technology cost data
from WEM and IEA’s Mobility Model. According to rank, available biomass feedstock supplies are allocated.
Domestic supply of biofuels is limited by refining capacity. In the near term, this is restricted by existing refineries
and those already under construction or planned.

Global trade
The model uses a global trade matrix to match unsatisfied demand with available supply on a least-cost basis,
including transportation costs. Transportation costs between regions include both average over-land and by-sea
costs. Three products are traded: ethanol, biodiesel and solid biomass pellets. The latter are high-density uniform
products that can be made from residues and other feedstock, and their uniformity and density make handling
and transportation easier and less expensive over long distances compared with other bioenergy resources. The
conversion of biomass feedstock to biofuels occurs in the exporting region, therefore conversion costs are
calculated based on the technology costs in the exporting region. Importing regions choose suppliers based on
least-cost available supplies (including transportation costs). Exporting regions make supplies available to
importing regions willing to pay the highest price.

6.4.2 Biogas and biomethane supply module

Biogas and biomethane supply potential has been assessed considering a wide variety of feedstock, grouped in
six categories: crop residues, animal manure, municipal solid wastes (MSW), forest product residues, wastewater
and industrial wastes.
The feedstock supply potentials are built on a wide range of data originating largely from the Food and
Agriculture Organization of the United Nations (FAO) database and OECD-FAO study (OECD/FAO, 2018) for
wheat, maize, rice, other coarse grains, sugar beet, sugar cane, soybean, and other oilseeds, cattle, pig, poultry
and sheep, log felling residues, wood processing residues and distiller dried grains (DDGs), a by-product of
ethanol production from grains and from a World Bank study (World Bank, 2018) for different categories of
organic municipal solid waste such as food and green waste, paper and cardboard, and wood. Wastewater
includes only municipal wastewater and is based on the output data from the Water module developed by the
World Energy Outlook team.
Biogas is produced by anaerobic digestion. Five technologies of centralised biogas production plants are
modelled: landfill gas recovery system, digester in municipal wastewater treatment plant and three centralised
co-digestion plants (small-, medium and large-scale). In addition, two types of household-scale digester are
modelled in the residential sector of the World Energy Model, to account for rural and decentralised biogas
production in rural areas of developing economies.

61
For biomethane, two production pathways are considered: upgrading of biogas produced by anaerobic digestion
and thermal gasification and methanation of lignocellulosic biomass.

For each technology technical and economic parameters, e.g. efficiency, lifetime, overnight capital cost or
operational costs are collected to assess the production costs.

The combination of the assessment of the supply potential and the economic evaluation of the different biogas
and biomethane processes were used to assess biogas and biomethane supply cost curves. For a given year, it is
made of the aggregation of biomethane potential and associated levelised cost of production for every region,
feedstock and technology. Information provided by supply curves is then used to assess the cost-competitiveness
of the two main uses of biogas and biomethane: electricity and heat generation and injection in the gas grid.
Supply curves are used to calculate GHG emissions potential savings and related abatement cost to understand
the future role of carbon pricing on biogas and biomethane development.

7 Emissions
7.1 CO2 emissions
As energy-related CO2 emissions account for the lion's share of global greenhouse gas emissions, one of the
important outputs of the WEM is region by region CO2 emissions from fuel combustion. For each WEM region,
sector and fuel, CO2 emissions from fuel combustion are calculated by multiplying energy demand by an implied
CO2 content factor. The implied CO2 content factors for coal, oil and gas differ between sectors and regions,
reflecting the product mix and efficiency. They have been calculated as an average of the past three years from
IEA energy-related sectoral approach CO2 data for all WEM regions and are assumed to remain constant over the
projection period (IEA, 2017d and 2018b).

7.2 Non-CO2 greenhouse gases and CO2 process emissions

For the WEO Special Report Energy and Climate Change, a detailed analysis of process-related CO2 emissions
from various industrial sources by WEM region was conducted. For the estimation a Tier 1 or Tier 2 method has
been used, which in general means that emissions have been estimated based on the production of industrial
materials and an emissions factor from the 2006 IPCC Guidelines for National Greenhouse Gas Inventories. So
far the analysis is limited to the most important sources of industrial process emissions:
 Mineral industry: cement, lime, limestone use, soda ash use
 Metal industry: primary aluminium
 Chemical industry: ammonia, methanol, ethylene, soda ash
 Non-energy products: lubricants and paraffins
 Transformation: coal-to-liquids, coal-to-gas and gas-to-liquids.

The WEM models all energy-related GHG emissions, both CO2 and non-CO2. Non-CO2 emissions originating from
non-energy sectors rely on the scenarios from the IPCC 5th Assessment Report scenario database. This database
contains projections of non-CO2 emissions over the 21st century under a wide range of scenarios. The CO 2 and
non-CO2 emissions modelled within WEM are benchmarked against scenarios from this database to provide
commensurate projections for all other GHGs. This includes projections for other methane emissions, nitrous
oxide (N2O), and F-gases. The last category includes hydrofluorocarbons (HFCs), perfluorocarbons (PFCs) and
sulphur hexafluoride (SF6) from several sectors, mainly industry. Where needed, the trajectory of CO 2 emissions
from land-use change and forestry (LULUCF) are again taken from scenario databases provided by the IPCC.

62
 7.3 Air pollution
Emissions of major air pollutants resulting from the WEO energy scenarios have been estimated in co-operation
with the International Institute for Applied Systems Analysis (IIASA). Using the IIASA GAINS model, estimates
have been made for the following local air pollutants: sulphur dioxide (SO2), nitrogen oxides (NOx) and PM2.5.18
More information can be found in the WEO Special Report on Energy and Air Pollution as well as in a previous
detailed report outlining the approach, results and information about health impacts, as well as pollution control
costs.

7.4 Oil and gas methane emissions model

7.4.1 Global estimate of methane emissions from oil and gas operations
Our approach to estimating methane emissions from global oil and gas operations relies on generating country-
specific and production type-specific emission intensities that are applied to production and consumption data
on a country-by-country basis. Our starting point is to generate emission intensities for upstream and
downstream oil and gas in the United States (

Table 8). The 2017 US Greenhouse Gas Inventory is used for this along with a range of other data sources,
including our survey of companies and countries. The hydrocarbon-, segment- and production-specific emission
intensities are then further segregated into fugitive, vented and incomplete flaring emissions to give a total of
19 separate emission intensities.

Table 8: Categories of emission sources and emissions intensities in the United States
Intensity (mass CH4/mass
Hydrocarbon Segment Production Type Emissions Type oil or gas)
Oil Upstream Onshore conventional Vented 0.35%
Oil Upstream Onshore conventional Fugitive 0.03%
Oil Upstream Offshore Vented 0.16%
Oil Upstream Offshore Fugitive 0.01%
Oil Upstream Unconventional oil Vented 0.53%
Oil Upstream Unconventional oil Fugitive 0.04%
Oil Downstream Vented 0.003%
Oil Downstream Fugitive 0.001%
Oil Onshore conventional Incomplete-flare
Oil Offshore Incomplete-flare
Oil Unconventional gas Incomplete-flare
Gas Upstream Onshore conventional Vented 0.35%
Gas Upstream Onshore conventional Fugitive 0.20%
Gas Upstream Offshore Vented 0.12%
Gas Upstream Offshore Fugitive 0.07%
Gas Upstream Unconventional gas Vented 0.60%
Gas Upstream Unconventional gas Fugitive 0.34%
Gas Downstream Vented 0.12%
Gas Downstream Fugitive 0.23%
Note: Emission intensities are given here as the energy ratio of tonne of oil-equivalent (toe) methane emitted to tonne of oil equivalent fuel.
For natural gas, since this is not 100% methane, ratios differ slightly if given as the volume ratio or mass ratio. For example, the emission
intensity for upstream onshore conventional vented emissions in the United States if stated as a volume ratio is 0.36% rather than 0.38% as
shown for the energy ratio.

18Fine particulate matter is particulate matter that is 2.5 micrometres in diameter and less; it is also known as PM2.5 or
respirable particles because they penetrate the respiratory system further than larger particles.

63
The US emissions intensities are then scaled to provide emission intensities in all other countries. This scaling is
based upon a range of auxiliary country-specific data. For the upstream emission intensities, the scaling is based
on the age of infrastructure and types of operator within each country (namely international oil companies,
independent companies or national oil companies). For downstream emission intensities, country-specific scaling
factors were based upon the extent of oil and gas pipeline networks and oil refining capacity and utilisation. The
strength of regulation and oversight, incorporating government effectiveness, regulatory quality and the rule of
law as given by the Worldwide Governance Indicators compiled by the World Bank (2017), affects the scaling of
all intensities. Some adjustments were made to the scaling factors in a limited number of countries to take into
account other data that were made available (where this was considered to be sufficiently robust).

Table 9 provides the resultant scaling factors in the top oil and gas producers (the countries listed cover 95% of
global oil and gas production). These scaling factors are directly used to modify the emissions intensities in

Table 8. For example, the vented emission intensity of onshore conventional gas production in Russia is taken as
0.38%×1.9 = 0.72%. These intensities are finally applied to the production (for upstream emissions) or
consumption (for downstream emissions) of oil and gas within each country.
Table 9: Scaling factors applied to emission intensities in the United States
Oil and gas
Oil Gas
Country production in 2019
mtoe Upstream Downstream Upstream Downstream
United States 1404 1.0 1.0 1.0 1.0
Russia 1165 1.9 1.6 1.6 2.0
Saudi Arabia 661 1.5 1.0 1.0 1.5
Canada 422 0.9 0.7 0.7 1.0
Iran 353 2.9 2.8 2.8 2.7
China 335 1.7 1.4 1.4 1.6
Iraq 261 6.0 7.5 7.5 6.2
United Arab Emirates 251 1.6 0.8 0.8 1.5
Qatar 230 1.3 1.0 1.0 1.5
Norway 192 0.1 0.0 0.0 0.0
Kuwait 166 1.8 1.4 1.4 1.7
Brazil 160 1.9 1.7 1.7 1.8
Algeria 148 2.8 2.8 2.8 2.7
Nigeria 138 2.5 2.7 2.7 2.6
Mexico 131 1.6 1.2 1.2 1.6
Kazakhstan 129 2.0 1.4 1.4 1.9
Australia 116 1.0 0.6 0.6 0.8
Indonesia 102 1.7 1.4 1.4 1.6
Malaysia 94 1.2 0.9 0.9 1.1
United Kingdom 88 1.1 0.7 0.7 1.0
Egypt 86 2.2 2.2 2.2 2.1
Oman 83 1.6 1.2 1.2 1.6
Venezuela 80 9.6 11.8 11.8 9.6
Turkmenistan 79 9.1 11.0 11.0 8.9
Angola 76 2.6 3.0 3.0 3.2
United States 1404 1.0 1.0 1.0 1.0
Russia 1165 1.9 1.6 1.6 2.0
Saudi Arabia 661 1.5 1.0 1.0 1.5
Canada 422 0.9 0.7 0.7 1.0
Iran 353 2.9 2.8 2.8 2.7
China 335 1.7 1.4 1.4 1.6
Iraq 261 6.0 7.5 7.5 6.2
United Arab Emirates 251 1.6 0.8 0.8 1.5
Qatar 230 1.3 1.0 1.0 1.5

64
7.4.2 Marginal abatement cost curves
To construct the marginal abatement cost curves presented in WEO-2018, the 19 emissions sources listed in

Table 8 were further separated into 86 equipment-specific emissions sources (Table 10). The allocation of
emissions from each of the 19 emissions sources to these 86 equipment-specific sources was generally based on
proportions from the United States. However a number of modifications were made for countries based on other
data sources and discussions with relevant stakeholders. Some of the largest changes made were for the
proportion of emissions from: pneumatic controllers (which are less prevalent in many countries outside North
America), LNG liquefaction (which were assumed to be larger in LNG exporting countries), and associated gas
venting.

Table 10: Equipment-specific emissions sources used in the marginal abatement cost curves
Equipment source Hydrocarbon Segment
Large Tanks w/Flares Oil Upstream
Large Tanks w/VRU Oil Upstream
Large Tanks w/o Control Oil Upstream
Small Tanks w/Flares Oil Upstream
Small Tanks w/o Flares Oil Upstream
Malfunctioning Separator Dump Valves Oil Upstream
Pneumatic Devices, High Bleed Oil Upstream
Pneumatic Devices, Low Bleed Oil Upstream
Pneumatic Devices, Int Bleed Oil Upstream
Chemical Injection Pumps Oil Upstream
Vessel Blowdowns Oil Upstream
Compressor Blowdowns Oil Upstream
Compressor Starts Oil Upstream
Associated Gas Venting Oil Upstream
Well Completion Venting (less HF Completions) Oil Upstream
Well Workovers Oil Upstream
HF Well Completions, Uncontrolled Oil Upstream
HF Well Completions, Controlled Oil Upstream
Pipeline Pigging Oil Upstream
Tanks Oil Downstream
Truck Loading Oil Downstream
Marine Loading Oil Downstream
Rail Loading Oil Downstream
Pump Station Maintenance Oil Downstream
Pipeling Pigging Oil Downstream
Uncontrolled Blowdowns Oil Downstream
Asphalt Blowing Oil Downstream
Process Vents Oil Downstream
CEMS Oil Downstream
Production Compressor Vented Gas Upstream
Gas Well Completions without Hydraulic Fracturing Gas Upstream
Gas Well Workovers without Hydraulic Fracturing Gas Upstream
Hydraulic Fracturing Completions and Workovers that vent Gas Upstream
Hydraulic Fracturing Completions and Workovers with RECs Gas Upstream
Well Drilling Gas Upstream
Pneumatic Device Vents (Low Bleed) Gas Upstream
Pneumatic Device Vents (High Bleed) Gas Upstream
Pneumatic Device Vents (Intermittent Bleed) Gas Upstream
Chemical Injection Pumps Gas Upstream
Kimray Pumps Gas Upstream

65
 Equipment source Hydrocarbon Segment
Dehydrator Vents Gas Upstream
Large Tanks w/VRU Gas Upstream
Large Tanks w/o Control Gas Upstream
Small Tanks w/o Flares Gas Upstream
Malfunctioning Separator Dump Valves Gas Upstream
Gas Engines Gas Upstream
Well Clean Ups (LP Gas Wells) - Vent Using Plungers Gas Upstream
Well Clean Ups (LP Gas Wells) - Vent Without Using Plungers Gas Upstream
Vessel BD Gas Upstream
Pipeline BD Gas Upstream
Compressor BD Gas Upstream
Compressor Starts Gas Upstream
G&B Station Episodic Events Gas Upstream
Pressure Relief Valves Gas Upstream
Mishaps Gas Upstream
Recip. Compressors Gas Upstream
Centrifugal Compressors (wet seals) Gas Upstream
Centrifugal Compressors (dry seals) Gas Upstream
Dehydrators Gas Upstream
AGR Vents Gas Upstream
Pneumatic Devices Gas Upstream
Blowdowns/Venting Gas Upstream
Reciprocating Compressor Gas Downstream
Centrifugal Compressor (wet seals) Gas Downstream
Centrifugal Compressor (dry seals) Gas Downstream
Reciprocating Compressor Gas Downstream
Dehydrator vents (Transmission) Gas Downstream
Dehydrator vents (Storage) Gas Downstream
Pneumatic Devices (High Bleed) Gas Downstream
Pneumatic Devices (Intermittent Bleed) Gas Downstream
Pneumatic Devices (Low Bleed) Gas Downstream
Pneumatic Devices (High Bleed) Gas Downstream
Pneumatic Devices (Intermittent Bleed) Gas Downstream
Pneumatic Devices (Low Bleed) Gas Downstream
Pipeline venting Gas Downstream
Station Venting Transmission Gas Downstream
Station Venting Storage Gas Downstream
LNG Reciprocating Compressors Vented Gas Downstream
LNG Centrifugal Compressors Vented Gas Downstream
LNG Station venting Gas Downstream
LNG Reciprocating Compressors Vented Gas Downstream
LNG Centrifugal Compressors Vented Gas Downstream
LNG Station venting Gas Downstream
Pressure Relief Valve Releases Gas Downstream
Pipeline Blowdown Gas Downstream
Mishaps (Dig-ins) Gas Downstream

The abatement options included in the marginal abatement cost curves to reduce emissions from these sources
are listed in Table 11. We are unable to provide the specific costs and applicability factors for these as it is based
on proprietary information gathered by ICF (although see ICF (2016a) and ICF (2016b) for data that has made
available publically). Costs were again based upon information from the United States. However labour costs,
whether the equipment is imported or manufactured domestically (which impacts the capital costs and whether
or not import taxes are levied), and capital costs were modified based on country-specific or region-specific

66
information. Similarly the applicability factors are modified based on other data that is available publically (for
example that solar-powered electric pumps cannot be deployed as widely in high-latitude countries).

Leak detection and repair (LDAR) programmes are the key mechanism to mitigate fugitive emissions from the
production, transmission or distribution segments of the value chain. The costs of inspection differ depending
on the segment in question since it takes longer to inspect a compressor on a transmission pipeline than in a
production facility. It is assumed that inspections can be carried out annually, twice a year, quarterly or monthly,
with each option included as a separate mitigation option in the marginal abatement cost curves. Annual
inspections are assumed to mitigate 40% of fugitive emissions, biannual inspections mitigate an additional 20%,
quarterly inspections mitigate an additional 10%, and monthly inspections mitigate an additional 5%.
Implementing a monthly LDAR programme therefore reduces fugitive emissions by 85%; the remaining 15%
cannot be avoided. As the frequency of implementing each programme increases, so does the cost per unit of
methane saved. For example, while the incremental cost of a biannual inspection programme is the same as that
of an annual inspection, the incremental volume of methane saved is lower (20% rather than 40%). Nevertheless,
LDAR programmes remain some of the most cost-effective mitigation options available, i.e. they tend to comprise
a large proportion of the positive net present value options in countries.

Table 11: Abatement options for methane emissions from oil and gas operations
Abatement option
Blowdown Capture and Route to Fuel System (per Compressor)
Blowdown Capture and Route to Fuel System (per Plant)
Early replacement of high-bleed devices with low-bleed devices
Early replacement of intermittent-bleed devices with low-bleed devices
Install Flares-Completion
Install Flares-Portable
Install Flares-Portable Completions Workovers WO HF
Install Flares-Portable WO Plunger Lifts
Install Flares-Stranded Gas Venting
Install Flares-Venting
Install New Methane Reducing Catalyst in Engine
Install Non Mechanical Vapor Recovery Unit
Install Plunger Lift Systems in Gas Wells
Install small flare
Install Vapor Recovery Units
LDAR Gathering
LDAR LDC - Large
LDAR LDC - MRR
LDAR Processing
LDAR Reciprocating Compressor Non-seal
LDAR Transmission
LDAR Wells
Mechanical Pumping for Liquids Unloading
Pipeline Pump-Down Before Maintenance
Redesign Blowdown Systems and Alter ESD Practices
Reduced Emission Completion
Replace Kimray Pumps with Electric Pumps
Replace Pneumatic Chemical Injection Pumps with Electric Pumps
Replace Pneumatic Chemical Injection Pumps with Solar Electric Pumps
Replace with Instrument Air Systems
Replace with Electric Motor
Replace with Servo Motors
Replace with Solenoid Controls
Replacement of Reciprocating Compressor Rod Packing Systems
Route to existing flare - Large Dehydrators

67
 Abatement option
Route to existing flare - Large Tanks
Route to flare - Small Dehydrators
Route to existing flare - Small Tanks
Route Vent Vapors to tank
Wet Seal Degassing Recovery System for Centrifugal Compressors
Wet Seal Retrofit to Dry Seal Compressor
Microturbine
Mini-LNG
Mini-GTL
Mini-CNG

7.4.3 Well-head prices used in net present value calculation

Since natural gas is a valuable product, the methane that is recovered can often be sold. This means that
deploying certain abatement technologies can result in overall savings if the net value received for the methane
sold is greater than the cost of the technology. Well-head prices are used in each country to determine the value
of the methane captured. As described in WEO-2019, the marginal abatement cost curves examine this issue
from a global, societal perspective. The credit obtained for selling the gas is therefore applied regardless of the
contractual arrangements necessary and the prices assume that there are no domestic consumption subsidies
(as the gas could be sold on the international market at a greater price). The well-head gas prices used could
therefore be substantially different from subsidised domestic gas prices.

The natural gas import prices listed in WEO-2019 (IEA, 2019b) are the starting point for the well-head prices
within each country. To estimate well-head prices over time, each country is assigned to be either an importer
or an exporter based on the trends seen in the Stated Policies Scenario. For importing countries, any gas that
would be saved from avoiding leaks would displace imports. The well-head price is therefore taken as the import
price minus the cost of local transport and various taxes that may be levied (assumed to be around 15% of the
import price). For exporting countries, the relevant well-head price is taken as the import price in their largest
export market net-backed to the emissions source. For the net-back, allowance is made for transport costs
(including liquefaction and shipping or pipeline transport), fees and taxes. For example, in Russia the export price
is taken as the import price in Europe ($4.9/MBtu in 2016). Export taxes of 40% are then subtracted along with
a further $0.5/MBtu to cover the cost of transport by pipeline. This gives a well-head gas price in Russia in 2016
of $2.4/MBtu. In the United States and Canada, the well-head price is taken as the Henry Hub price minus 15%
(to cover the cost of local transportation and fees).
The costs and revenue for each technology or abatement measure is converted into net present value using a
discount rate of 10% and divided by the volume of emissions saved to give the cost in dollars per million British
thermal units (MBtu).

7.4.4 Other notes on marginal abatement cost curves

To aid visualisation of the marginal abetment cost curves, the costs and savings from multiple technologies are
aggregated together. Within each country, the abatement options that could be applied to each of the 19
emission sources listed in

Table 8 are aggregated into three cost steps. These steps roughly represent the cheapest 50% of reductions, the
next 30% of reductions and the final of 20% reductions.

68
7.4.5 Methane Tracker Update
There are several emerging technologies and approaches to measurement that appear promising to elevate data
available on oil and gas methane emissions—among them are satellites and other aerial detection instruments
utilised during measurement campaigns.

Confirming and reconciling bottom-up estimates with direct emissions measurements, via aerial instruments or
otherwise, is the best option to hone accurate emissions information and overcome shortcomings associated
with any single approach. In this regard, using stationary monitors, ground vehicles, or aerial instruments such
as satellites, drones, and planes, can reduce the risk that bottom-up estimates significantly underestimate
emissions from a site. In order to have the greatest impact on improving estimation techniques, site-level studies
should be sufficiently representative geographically and temporally, publicly reported, and independently
verified.

The IEA has stayed abreast of instances of these emerging measurement strategies and worked to integrate
results from credible sources into the Methane Tracker where data has become available. The 2020 Tracker
update reflects major downward revisions to emissions levels in a handful of jurisdictions—notably Norway and
the Netherlands. This was the result of a series of measurement campaigns of methane emissions from oil and
gas production in the North Sea—these efforts yielded improvements to the process of inventorying emissions
and confirmed estimates generated and reported by Norwegian and Dutch industry operators.

8 Investment
8.1 Investment in the energy supply chain
Investment is measured as the ongoing capital expenditures in energy supply capacity. Projections of supply-side
investment requirements by scenario are derived from the WEM energy supply and demand modules, which
incorporate economic analysis of the net change in investment by energy suppliers and energy consumers; the
net change in energy import bills and export revenues; and how the cost to consumers of investing in more
energy-efficient equipment compares with the savings they make through lower expenditure on energy bills. All
investments and consumers’ savings in energy bills are expressed in year-2019 dollars. The calculation of the
investment requirements for power generation and fuel supply involved the following steps for each region:

 New capacity needs for production, transportation and (where appropriate) transformation were
calculated on the basis of projected supply trends, estimated rates of retirement of the existing supply
infrastructure and decline rates for oil and gas production.
 Unit capital cost estimates were compiled for each component in the supply chain. These costs were
then adjusted for each year of the projection period using projected rates of change based on a detailed
analysis of the potential for technology-driven cost reductions and on country-specific factors.
 Incremental capacity needs were multiplied by unit costs to yield the amount of investment needed as
if the assets were constructed and became operational on an overnight basis.
 Finally, using technology and country/region-specific spending profiles, overnight investment needs
were then distributed uniformly across construction lead times estimated for each asset.
The results are presented by decade. The estimates of investment in the current decade take account of projects
that have already been decided and expenditures are already ongoing. The definition of investment changed in
WEO-2019 from previous editions where investment was defined as overnight capital expenditure. This approach
based on capital spending can differ across supply areas. For some sectors, such as power generation, the
investment is spread out evenly from the year in which a new plant or upgrade of an existing one begins its
construction to the year in which it becomes operational. For other sources, such as upstream oil and gas and

69
liquefied natural gas (LNG) projects, investment reflects the capital spending incurred over time as production
from a new source ramps up or to maintain output from an existing asset.

For the purposes of this study, investment is defined as capital expenditure only. It does not include spending
that is usually classified as operation and maintenance.

8.1.1 Short-term oil and natural gas upstream investment

Projections of upstream investment are based on a combination of bottom-up and top-down approaches. The
former involves a detailed analysis of the plans and prospects for oil and gas industry investment in the future,
with the aim of determining how much the industry is planning to invest in response to current prices and to the
need for new capacity and of assessing the resulting additions to production capacity.

This analysis is based on a survey of the capital-spending programmes of 70 of the largest upstream oil and gas
companies (national and international companies and pure exploration and production companies), covering
actual capital spending from 2000 to 2019 and their plans or forecasts of upcoming spending when available.
Companies were selected on the basis of their size as measured by their production and reserves, though
geographical spread and data availability also played a role. The surveyed companies account for around three-
quarters of world oil and gas production. Total industry investment was calculated by adjusting upwards the
spending of the 70 companies, according to their share of world oil and gas production for each year. Data was
obtained from companies’ annual and financial reports, corporate presentations, press reports, trade
publications and direct contacts in the industry.

8.1.2 Long-term investment in fuel supply

Projections of long-term oil, gas, coal and biofuels and biogases supply-side investment requirements are
generated in the respective supply-side modules. The methodology establishes a direct link over time between
new production capacity brought on stream, the cash flow generated and the investments required. The cost of
new capacity is estimated from a set of variables: size of the reserves, degree of depletion, location type of
resource, technology employed, technology learning, and underlying assumptions for cost changes (which are a
function of oil prices in the oil and gas supply-side modules). A more detailed historical estimate was made of
investments associated with carbon capture, utilisation and storage (CCUS), including for natural gas processing
and in power and industry applications (see below), and for projects under advanced development stages.

8.1.3 Power sector investment

Large investments in the power sector will be needed over the Outlook period to meet rising electricity demand
and to replace or refurbish obsolete generating assets and network infrastructure. The overnight investments in
generating assets are a straightforward calculation multiplying the capital cost ($/kW) for each generating
technology by the corresponding capacity additions for each modelled region/country. Investment outlays are
then spread over time based on spending profiles that begin at the start of construction and finish when an asset
becomes operational.
The capital costs assumed in the power generation sector are based on a review of the latest country data
available and on assumptions of their evolution over the projection period. They represent overnight costs for
all technologies. For renewable sources and for plants fitted with carbon capture, utilisation and storage (CCUS)
facilities, the projected investment costs result from the various levels of deployment in the different scenarios.
A spreadsheet outlining the indicative overnight costs and other relevant investment assumptions for all
technologies by region may be found on the WEO model documentation page
(http://www.iea.org/weo/weomodel/). For investment in transmission and distribution networks, please refer
to section 4.3.

70
 8.2 Demand-side investments
Demand-side investments are consumers’ outlays for the purchase of durable goods, that is, end-use equipment.
Ongoing spending associated is assumed to occur in the same year as when assets become operational. The
World Energy Model includes detailed analysis of project procurement capital, i.e. the money spent by end-users
on energy-consuming products. This does not include all of the spending, only the amount that is spent (including
taxes and freight costs) to procure equipment that is more efficient than a baseline. The investment cost includes
labour costs that are directly related to an installation, while additional costs can arise from administrative
procedures, legal protection and border clearances, which are also included in the cost estimate. In other words,
this calculation reflects the additional amount that consumers have to pay for higher energy efficiency over the
projection period.

Across the WEM regions and for each end-use sector (industry, transport and buildings), the investment needed
to move to greater efficiency levels have been analysed. The analysis is based on investment cost, stock turnover
and the economic return required across sub-sectors in industry, across modes of transport and across end-uses
in buildings.
In the road transport sector, the costs of efficiency improvements and of a switch to alternative fuel vehicles are
used as an input to the model to determine each option’s cost-competitiveness. Based on the outcome of this
analysis, the investment needs are then determined by multiplying the number of vehicles sold in each year by
the costs of each vehicle. Detailed figures from Airbus/Boeing were used to assess the need for new planes and
the increased capital cost of increasing efficiency in the aviation sector.

Outputs include the additional annual capital needs for the 26 regions and three end-use sectors. The impact of
the energy savings on consumers’ bills is also analysed. The sectoral end-user prices (including taxes) have been
used to assess the overall impact of the policies on consumers over time. The results also include the impact on
main importing countries.
Demand-side investments include outlays for renewables assets used primary in heating applications, such as
solar thermal and bioenergy-based heating equipment.

In addition to energy efficiency and renewables, investments in other end-use are derived from projections for
CCUS (carbon capture, utilisation and storage) in industry, spending to meet the incremental cost of EVs and
investment in EV charging infrastructure. While a portion of EV charging infrastructure – that for installed for
private use – is allocated to demand-side investments, publicly-accessible charging infrastructure is included
under power sector distribution networks.

8.3 Financing for investments

Building upon analysis carried out in the World Energy Investment report, a more detailed assessment of the
sources of finance and financing costs associated with investments was carried out in some sectors and regions.
While project developers act as the primary actors investing in energy assets, their success depends on a having
robust inter-connected system of secondary financial sources and intermediaries, diverse investment vehicles to
facilitate flows and clear signals for action, based on profit expectations and risk profile.
An analysis of the ownership of clean-energy related investment in the Sustainable Development Scenario was
carried out based on estimates of the share carried out by state-owned enterprises (SOEs) and government
sponsors (public sources) compared with that led by private actors.
How companies across the energy sector fund their overall operations and growth varies significantly, with much
depending on the type of developer and investor, and the risk and return profile of the asset. In this Outlook, we
have carried out our first detailed analysis of the capital structure of energy investments, analysing the mix of

71
debt and equity that developers, households and project companies use in the primary financing of projects and
assets. The estimates are based on current observed capital structures by sector, derived from financing data on
companies and consumers (for balance sheet financing) and assets (for project finance), applied to investment
projections. Understanding the capital structure of energy investments can be very useful for policy makers, not
least as a guide to where limited public finance can have the greatest impact.

The cost of financing is expressed through the weighted average cost of capital, which measures the required
return on equity and the cost of debt issuance, weighted according to a project or company’s capital structure.
Within WEO-2020, more detailed financing cost analysis was performed for utility-scale solar PV projects, shale
companies , coal mining companies, and for economy-wide comparisons of advanced with emerging market and
developing economies. These assessments were derived from country level data on risk-free rates, market risk
premiums and project/company-specific risk premiums obtained from Bloomberg, Thomson Reuters and
Damodaran Online, as well as sector-specific sources and analysis.

Finally, more analysis was carried out on the role of the capital market and secondary finance transactions for
energy investments. This included assessments from a broader standpoint the recent dramatic rise of sustainable
financial flows, and related regulatory developments, and how this trend relates to energy investment, as well
as the increased trend over the past decade of acquisitions and refinancing of energy assets.

9 Energy access
9.1 Defining modern energy access
There is no single internationally-accepted and internationally-adopted definition of modern energy access.

Yet significant commonality exists across definitions, including:

 Household access to a minimum level of electricity

 Household access to safer and more sustainable (i.e. minimum harmful effects on health and the
environment as possible) cooking and heating fuels and stoves

 Access to modern energy that enables productive economic activity, e.g. mechanical power for
agriculture, textile and other industries
 Access to modern energy for public services, e.g. electricity for health facilities, schools and street
lighting
All of these elements are crucial to economic and social development, as are a number of related issues that are
sometimes referred to collectively as "quality of supply", such as technical availability, adequacy, reliability,
convenience, safety and affordability.
The data and projections presented in the WEO focus on two elements of energy access: a household having
access to electricity and to clean cooking facilities. The IEA defines energy access as "a household having reliable
and affordable access to both clean cooking facilities and to electricity, which is enough to supply a basic bundle
of energy services initially, and with the level of service capable of growing over time". This energy access
definition serves as a benchmark to measure progress towards goal SDG 7.1 and as a metric for our forward-
looking analysis.
Electricity access entails a household having initial access to sufficient electricity to power a basic bundle of
energy services – at a minimum, several lightbulbs, phone charging, a radio and potentially a fan or television –
with the level of service capable of growing over time. In our projections, the average household who has gained

72
access has enough electricity to power four lightbulbs operating at five hours per day, one refrigerator, a fan
operating 6 hours per day, a mobile phone charger and a television operating 4 hours per day, which equates to
an annual electricity consumption of 1 250 kWh per household with standard appliances, and 420 kWh with
efficient appliances. This service-level definition cannot be applied to the measurement of actual data simply
because the level of data required does not exist in a large number of cases. As a result, our electricity access
databases focus on a simpler binary measure of those that have a connection to an electricity grid, or have a
renewable off- or mini-grid connection of sufficient capacity to deliver the minimum bundle of energy services
mentioned above.

Access to clean cooking facilities means access to (and primary use of) modern fuels and technologies, including
natural gas, liquefied petroleum gas (LPG), electricity and biogas, or improved biomass cookstoves (ICS) that have
considerably lower emissions and higher efficiencies than traditional three-stone fires for cooking. Currently,
very few ICS models attain this lower emissions target, particularly under real-world cooking conditions.
Therefore, our clean cooking access database refers to households that rely primarily on fuels other than biomass
(such as fuelwood, charcoal, tree leaves, crop residues and animal dung), coal or kerosene for cooking. For our
projections, only the most improved biomass cookstoves that deliver significant improvements are considered
as contributing to energy access. The main sources are the World Health Organisation (WHO) Household Energy
Database and the IEA Energy Balances.

9.2 Outlook for modern energy access

9.2.1 Outlook for electricity access

The IEA’s electricity access database19 provides valuable information about the current electrification rates in a
large number of countries. In order to provide an outlook for electricity access in the next decades, a model able
to generate projections of electrification rates by region has been developed. The projections are based on an
econometric panel model that regresses historic electrification rates of different countries over many variables,
to test their level of significance. Variables that were determined statistically significant and consequently
included in the equations are per-capita income, demographic growth, urbanisation, fuel prices, level of
subsidies, technological advances, energy consumption, and energy access programmes.

9.2.2 Outlook for clean cooking access

Our baseline data on the traditional use of biomass for cooking is based on the World Health Organization’s
(WHO) Global Health Observatory estimates of reliance on solid fuels. 20 To provide an outlook for the number of
people relying on the traditional use of biomass in the next decades, a regional model was developed under
different assumptions. Reliance on biomass rates of different countries is projected using an econometric panel
model estimated from a historical time series. Variables that were determined statistically significant and
consequently included in the equations are per-capita income, demographic growth, urbanisation level, level of
prices of alternative modern fuels, subsidies to alternative modern fuel consumption, technological advances
and clean cooking programmes.

For more detail on the energy access analysis and methodology see also the dedicated section on the WEO
website: https://www.iea.org/topics/energy-access.

19 https://www.iea.org/reports/sdg7-data-and-projections/access-to-electricity
20 For more information, see www.who.int/gho/phe/indoor_air_pollution/en/index.html

73
 9.3 Affordability of basic electricity services
For WEO-2020, a new analysis was conducted on the impact of the Covid-19 pandemic on the affordability of
basic electricity services for households in Africa and Developing Asia. Using poverty data from Lakner et al.
(2020), as well as country electricity prices, we analysed the extent to which Covid-19 induced poverty could
bring about energy poverty if households become unable to afford basic electricity services.
We considered two bundles of electricity services: an essential bundle (including four lightbulbs operating four
hours per day, a fan three hours per day and a television two hours per day; equating to 500 kilowatt-hours (kWh)
per household per year with standard appliances), and an extended bundle (including the essential bundle plus
one refrigerator, and double hours for the fan and the television; equating to 1 250 kWh per household per year
with standard appliances). The number of people at risk of losing basic electricity services was estimated by
combining data on the costs of these bundles in different countries with data on the number of additional
households pushed across different poverty lines ($1.90/day, $3.20/day or $5.50/day) as a result of the crisis.
We considered a household at risk of losing ability to pay when it represents over 5% of the household spending.

74
Annex 1: WEM regional definitions
A1.1 World Energy Outlook (WEO) publication
In several tables of this methodology document, as well as in the WEO publication itself, results from the WEM
model are often presented with the below regional groupings.
Advanced economies: OECD regional grouping and Bulgaria, Croatia, Cyprus21,22, Malta and Romania.

Africa: North Africa and sub-Saharan Africa regional groupings.

Asia Pacific: Southeast Asia regional grouping and Australia, Bangladesh, China, India, Japan, Korea, Democratic
People’s Republic of Korea, Mongolia, Nepal, New Zealand, Pakistan, Sri Lanka, Chinese Taipei, and other Asia
Pacific countries and territories.23

Caspian: Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan and Uzbekistan.
Central and South America: Argentina, Plurinational State of Bolivia (Bolivia), Brazil, Chile, Colombia, Costa Rica,
Cuba, Curaçao, Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Jamaica, Nicaragua,
Panama, Paraguay, Peru, Suriname, Trinidad and Tobago, Uruguay, Bolivarian Republic of Venezuela (Venezuela),
and other Central and South American countries and territories.24

China: Includes the (People's Republic of) China and Hong Kong, China.

Developing Asia: Asia Pacific regional grouping excluding Australia, Japan, Korea and New Zealand.

Emerging market and developing economies: All other countries not included in the “advanced economies”
regional grouping.

Eurasia: Caspian regional grouping and the Russian Federation (Russia).

Europe: European Union regional grouping and Albania, Belarus, Bosnia and Herzegovina, North Macedonia,
Gibraltar, Iceland, Israel25, Kosovo, Montenegro, Norway, Serbia, Switzerland, Republic of Moldova, Turkey,
Ukraine and United Kingdom.

21Note by Turkey: The information in this document with reference to “Cyprus” relates to the southern part of the Island.
There is no single authority representing both Turkish and Greek Cypriot people on the Island. Turkey recognises the Turkish
Republic of Northern Cyprus (TRNC). Until a lasting and equitable solution is found within the context of the United Nations,
Turkey shall preserve its position concerning the “Cyprus issue”.
22Note by all the European Union Member States of the OECD and the European Union: The Republic of Cyprus is recognised
by all members of the United Nations with the exception of Turkey. The information in this document relates to the area
under the effective control of the Government of the Republic of Cyprus.
23Individual data are not available and are estimated in aggregate for: Afghanistan, Bhutan, Cook Islands, Fiji, French
Polynesia, Kiribati, Macau (China), Maldives, New Caledonia, Palau, Papua New Guinea, Samoa, Solomon Islands, Timor-Leste
and Tonga and Vanuatu.
24 Individual data are not available and are estimated in aggregate for: Anguilla, Antigua and Barbuda, Aruba, Bahamas,
Barbados, Belize, Bermuda, Bonaire, British Virgin Islands, Cayman Islands, Dominica, Falkland Islands (Malvinas), French
Guiana, Grenada, Guadeloupe, Guyana, Martinique, Montserrat, Saba, Saint Eustatius, Saint Kitts and Nevis, Saint Lucia, Saint
Pierre and Miquelon, Saint Vincent and Grenadines, Saint Maarten, Turks and Caicos Islands.
25 The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use of such
data by the OECD and/or the IEA is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements
in the West Bank under the terms of international law.

75
European Union: Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France,
Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal,
Romania, Slovak Republic, Slovenia, Spain, Sweden.

IEA (International Energy Agency): OECD regional grouping excluding Chile, Iceland, Israel, Latvia, Lithuania and
Slovenia.

Latin America: Central and South America regional grouping and Mexico.

Middle East: Bahrain, Islamic Republic of Iran (Iran), Iraq, Jordan, Kuwait, Lebanon, Oman, Qatar, Saudi Arabia,
Syrian Arab Republic (Syria), United Arab Emirates and Yemen.

Non-OECD: All other countries not included in the OECD regional grouping.

Non-OPEC: All other countries not included in the OPEC regional grouping.

North Africa: Algeria, Egypt, Libya, Morocco and Tunisia.

North America: Canada, Mexico and United States.

OECD (Organisation for Economic Co-operation and Development): Australia, Austria, Belgium, Canada, Chile,
Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy,
Japan, Korea, Latvia, Lithuania, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal,
Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, United Kingdom and United States. Colombia
became a member of the OECD in April 2020, and its membership is not yet reflected in WEO projections for the
OECD.

OPEC (Organisation of the Petroleum Exporting Countries): Algeria, Angola, Republic of the Congo (Congo),
Equatorial Guinea, Gabon, Islamic Republic of Iran (Iran), Iraq, Kuwait, Libya, Nigeria, Saudi Arabia, United Arab
Emirates and Bolivarian Republic of Venezuela (Venezuela), based on membership status as of October 2020.

Southeast Asia: Brunei Darussalam, Cambodia, Indonesia, Lao People’s Democratic Republic (Lao PDR), Malaysia,
Myanmar, Philippines, Singapore, Thailand and Viet Nam. These countries are all members of the Association of
Southeast Asian Nations (ASEAN).

Sub-Saharan Africa: Angola, Benin, Botswana, Cameroon, Republic of the Congo (Congo), Côte d’Ivoire,
Democratic Republic of the Congo, Eritrea, Ethiopia, Gabon, Ghana, Kenya, Mauritius, Mozambique, Namibia,
Niger, Nigeria, Senegal, South Africa, South Sudan, Sudan, United Republic of Tanzania (Tanzania), Togo, Zambia,
Zimbabwe and other African countries and territories.26

26Individual data are not available and are estimated in aggregate for: Burkina Faso, Burundi, Cabo Verde, Central African
Republic, Chad, Comoros, Djibouti, Equatorial Guinea, Kingdom of Eswatini, Gambia, Guinea, Guinea-Bissau, Lesotho, Liberia,
Madagascar, Malawi, Mali, Mauritania, Réunion, Rwanda, Sao Tome and Principe, Seychelles, Sierra Leone, Somalia and
Uganda.

76
Figure 25: World Energy Outlook regional groupings

Note: 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.

A1.2 Oil and natural gas supply modules

The WEM oil and gas supply modules are made up of 120 regional models of which 101 countries have production
levels modelled on an individual basis. Trade volumes broken downy by pipeline and liquefied natural gas are
modelled only for the following 20 areas: Canada, Mexico, United States, Brazil, Other Latin America, European
Union, Other OECD Europe, Other Easter Europe, North Africa, West Africa, East Africa, Russia, Caspian, Middle
East, OECD Asia, OECD Oceania, China, India, Southeast Asia, and Other Asia. The 101 countries for which oil and
gas production levels are modelled individually can be categorised into the 20 gas trade modelling areas in the
following manner:

Canada: Canada.
Mexico: Mexico.
United States: United States.
Brazil: Brazil.
Other Latin America: Argentina, Bolivia, Chile, Colombia, Cuba, Ecuador, Paraguay, Peru, Trinidad and Tobago,
Uruguay, and Venezuela.
European Union: Denmark, Estonia, France, Germany, Greenland, Italy, Netherlands, Poland, Romania, Slovenia,
Sweden, and United Kingdom.
Other OECD Europe: Israel (included in this region for statistical reasons) and Norway.
Other Eastern Europe: Ukraine.
North Africa: Algeria, Libya, Egypt, Tunisia, and Morocco.

77
West Africa: Angola, Benin, Cameroon, Central African Republic, Chad, Congo, Democratic Republic of Congo,
Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Ivory Coast, Liberia, Mauritania, Niger,
Nigeria, Senegal, Sierra Leone, and Togo.
East Africa: Botswana, Eritrea, Ethiopia, Kenya, Madagascar, Mozambique, Namibia, Seychelles, Somalia, South
Africa, South Sudan, Sudan, Tanzania, and Uganda.
Russia: Russia.
Caspian: Azerbaijan, Kazakhstan, Turkmenistan, and Uzbekistan.
Middle East: Bahrain, Iran, Iraq, Jordan, Kuwait, Lebanon, Oman, Qatar, Saudi Arabia, Syria, United Arab
Emirates, and Yemen. Data for Saudi Arabia and Kuwait include 50% each of production from the Neutral Zone.
OECD Asia: Japan and Korea.
OECD Oceania: Australia and New Zealand.
China: China.
India: India.
Southeast Asia: Brunei Darussalam, Indonesia, Malaysia, Philippines, Thailand, and Viet Nam.
Other Asia: Bangladesh and Pakistan.

A1.3 Coal supply module

19 countries are modelled on an individual basis in the WEM coal supply module: Australia, Brazil, Canada,
Chile, China, Colombia, India, Indonesia, Japan, Korea, Mexico, Mongolia, Mozambique, New Zealand, Russia,
South Africa, the United States, Venezuela and Viet Nam.

Annex 2: References
Aboumahboub et al. (2010): Optimal Configuration of a Renewable-based Electricity Supply Sector, WSEAS
Transactions on Power Systems (ISSN: 1790-5060), 2, p. 120-129, http://www.wseas.us/e-
library/transactions/power/2010/89-612.pdf, accessed 28 June 2011

Airbus SAS (2004), Global Market Forecast 2004-2023, Airbus SAS, Blagnac, France.

AIM (Aviation Integrated Model), developed at UCL, https://www.ucl.ac.uk/energy-models/models/aim

Allen et al. (2016), New use of global warming potentials to compare cumulative and short-lived climate
pollutants, Nature Climate Change (6).
Anandarajah, G. et al. (2011), TIAM-UCL Global Model Documentation, UKERC Working Paper
UKERC/WP/ESY/2011/001, UK Energy Research Centre, London.
APP (Asia-Pacific Partnership on Clean Development and Climate) (2009), Energy Efficiency and Resource
Saving Technologies in Cement Industry, APP, Washington D.C.,
http://www.asiapacificpartnership.org/english/cement_tf_docs.aspx

BGR (German Federal Institute for Geosciences and Natural Resources) (2014), Energiestudie 2014, Reserven,
Ressourcen und Verfügbarkeit von Energierohstoffen, (Energy Study 2014, Reserves, Resources and Availability
of Energy Resources), BGR, Hannover, Germany.

Boeing (2005), Boeing Current Market Outlook 2005, Boeing, Chicago.

BP (2015), BP Statistical Review of World Energy 2015, BP, London.

78
Breiman, L. et al. (1984), Classification and Regression Trees, Wadsworth International, Belmont, United States.

CNI (Confederação nacional da indústria), IEL (Instituto Euvaldo Lodi), Eletrobras (2010a), Oportunidades de
eficiência energética para a indústria – setor cimenteiro [Opportunities for energy efficiency for industry –
cement sector], CNI, Brasilia, Brazil

- (2010b), Oportunidades de eficiência energética para a indústria – setor papel e celulose [Opportunities for
energy efficiency for industry – pulp and papersector], CNI, Brasilia, Brazil

CSI (Cement Sustainability Initiative) (2013), Existing and Potential Technologies for Carbon Emissions
Reductions in the Indian Cement Industry, CSI, Geneva.
CSI (Cement Sustainability Initiative) and ECRA (European Cement Research Academy) (2009), Development of
State of the Art-Techniques in Cement Manufacturing: Trying to Look Ahead, CSI/ECRA Technology Paper, CSI,
Geneva.
Dargay, J., D. Gately and M. Sommer (2006), “Vehicle Ownership and Income Growth, Worldwide: 1960-2030”,
Energy Journal, Vol. 28, No. 4, Elsevier, Amsterdam.

EC (European Commission) (2001), Integrated Pollution Prevention and Control (IPPC) – Reference Document on
Best Available Techniques in the Chlor-Alkali Manufacturing Industry, EC, Brussels.

- (2003), Integrated Pollution Prevention and Control (IPPC) – Reference Document on Best Available Techniques
in the Large Volume Organic Chemical Industry, EC, Brussels.
- (2010), Integrated Pollution Prevention and Control (IPPC) – Reference Document on Best Available Techniques
in the Pulp and Paper Industry, EC, Brussels.

Econoler, et al. (2011), Cooling Benchmarking Study Report, Collaborative Labeling and Appliance Standards
Program (CLASP), Brussels.
ENTSO-G (2017), Ten-Year Network Development Plan 2017, ENTSO-G, Brussels, Belgium.

- (2018), Ten-Year Network Development Plan 2018, ENTSO-G, Brussels, Belgium.

Eurofer (2014), A steel roadmap for a low carbon Europe 2050, The European Steel Association, Brussels.

EV Volumes (2020), https://www.ev-volumes.com/

EWITS (2011): Eastern Wind Integration and Transmission Study, prepared for: The National Renewable Energy
Laboratory, prepared by: EnerNex Corporation,
http://www.nrel.gov/wind/systemsintegration/pdfs/2010/ewits_final_report.pdf , accessed 28 June 2011. Data
available on http://www.nrel.gov/wind/systemsintegration/ewits.html, accessed 28 June 2011
FAO (2019), Paper and paperboard production, Food and Agriculture Organization of the United Nations,
Rome.
Fraunhofer ISI and FfE (2003), Möglichkeiten, Potenziale, Hemmnisse und Instrumente zur Senkung des
Energieverbrauchs branchenübergreifender Techniken in den Bereichen Industrie und Kleinverbrauch
[Possibilities, potentials, barriers and instruments for the reduction of energy consumption from cross-cutting
technologies in industry and small consumers], Fraunhofer Institute Systemtechnik und Innovationsforschung
and Forschungsstelle für Energiewirtschaft e.V., Karlsruhe and München, Germany

Fraunhofer ISI et al (2009), Study on the Energy Savings Potentials in EU Member States, Candidate Countries
and EEA Countries, Karlsruhe

79
Gately, D. (2006), “What Oil Export Levels Should We Expect from OPEC?”, Energy Journal, Vol. 28, No. 2,
Elsevier, Amsterdam.

GBPN (Global Buildings Performance Network) and CEU (Central European University) (2012), Best Practice
Policies for Low Carbon & Energy Buildings - Based on Scenario Analysis, GBPN and CEU, Paris and Budapest.

Heide et al. (2010): Seasonal optimal mix of wind and solar power in a future, highly renewable Europe.
Renewable Energy Vol. 35, p. 2483 - 2589, http://www.mng.org.uk/gh/resources/Heide_et_al2.pdf, accessed
28 June 2011

IATA (International Air Transport Association) (2012), Technology Roadmap Report 4th edition, IATA, Geneva,
available at http://www.iata.org/whatwedo/environment/Documents/technology-roadmap-2013.pdf

ICAO (International Civil Aviation Organization) (2019), Annual Report 2018, Air transport Statistics, Montreal,
available at https://www.icao.int/annual-report-
2018/Documents/Annual.Report.2018_Air%20Transport%20Statistics.pdf

ICF (2016a), Economic Analysis of Methane Emission Reduction Potential from Natural Gas Systems. Fairfax, VA,
United States: ICF.

− (2016b), Summary of Methane Emission Reduction Opportunities Across North American Oil and Natural Gas
Industries. Fairfax, VA, United States: ICF.
IEA (International Energy Agency) (2007), Tracking Industrial Energy Efficiency and CO2 Emissions, OECD/IEA,
Paris.

− (2009), Energy Technology Transitions for Industry, OECD/IEA, Paris.

− (2010), Energy Technology Perspectives 2010, OECD/IEA, Paris.
− (2011), Technology Roadmap: Energy-Efficient Buildings: Heating and Cooling Equipment, OECD/IEA, Paris

− (2012a), Energy Technology Perspectives 2012 – Pathways to a Clean Energy System, OECD/IEA, Paris.
− (2012b), Technology Roadmap: Fuel Economy of Road Vehicles, OECD/IEA, Paris.

− (2013a), Transition to Sustainable Building, OECD/IEA, Paris

− (2013b), Technology Roadmap – Energy and GHG Reductions in the Chemical Industry via Catalytic Processes,
OECD/IEA, Paris.

− (2013c), Technology Roadmap – Low-Carbon Technology for the Indian Cement Industry, OECD/IEA, Paris.
− (2014a), CO2 Emissions from Fuel Combustion 2014, OECD/IEA, Paris.

– (2014b), World Energy Investment Outlook: World Energy Outlook Special Report, OECD/ IEA, Paris.

− (2015a), WEO 2015 Special Report on Energy and Climate Change, OECD/IEA, Paris.
− (2015b), World Energy Outlook 2015, OECD/IEA, Paris.

– (2016a), WEO 2015 Special Report on Energy and Air Pollution, OECD/IEA, Paris
− (2017a), “Energy Sector Investment to Meet Climate Goals”, in Perspectives for the Energy Transition –
Investment Needs for a Low-Carbon Energy System, 51–120,
www.iea.org/publications/insights/insightpublications/PerspectivesfortheEnergyTransition.pdf
− (2017b), Energy Access Outlook: from Poverty to Prosperity, World Energy Outlook Special Report, OECD/IEA,
Paris

80
− (2017c), Energy Technology Perspectives - Catalysing Energy Technology Transformations, OECD/IEA, Paris.

− (2017d), World Energy Outlook 2017, OECD/IEA, Paris.

− (2018a), Global Electric Vehicle Outlook 2018, OECD/IEA, Paris.

− (2018b), World Energy Outlook 2018, OECD/IEA, Paris.

− (2018c), Technology Roadmap - Low-Carbon Transition in the Cement Industry, OECD/IEA, Paris.

− (2019a), Material efficiency in clean energy transitions, IEA, Paris.

− (2019b), World Energy Outlook 2019, IEA, Paris.

− (2019c), The Future of Hydrogen 2019, IEA, Paris.

− (2019d), World Energy Investment 2019, IEA, Paris.

− (2020a), Sustainable Recovery, IEA, Paris.

− (2020b), Energy Technology Perspectives 2020, IEA, Paris.

IEA CCC (IEA Clean Coal Centre) (2012), CO2 abatement in the Iron and Steel Industry, IEA CCC, London.

IIP and IFC (2014), Waste Heat Recover for the Cement Sector: Market and Supplier Analysis, Institute for
industrial Productivity and International Finance Corporation, Washington D.C.

IMO (2014). Reduction of GHG emissions from ships - third IMO GHG study 2014. United Kingdom: London.

International Aluminium Institute (IAI) (2019), Primary Aluminium Production, London, http://www.world-
aluminium.org/statistics/

IPCC (Intergovernmental Panel on Climate Change) (2018), Global Warming of 1.5 °C. An IPCC Special Report on
the impacts of global warming of 1.5 °C above pre-industrial levels and related global greenhouse gas emission
pathways, in the context of strengthening the global response to the threat of climate change, sustainable
development and efforts to eradicate poverty, [Masson-Delmotte, V., P. Zhai, H.-O. Pörtner, D. Roberts, J. Skea,
P.R. Shukla, A. Pirani, W. Moufouma-Okia, C. Péan, R. Pidcock, S. Connors, J.B.R. Matthews, Y. Chen, X. Zhou,
M.I. Gomis, E. Lonnoy, T. Maycock, M. Tignor, and T. Waterfield (eds.), World Meteorological Organization,
Geneva, https://www.ipcc.ch/sr15/download/#full.

IPTS (Institute for Prospective Technological Studies) (2011), Best Available Techniques (BAT) Reference
Document for Iron and Steel Production, European Commission, Seville, Spain.
Kannan, R. et al. (2007), UK MARKAL Model Documentation, UK Energy Research Centre, London,
http://www.ukerc.ac.uk/support/tiki-index.php?page=ES_MARKAL_Documentation_2010.
Kaza, S. et al. (2018), What a Waste 2.0, A Global Snapshot of Solid Waste Management to 2050 , International
Bank for Reconstruction and Development, The World Bank, Washington, DC
Ke, J. et al. (2012), “Potential Energy Savings and CO2 Emissions Reduction of China’s Cement Industry”, Energy
Policy, Vol. 45, Elsevier, Amsterdam, pp. 739-751.
Kermeli, K et al. (2015), “Energy efficiency improvement and GHG abatement in the global production of
primary aluminium”, Energy Efficiency, Vol. 8 (4), Springer, Heidelberg, pp. 629-666.

Kesicki, F and A. Yanagisawa (2014), “Modelling the potential for industrial energy efficiency in IEA’s World
Energy Outlook”, Energy Efficiency, Springer, Heidelberg

81
Koplow, D. (2009), Measuring Energy Subsidies Using the Price-Gap Approach: What Does It Leave Out?,
International Institute for Sustainable Development, Winnipeg, Canada.

Kramer, K. et al. (2009), Energy Efficiency Improvement and Cost Saving Opportunities for the Pulp and Paper
Industry, Lawrence Berkeley National Laboratory, Berkeley, United States.

Krause, J. et al (2017), Light Duty Vehicle CO2 Emission Reduction Cost Curves and Cost Assessment -the DIONE
Model, Publications Office of the European Union, Luxembourg.

Krause, J. and Donati, A (2018), Heavy duty vehicle CO2 emission reduction cost curves and cost assessment –
enhancement of the DIONE model CO2, Publications Office of the European Union, Luxembourg.

Lakner et al. (2020), How Much Does Reducing Inequality Matter for Global Poverty?, Global Poverty
Monitoring Technical Note, World Bank, Washington, DC.

LBNL (Lawrence Berkeley National Laboratory) (2012), Bottom Up Energy Analysis System - Methodology and
Results, Lawrence Berkeley National Laboratory and The Collaborative Labeling and Appliance Standards
Program.

Liu, G., Bangs, C.E. and D. B. Müller (2013), „Stock dynamics and emission pathways of the global aluminium
cycle”, Nature Climate Change, Vol. 3, Nature, London, pp. 338-342

Kong, L., et al. (2013), Analysis of Energy-Efficiency Opportunities for the Pulp and Paper Industry in China,
Lawrence Berkeley National Laboratory and South China University of Technology, Berkeley, United States.

Korkovelos, A., et al. (2018), A Geospatial Assessment of Small-Scale Hydropower Potential in Sub-Saharan
Africa. Energies, 11(11), 3100.
Madlool, N.A., et al. (2011), “A critical review on energy use and savings in the cement industries”, Renewable
and Sustainable Energy Reviews, Vol. 15, Elsevier, Amsterdam, pp. 2042-2060.

Martin, N., et al. (2000), Opportunities to Improve Energy Efficiency and Reduce Greenhouse Gas Emissions in
the U.S. Pulp and Paper Industry, Lawrence Berkeley National Laboratory, Berkeley, United States.

Martinez, L., Kauppila, J., & Castaing Gachassin, M. (2014). International freight and related CO2 emissions by
2050: A new modelling tool Organisation for Economic Co-operation and Development.
doi:10.1787/5jrw1kslrm9t-en
McNeil, M. and V. Letschert (2007), “Future Air Conditioning Energy Consumption in Developing Countries and
What Can be Done about it: The Potential of Efficiency in the Residential Sector”, European Council for an
Energy Efficient Economy (ECEEE) 2007 Summer Study, ECEEE, pp. 1311-1322.
METI (2018), “Forecast of Global Supply and Demand Trends for Petrochemical Products (for the period 2009
to 2022)”, Japan Ministry of Economy, Trade and Industry, Tokyo
Moya, J., N. Pardo and A. Mercier (2011), “The Potential for Improvements in Energy Efficiency and CO2
Emissions in the EU27 Cement Industry and the Relationship with the Capital Budgeting Decision Criteria”,
Journal of Cleaner Production, Vol. 19, No. 11, Elsevier, Amsterdam, pp. 1207-1215.
NEA/IEA (Nuclear Energy Agency/International Energy Agency) (2010), Projected Costs of Generating Electricity:
2010 Update, OECD, Paris.

NEDO (New Energy and Industrial Technology Development Organisation) (2008), Japanese Technologies for
Energy Savings / GHG Emissions Reduction – 2008 Revised Edition, NEDO, Kawasaki City, Japan.

82
Neelis, M., E. Worrell and E. Masanet (2008), Energy Efficiency Improvement and Cost Saving Opportunities for
the Petrochemical Industry, Lawrence Berkeley National Laboratory, Berkeley, United States.

Neelis, M., et al. (2012), Climate protection with rapid payback – Energy and CO2 savings potential of industrial
insulation in EU27, Ecofys and European Industrial Insulation Foundation, Utrecht, The Netherlands and Gland,
Switzerland.

NRC (National Research Council) (2010), Technologies and Approaches to Reducing the Fuel Consumption of
Medium- and Heavy-Duty Vehicles, The National Academy Press, Washington, DC.

NREL (2010): National Solar Radiation Data Base, 1991-2005 Update,

http://rredc.nrel.gov/solar/old_data/nsrdb/, accessed 28 June 2011

OAG (Official aviation Guide) (2020), https://www.oag.com/

Oda, J, et al. (2012), “International Comparisons of Energy Efficiency in Power, Steel and Cement Industries”,
Energy Policy, Vol. 44, Elsevier, Amsterdam, pp. 118-129.

OGJ (Oil and Gas Journal) (2014), “Global Reserves, Oil Production Show Increases for 2014”, OGJ, Pennwell
Corporation, Oklahoma City, United States.

OECD (Organisation for Economic Co-operation and Development) (2012a), The OECD ENV-Linkages Modelling
Framework, OECD, Paris.

− (2012b), OECD Environmental Outlook to 2050, OECD, Paris.

OECD/FAO (2018), OECD-FAO Agricultural Outlook 2018-2027, OECD Publishing, Paris/Food and Agriculture
Organization of the United Nations, Rome
Okazaki, T. and M. Yamaguchi (2011), “Accelerating the Transfer and Diffusion of Energy Saving Technologies
Steel Sector Experience – Lessons Learned”, Energy Policy, Vol. 39, Elsevier, Amsterdam, pp. 1296-1304.

Pardo, N., J.A. Moya and K. Vatopoulos (2012), Prospective Scenarios on Energy Efficiency and CO2 Emissions in
the EU Iron & Steel Industry, European Commission Joint Research Centre, Petten, Netherlands

Ren, T., M.K. Patel and K. Blok (2008), “ Steam cracking and methane to olefins: Energy use, CO2 emissions and
production costs “, Energy, Vol. 33, Elsevier, Amsterdam, pp. 817-833.
Resch, G., et al. (2004), “Forecasts of the Future Deployment of Renewable Energy Sources for Electricity
Generation – A Worldwide Approach”, Energy Economics Group Working Paper, Vienna University of
Technology, Vienna.

RISI (Resource Information System Inc) (2019), Pulp and paper industry database, RISI, Boston.
Sander+Partner GmbH (2010) http://www.sander-partner.ch/en/index.html, accessed 28 June 2011

Saygin, D., et al. (2009), Chemical and petrochemical sector – Potential of best practice technology and other
measures for improving energy efficiency, OECED/IEA, Paris.

Saygin, D., et al. (2011), “Potential of best practice technology to improve energy efficiency in the global
chemical and petrochemical sector”, Energy, Vol. 36, Elsevier, Amsterdam, pp. 5779-5790.
Schipper L., C. Marie-Liliu, and R. Gorham (2000), Flexing the link between Transpor and Greenhouse Gas
Emissions: a path for the World Bank, International Energy Agency/OECD, Paris.

Suranjana et al. (2010): The NCEP Climate Forecast System Reanalysis. Bull. Amer. Soc. Vol 91, p. 1015 – 1057

83
Tsinghua University (2008), Assisting Developing Country Climate Negotiators through Analysis and Dialogue:
Report of Energy Saving and CO2 Emission Reduction Analysis in China Cement Industry, Tsinghua University,
Beijing.

Ueckerdt, F., et. al. (2016). Decarbonizing global power supply under region-specific consideration of
challenges and options of integrating variable renewables in the REMIND model. Energy Economics.

UNIDO (United Nations Industrial Development Organization) (2010), Global Industrial Energy Efficiency
Benchmarking – An Energy Policy Tool, UNIDO, Vienna.

− (2011), Industrial Development Report 2011 – Industrial Energy Efficiency for Sustainable Wealth Creation,
UNIDO, Vienna.

UNPD (United Nations Population Division) (2019), World Population Prospects: The 2019 Revision, UNPD, New
York.

UNWTO (2012). UNWTO Tourism Highlights 2012. United Nations World Tourism Organization. Available at:
http://mkt.unwto.org/en/publication/unwto‐tourism‐highlights‐2013‐edition.

US DOE/EIA/ARI (Department of Energy/Energy Information Administration)/(Advanced Resources

International) (2013), Technically Recoverable Shale Oil and Shale Gas Resources: An Assessment of 137 Shale
Formations in 41 Countries Outside the United States, US DOE/EIA, Washington, DC.
US EPA (United States Environmental Protection Agency) (2010a), Available and Emerging Technologies for
Reducing Greenhouse Gas Emissions from the Iron and Steel Industry, US EPA, Washington, DC.

− (2010b), Available and Emerging Technologies for Reducing Greenhouse Gas Emissions from the Portland
Cement Industry, US EPA, Washington D.C.

− (2010c), Available and Emerging Technologies for Reducing Greenhouse Gas Emissions from the Pulp and
Paper Manufacturing Industry, US EPA, Washington, DC.

− (2013), Draft Inventory of US Greenhouse Gas Emissions and Sinks: 1990-2011, US EPA, Washington, DC.

USGS (United States Geological Survey) (2012a), “Assessment of Potential Additions to Conventional Oil and Gas
Resources of the World (Outside the United States) from Reserves Growth”, Fact Sheet FS2012-3052, USGS, Boulder,
United States.

− (2012b), “An Estimate of Undiscovered Conventional Oil and Gas Resources of the World”, Fact Sheet 2012-3042,
USGS, Boulder, United States.

− (2018a), Nitrogen (Fixed) – Ammonia, Boulder, United States.

− (2018b), Cement, USGS, Boulder, United States.

Waide, P. (2011), Opportunities for Success and CO2 Savings from Appliance Energy Efficiency Harmonization,
Navigant Consulting and CLASP, London and Brussels.

Wen, Z. and H. Li (2014), “Analysis of potential energy conservation and CO2 emissions reduction in China’s
non-ferrous metals industry from a technology perspective”, International Journal of Greenhouse Gas Control,
Vol. 28, Elsevier, Amsterdam, pp. 45-56.
WEPROG (2010), Weather and wind Energy Prognosis http://www.weprog.com/information, accessed 28 June
2011
WHO (2018),Household energy database http://www.who.int/airpollution/data/household-energy-
database/en/, accessed 12 November 2018

84
World Steel Association (2019), Steel Production 2018, World Steel Association, Brussels.

WWITS (2010), Western Wind and Solar Integration Study, prepared for: The National Renewable Energy
Laboratory, prepared by: GE Energy,
http://www.nrel.gov/wind/systemsintegration/pdfs/2010/wwsis_final_report.pdf , accessed 28 June 2011.
Data available on http://www.nrel.gov/wind/systemsintegration/wwsis.html, accessed 28 June 2011

Zhou, W., et al. (2010), “CO2 emissions and mitigation potential in China’s ammonia industry”, Energy Policy,
Vol. 38, Elsevier, Amsterdam, pp. 3701-3709.

85