Energy Technology Perspectives 2020 (ETP 2020) applies a combination of scenario techniques from now to 2070 to assess possible long-term evolutions of the energy sector. The first one is “backcasting”, or normative scenarios, which lay out plausible pathways to a desired end state. It allows us to identify milestones that need to be reached and trends that need to change promptly in order for the end goal to be achieved. The second one is about exploratory scenarios, which look at what is likely to happen on the basis of a set of assumptions. The advantage of exploratory scenarios, where the end state is a result of the analysis, is that it allows greater consideration of short-term constraints.

The analysis and modelling in ETP-2020 aim to identify an economical way for society to reach the desired outcome, but the scenario results do not necessarily reflect the least-cost way of doing so. In essence this is because an unconstrained least-cost approach may be rather theoretical, and may fail to take account of all the issues that need to be considered in practice. Many important issues cannot directly be captured in a cost optimisation framework, including, for example, political or individual preferences, feasible ramp-up rates, capital constraints and public acceptance. Therefore, doing a pure least-cost analysis for some end-use sectors (in particular buildings and transport) is not always suitable. Nor does such an analysis account for secondary effects resulting from climate change, such as adaptation costs. By combining varied modelling approaches, together with extensive expert consultation, we can model as accurately as possible the realities of the different sectors and develop real-world impact and in-depth insights.

All technologies modelled in ETP-2020 are already commercially available or at a relatively advanced stage of development, so that the technology has at least reached the large prototype phase and there is information about its expected performance and costs at scale. Costs for many of these technologies are expected to fall over time, helping to make a low-carbon future economically feasible. To make the results more robust, the analysis pursues a portfolio of technologies within a framework of cost minimisation within technical, economical and regulatory constraints. This offers a hedge against the real risks associated with the pathways: if one technology or fuel fails to fulfil its expected potential, it can more easily be compensated by another if its share in the overall energy mix is low. The tendency of the energy system to comprise a portfolio of technologies becomes more pronounced as carbon emissions are reduced, since the technology options for emissions reductions and their deployment potential typically depend on the local conditions in a country. Such scenario analysis can also help identify R&D priorities for less mature technology options to reduce the performance and cost gap with incumbent technologies and facilitate their market entry and wider deployment. 

The ETP model, which is the primary analytical tool used in ETP 2020, supports integration and manipulation of data from four soft-linked models:

  • energy conversion
  • industry
  • transport
  • buildings (residential and commercial/services).

The model makes it possible to explore outcomes that reflect variables in energy supply (using the energy conversion model) and in the three end-use sectors with the highest energy demand and largest greenhouse gas emissions (using models for industry, transport and buildings). Figure A.1 illustrates the interplay of these elements in the processes by which primary energy is converted into the final energy that is used in these three demand-side end-use sectors.

Figure A.1: Structure of the ETP model

Openexpand
Figure A.1: Structure of the ETP model

Coverage

The ETP-TIMES Supply model is a bottom-up, technology-rich model that depicts primary energy supply and transformation to final energy demand up to 2070. Primary energy supply covers conventional and unconventional fossil energy resources (hard coal, lignite, oil and natural gas), nuclear fuel supply and renewable resources (biomass, solar, wind, geothermal and marine energy). The transformation sector includes power and heat generation, oil refining, synthetic fuel production, liquid and gaseous biofuel production as well as the production of hydrogen and hydrogen-based fuels (synthetic hydrocarbon fuels from hydrogen and CO2, ammonia). Direct air capture of CO2 from the atmosphere, though a cross-cutting technology option, is also part of the supply model.

The model covers 28 regions, representing either individual countries, such as India, or aggregates of several countries, such as the Middle East. The model regions are linked by trade in fossil energy carriers (crude oil, petroleum products, coal, pipeline gas, liquefied natural gas), biofuels (biodiesel, bioethanol), liquefied hydrogen, synthetic hydrocarbon fuels from hydrogen, ammonia and electricity.

Methodology

The ETP-TIMES Supply model is based on the TIMES (The Integrated MARKAL-EFOM System) model generator, which has been developed and is being maintained by the Energy Technology Systems Analysis Programme, one of the International Energy Agency’s (IEA) Technology Collaboration Programmes. TIMES allows an economic representation of local, national and multiregional energy systems on a technologically detailed basis (Loulou et al., 2016).

Starting from the current situation (e.g. existing capacity stock, operating costs and conversion efficiencies) in the conversion sector (see Figure A.1), the supply model integrates the technical and economic characteristics of new technologies that could be added to the energy system in the future. The model can then determine the least‑cost technology mix needed to meet the final energy demand calculated in the ETP end-use sector models for agriculture, buildings, industry and transport (Figure A.2).

Figure A.2: Structure of the ETP-TIMES model for the conversion sector

Openexpand
Figure A.2: Structure of the ETP-TIMES model for the conversion sector

Overall, around 400 technologies are considered in the conversion sector. Technologies are described in terms of their technical and economic parameters, such as conversion efficiencies or specific investment costs. Learning curves are used for new technologies to link future cost developments with cumulative capacity deployment.

Electricity demand is divided into non-urban and urban, with urban further divided into five city classes by population size to reflect local differences in the technical potential for rooftop solar photovoltaics (PV) and municipal solid waste. Renewable energy resources for onshore and offshore wind, solar PV and solar thermal electricity are based on geospatial assessments for the different model regions, differentiating the potential in those regions, taking account of the capacity factor of the resources (in addition, for offshore wind, the water depth and distance to the coast) and of the distance from the location of the resources to the city classes (five distance categories) as an approximation for the transmission costs needed to use these resources. To capture the impact on investment decisions of variations in electricity and heat demand, as well as the variation in generation from certain renewable technologies, a year is divided into four seasons, with each season being represented by a typical day, which again is divided into eight daily load segments of three hours. The ETP-TIMES Supply model also takes into account additional constraints in the energy system such as emissions reduction goals, and it can decide endogenously to retire capacity before the end of the technical lifetime of a technology, as well as to add capacity. The model’s results provide detailed information on future energy flows and their related emissions impacts, required technology additions and the overall cost of the supply-side sector.

For a more detailed analysis of the operational aspects of the electricity sector, the long-term ETP-TIMES Supply model has been supplemented with a linear dispatch model. This model uses the outputs of the ETP-TIMES Supply model, in particular the installed capacities, to analyse the operation of the electricity system for a specific model region and year. This allows for detailed assessment of an entire year with one-hour time resolution using datasets for wind production, solar PV production and hourly electricity demand. The dispatch model’s results are incorporated into the long-term ETP-TIMES Supply model.

Given the hourly demand curve and a set of technology-specific operational constraints, the linear dispatch model determines the optimal hourly generation profile, as illustrated in Figure A.3 over a two-week period. To increase the flexibility of its modelling of the electricity system, the linear dispatch model can indicate investment needs in electricity storage or additional flexible generation technologies (such as gas turbines), and can include demand response from electricity use in the transport and buildings sectors in its analysis.

The linear dispatch model represents storage in terms of three steps: charge, store and discharge. The major operational constraints included in the model are capacity states, minimum generation levels and time, ramp-up and ramp-down, minimum downtime hours, annualised plant availability, cost considerations associated with start-up and partial-load efficiency penalties, and maximum storage reservoir capacity in energy terms (megawatt hours [MWh]).

Model limitations include challenges associated with a lack of comprehensive data on storage volume (MWh) for existing pumped hydro storage plants in some countries and regions. Electricity networks are not explicitly modelled, which precludes the study of the impacts of spatially dependent factors, such as the aggregation of variable renewable outputs with better interconnection.

Input data

Various forms of input data are required for the supply model. To replicate the energy flows in the historic base year, information (or assumptions) on installed capacities for existing technologies and assumptions about their technical and economic performance and operation are combined with the IEA’s World Energy Balances and Statistics. Assumptions are also made about the technical and economic characteristics of technologies in the future and about when new technologies may become available in the future, if not yet commercially available today. For primary energy supply, information on fossil energy reserves and resources and their production costs as well as renewable potentials and their temporal availability are further inputs into the model. Geospatial datasets were used to derive the regional potentials for solar PV, solar thermal energy and wind energy.

Information on existing installed capacities and their performance are collected from various sources, including the IEA medium-term market reports; the IEA Technology Collaboration Programmes commercial databases such as Platts’ World Electric Power Plant Database or Rystad Energy’s UCube; and reports from academia, industry and other organisations, such as the International Renewable Energy Agency.

Investment boundaries

Reported investments for supply-side sectors are based, as far as possible, on total costs: they include costs for equipment; labour; engineering, procurement and construction; as well as owner’s costs, but do not include interest costs during construction. Fixed operating and maintenance expenditures are not included under investments, though they are considered in the context of the economic characterisation of technologies in the model.

Coverage

For the purposes of the industry model, the industrial sector includes International Standard Industrial Classification (ISIC) Divisions 7, 8, 10-18, 20-32 and 41-43, and Group 099, covering mining and quarrying (excluding mining and extraction of fuels), construction, and manufacturing. Petrochemical feedstock use and blast furnace and coke oven energy use are also included within the boundaries of industry.

The industry sector is modelled using a hybrid approach: technology-rich models for five energy-intensive sub-sectors (iron and steel, chemicals and petrochemicals, cement, pulp and paper, and aluminium) combined with a cross-sectoral conversion device simulation model to cover the other industry sub-sectors (Figure A.4). The five technology-rich sub-sector models characterise the energy performance of process technologies at the process unit level, with global coverage using 40 countries and regions. The simulation model for the remaining industry sub-sectors characterises the stock of the main conversion devices (e.g. motors, heating equipment) used to provide various energy services required during the production of thousands of materials and products. 

Figure A.4: ETP industry model coverage

Openexpand
Figure A.4: ETP industry model coverage

Typically, raw material production is not included within the boundaries of the models, with the exception of the iron and steel sector, in which energy use for coke ovens and blast furnaces is covered. Due to the complexity of the chemicals and petrochemicals sector, the technology detail of the sub-model focuses on five products that represent about half of the sector’s energy use: ethylene, propylene, ammonia, methanol, benzene, toluene and xylene. The remaining industrial final energy consumption is accounted for in a simulation model that estimates energy consumption based on activity level.

Energy-intensive sub-sector model structure

For each of the five energy-intensity sectors, the modelling framework consists of a series of interacting sub-modules and a core technology model, the latter of which is implemented in the TIMES modelling framework (see Figure A.5). Each sub-module consists of an activity model, a stock model and a capacity model. Demand for materials is projected through interaction between an activity model and a stock model. The activity model uses country-level historical data on material consumption to calculate demand per capita, then projects forward total demand using population projections and industry value-added projections. The industry value-added projections inform the rate of growth in demand per capita.

The results of the activity model on demand projections feed into the stock model, which uses bottom-up material demand inputs from the ETP buildings, transport and supply models and complementary assumptions about other end-product shares and lifetimes to calculate the implied build-up of material stocks. Stock saturation in the stock model in turn informs per capita material demand saturation in the activity model through a series of iterations. Material efficiency strategies occurring in the other end-use sectors (e.g. building lifetime extension, vehicle light-weighting) are fed into the stock model via the bottom-up demand estimates, while material efficiency strategies within the industry boundary (e.g. manufacturing yield improvements, direct reuse and recycling) are modelled within the stock model. These strategies lead to reduced material demand, which is fed into the activity model via a material efficiency factor. The resulting activity projections from the activity model and scrap availability from the stock model feed into the main technology model.

Material trade between model regions is not modelled endogenously in the technology model, but rather is reflected in the activity projections developed in the activity and stock models. Apart from specific instances where announced policies or projected energy price signals provide relevant evidence to the contrary, trade patterns in material production and consumption are projected to follow current trends.

The capacity model contains data on historic and planned plant capacity additions and retrofits by plant type. Using assumptions about investment cycles, it calculates plant refurbishments and retirements. The resulting remaining capacity informs the main technology model. The capacity model also provides projections on the average age of plants at a given time.

The main technology model of each sector consists of a detailed representation of process technologies required for relevant production routes. Energy use and technology portfolios for each country or region are characterised in the base year using relevant energy use and material production statistics. Throughout the modelling horizon, demand for materials (as dictated by the activity model outputs) is met by technologies and fuels chosen through a constrained optimisation framework, with the objective function set to minimise overall system cost. System cost comprises energy costs and investments.

Figure A.5: ETP technology-rich industry sub-sector model structure

Openexpand
Figure A.5: ETP technology-rich industry sub-sector model structure

Changes in the technology and fuel mix, as well as efficiency improvements, are in part driven by a combination of exogenous assumptions on the penetration and energy performance of best available technologies, constraints on the availability of raw materials (such as scrap availability according to the stock model outputs), techno‑economic characteristics of the available technologies and process routes, and assumed progress on demonstrating innovative technologies at commercial scale. The results are sensitive to assumptions about how quickly physical capital is turned over (including retirements of existing capacity according to the capacity model outputs) and about the relative costs of the various technology options and fuels. A given scenario can also be subject to a CO2 emission trajectory that the model must adhere to. Model outputs include energy consumption, CO2 emissions, technology shares and investment requirements.

Investment boundaries

The boundaries for investments reporting include capital expenditure (CAPEX), and engineering, procurements and construction costs. For carbon capture, utilisation and storage (CCUS) technologies, CO2 transport and storage costs are also included. For material efficiency, investments in the most ambitious scenario are calculated as the difference with the baseline scenario, based on data on CO2 abatement costs for material efficiency strategies converted into costs for material savings. Fixed operating and maintenance expenditures (OPEX) are not included under reported investments, though they are considered in the context of the economic characterisation of technologies in the model.

Input data

Input data to the model come from a wide variety of sources. Sources for historical production and consumption include the World Steel Association, the International Fertilizer Association, the United States Geological Survey, the International Aluminium Institute and a number of proprietary sources. Data on the energy intensities of processes come from a variety of industry sources (e.g. the Getting the Numbers Right publication overseen by the Global Concrete and Cement Association), academic literature and industry contacts. CAPEX and OPEX similarly come from a combination of industry and academic sources. Fuel costs are based on energy price signal outputs from the ETP-TIMES Supply model.

Coverage

The buildings sector is modelled using a set of global simulation stock accounting modules, split into the residential and non-residential sub-sectors by end-use, fuel type and technology across 35 countries and/or regions. The residential sub-sector includes all energy-using activities in apartments and houses, including space heating and cooling, domestic hot water production, cooking, lighting, and the use of appliances and other electronic plug-loads. The non-residential sub-sector includes activities related to trade, finance, real estate, public administration, health, food and lodging, education, and other commercial services. This sub-sector is also referred to as the commercial and public service sector. It covers energy used for space heating, space cooling and ventilation; water heating; lighting; and a number of other miscellaneous energy-consuming equipment such as commercial appliances, office equipment, cooking devices, medical equipment and desalination equipment.

Figure A.6: Scope of the ETP buildings energy model

Openexpand
Figure A.6: Scope of the ETP buildings energy model

Model structure

For both the residential and non-residential sub-sectors, the model uses socio-economic drivers, such as population, gross domestic product (GDP), household income (approximated by average gross national income per capita), urbanisation and electrification rates, to project the major building energy demand indicators, including residential and non-residential floor space (in square metres [m²]), number of households and residential appliance ownership (including air conditioners, refrigerators, televisions and other white goods). As far as possible, country statistics are used for historical floor area, household numbers and appliance ownership rates. These data can be difficult to obtain, especially in developing countries, and the following methodology helped fill data gaps and project activity variables by country:

  • Identification of the set of socio-economic and climate variables that are the most closely correlated to activity variables (using a principal component analysis).
  • Clustering of countries by degree of similarity regarding socio-economic and climate variables to inform a country assessment based on data from countries that are likely to behave similarly (using a hierarchical clustering analysis).
  • Aggregation of the set of socio-economic indicators into a single driving parameter, where applicable (using a canonical variate analysis).
  • Definition of multi-parameter logistic functions to mathematically assess the relationship between the activity variables and the explanatory variables. These are S-shaped curves for floor area, number of employees and appliance ownership and inverse curves for household occupancy.

Once activity variables are estimated by country, a stock accounting model is used to estimate useful energy intensity across the various building end-uses with respect to assumed technology shares and equipment efficiencies relative to the IEA buildings sector energy balances for the residential and non-residential sub-sectors. These useful energy intensities (e.g. demand for space heating per unit of floor area in terms of delivered [i.e. useful] energy service) are then applied across the buildings end‑uses along with the projections for floor area, households and equipment ownership. The model takes into account the age of the buildings stock as well as the effects of ageing or refurbishment of buildings through corresponding degradation and improvement rates for the useful energy intensities. Changes in useful energy intensity (e.g. due to increasing household income leading to greater demand for thermal comfort or energy services) are also accounted for in the model inputs. Historical trends, country-level indicators and cross-country comparisons are used to estimate these expected changes in useful energy intensity over time.

For each of the derived useful energy demands, a suite of technology and fuel options are represented in the model, reflecting their current techno-economic characteristics (e.g. efficiencies and costs) as well as their future improvement potential. Depending on the current technology stock as well as on assumptions about the penetration and market shares of new technologies, the buildings sector model allows exploration of strategies and scenarios that cover the different useful energy demands to quantify the resulting final energy consumption and related CO2 emissions across the end‑uses of the buildings and the 35 model regions.

Special features

Special modules are constantly added to the IEA building energy model to enhance input granularity, provide a rationale for technology choices or complement its capabilities.

Since 2017, the model has used historical annual, monthly, daily and hourly climate data from 1990 to the base year from 13 000 weather stations. This is used primarily to estimate precise population-weighted cooling and heating degree days at the desired level of granularity, and this information is a direct input to the model. Temperature and humidity projections from representative carbon concentration pathways 4.5 and 2.6 are also used to assess the effect of climate change on heating and cooling demand.

Building upon local climate data, population density mapping and regional estimates of energy demand by end-use and sector provide a good basis for distributing heating and cooling demand at the local level and assessing clean technology deployment strategies. For instance, the assessment of heat and cold demand densities at the city or district level is key to making sound judgement calls on the decarbonisation potential of district energy systems (together with other variables such as the share of variable renewables in the electricity mix and the availability of waste heat sources). Local climate and population data are also used to derive heat pump energy performance.

Another special feature of the buildings energy model is the quantification of cement and steel demand for buildings construction and renovation. The stock accounting framework combined with other data sources on material intensities for construction, housing completions, structural frames, height and other indicators provides critical inputs to assessing material demand trends. The application of various material efficiency strategies is linked to the industry material demand modules. Strategies include lifetime extension through renovation within the stock accounting or the effect of structural optimisation, pre-casting, digitalisation, and reuse or waste reduction in the material accounting module.

Input data

The assumptions used in the ETP buildings model are developed using energy indicators that build on available historical data on energy and service-level trends. Shortfalls in detailed energy data currently constitute a major limiting factor for developing very detailed indicators in the buildings sector. Energy balances accounted for within the IEA World Energy Statistics and Balances include residential and non-residential energy use by fuel type. However, they do not include energy consumption by end-use or technology type, nor do they provide the buildings activity data (e.g. floor area and number of households) that constitute the essential denominator of any useful energy efficiency or service demand indicator. More disaggregated indicators are needed to inform bottom-up trends.

In 2009, ministers of the IEA member countries agreed to a new annual questionnaire dedicated to energy efficiency in order to improve the position. Since then, the IEA Secretariat has collected statistics on buildings end-use energy consumption and activity on an annual basis. These data are used to assemble indicators that constitute the basis of IEA buildings-related energy assessment and modelling. Although this is an important step forward, there are still major issues regarding missing data and data quality, especially for non-member countries, where the bulk of the buildings sector growth and energy demand increases is occurring.

Official country statistics are also updated regularly and provide valuable information on historical building energy trends. Databases that are reviewed annually include StatsCan (for Canada), RECS and CBECS (for the United States), NBS (for the People’s Republic of China), SEDLAC (for Latin America), Entranze, Enerdata, and EC Europa (for Europe). In parallel, policy signals driving energy efficiency trends are updated regularly to reflect additions in the IEA Policy Database. The IEA also exchanges data with industry associations, research institutes and other specialised organisations.

The IEA has also worked closely with partners, in particular Tsinghua University in China and the National Energy Board of Canada, to improve historical data resolution and global projections of the key buildings sector drivers through collaborative projects (e.g. The Future of Cooling in China, Heating and Cooling Strategies in Canada, etc.). These include, notably, household and floor area projections, building heating intensities by age, equipment stocks and residential appliance ownership estimates across air conditioners, and the six major appliance categories: refrigerators and freezers, clothes washers and dryers, dishwashers, and televisions.

Investment boundaries

Investments reporting in buildings include CAPEX and EPC. Energy efficiency investment trends are reflected through the evaluation of incremental rises in energy efficiency spending. For material efficiency, Sustainable Development Scenario investments are calculated as a difference between the Stated Policy Scenario and the Sustainable Development Scenario, based on data on CO2 abatement costs for material efficiency strategies converted into costs for material savings.

Model structure and input data

The Mobility Model (MoMo) is a technical-economic database spreadsheet and simulation model that enables detailed projections of transport activity, vehicle activity, energy demand and well-to-wheel greenhouse gas (GHG) and pollutant emissions according to user-defined policy scenarios to 2100.

The fundamental exogenous drivers for projections in MoMo are GDP, population, fuel prices and population density. MoMo projects the demand for passenger and freight activity in cities of three size categories and at the urban and non-urban level on the basis of Gompertz curves (for surface passenger activity). These “S”-curves are used to estimate vehicle ownership, mileage, and splits between private and public motorised transport (Figure A.7). Fiscal, regulatory, and other transport and land-use policies (including travel demand management and city planning, public transit investments, and emissions/traffic circulation restrictions, among others) are then modelled and used to alter these projections, based on changes seen in historical data and empirical studies in cities.

MoMo projects road freight activity based on log-log regressions, using fuel prices and GDP per capita as the key explanatory variables. Demand projections in international shipping (in tonne-kilometres) are provided by the International Transport Forum. For aviation, MoMo makes direct use of the Aviation Integrated Model, an open-source iterative cost minimisation model developed by researchers at University College London (UCL, 2020).

Input data

The MoMo modelling framework relies upon compiling and combining detailed data from various sources on vehicles in each of the countries and regions to estimate aggregate energy consumption, emissions and other energy-relevant metrics.

Historical data series have been collected by MoMo modellers from a wide variety of public and proprietary data sources for nearly two decades. National data are gathered primarily from the following organisations: national and international public institutions (e.g. the World Bank, the Asian Development Bank and Eurostat); national government ministries (e.g. departments of energy and transport, and statistical bureaus); federations, associations and non-governmental organisations (e.g. International Organization of Motor Vehicle Manufacturers, Japan Automobile Manufacturers Association, Korea Automobile Manufacturers Association and National Association of Automobile Manufacturers of South Africa); public research institutions (e.g. peer-reviewed papers and reports from universities and national laboratories); private research institutions (e.g. International Council on Clean Transportation); and private business and consultancies (e.g. IHS Markit, Segment Y, EVVolumes, and other major automotive market research and analysis organisations, in addition to major energy companies and automobile manufacturers themselves). Full details on data sources on a national or regional basis are documented in the regional data files of MoMo.

MoMo builds upon rich historical databases of road and rail vehicle market dynamics and operations. These databases compile activity on vehicle sales, stock, powertrain and fuel consumption (energy intensity), as well as estimated average mileage, scrappage and occupancy as well as load factors. Bottom-up estimates for each road vehicle category and powertrain, as shown in Figure A.7, are then calibrated and validated at the country and regional level as annual time series against the IEA Energy Balances for road vehicle consumption. In the case of rail, MoMo covers light and underground rail (and benefits from databases compiled by the International Association of Public Transport and the Institute for Transportation and Development Policy, as well as conventional passenger and freight rail, and high-speed rail (and benefits from a data-sharing agreement with the International Railway Union).

The road database covers 45 countries and regions and contains annual historical data stretching back at least to 1990, and in some countries as far back as 1970. MoMo reports average country and regional parameters, at urban and non-urban resolution, for 32 countries and regions, in 5-year time steps from 1990 to 2100.

Figure A.8: Road vehicle categories and powertrains covered in MoMo

Openexpand
Figure A.8: Road vehicle categories and powertrains covered in MoMo

In the case of shipping, MoMo builds upon public and proprietary data (the International Maritime Organization, the United Nations Conference on Trade and Development, Clarkson Research, Bloomberg) to build a vessel fleet-turnover model across five ship types: tankers, container ships, bulk carriers, general cargos and “other” ships. MoMo covers most major current and potential future maritime fuels (high sulphur fuel oils, very low sulphur fuel oils, diesel and biodiesel, liquefied natural gas and biomethane, ammonia, hydrogen, and electricity) and powertrains (internal combustion engine configurations, fuel cells, battery electric vessels).

Calibration of historical data with energy balances

The framework for estimating average and aggregate energy consumption for a given vehicle class can be neatly summarised by the ASIF identity (Schipper, Marie-Lilliu and Gorham, 2000):

where: F = total fuel use (megajoules [MJ] per year]; A = vehicle activity (vehicle kilometres [vkm] per year); I = energy intensity [MJ/vkm]; S = structure (shares of vehicle activity [%]); and i is an index of vehicle modes and classes. Vehicle activity can also be expressed as the product of vehicle stock (vehicles) and mileage (kilometre [km] per year). The energy used by each mode and vehicle class in a given year (MJ per year) can, therefore, be calculated as the product of three main variables: vehicle stock (S) (vehicles), mileage (M) (km/year) and fuel economy (FE) (MJ/vkm).

To ensure a consistent modelling approach is adopted across the modes, energy use is estimated based on stocks (via scrappage functions), utilisation (travel per vehicle), consumption (energy use per vehicle, i.e. fuel economy) and emissions (via fuel emission factors for CO2 and pollutants on a vehicle and well-to-wheel basis) for all modes. Final energy consumption, as estimated by the “bottom-up” approach described above, is then validated against and calibrated as necessary to the IEA Energy Balances (IEA, 2020). 

MoMo contains historical consumption estimates and scenario projections for fuel consumption by vehicle type and powertrain across a wide range of fuel options (gasoline and diesel, biofuels [ethanol and biodiesel via various production pathways], and synthetic alternatives to liquid fuels [coal-to-liquid, gas-to-liquid and other synthetic hydorcarbon fuels], gaseous fuels including natural gas [compressed natural gas and liquefied petroleum gas] and hydrogen via various production pathways, and electricity [with emissions according to the average national generation mix as modelled by the ETP-TIMES Supply model in the relevant scenario]).

To ease the manipulation and implementation of the modelling process, MoMo is split into modules that can be updated and elaborated upon independently. Figure A.8 shows how the modules interact with one another. By integrating assumptions on technology availability and cost in the future, the model reveals, for example, how costs could drop if technologies were deployed at a commercial scale and allows fairly detailed bottom-up “what-if” modelling.

Figure A.9: Structure of the ETP Transport model

Openexpand
Figure A.9: Structure of the ETP Transport model

Key elasticities have been included in MoMo since 2012. Price and income elasticities of fuel demand, for light-duty (passenger) road activity as well as road freight, based upon representative “consensus” literature values, are used to model vehicle activity and fuel consumption responses to changes in fuel prices – which are themselves driven by projections and policy scenarios (e.g. GHG or fuel taxes). These elasticities encompass the urban/non-urban split, and include the potential for municipal-level policies to reduce transport energy use.1 Elasticities also enable vehicle ownership to vary according to fuel prices and income, with GDP per capita acting as a proxy for income.

MoMo further enables estimation of scenario-based costs of vehicles, fuels and transport infrastructure, as well as the primary material inputs required for the construction of vehicles, related energy needs and resultant GHG emissions.

Investment boundaries

Vehicle platform, components and technology costs

Detailed cost modelling for plug-in hybrid accounts for initial (base year) costs, asymptotic (i.e. fully learned-out) costs and an experience parameter that defines the shape of cost reductions. These three parameters define learning functions that are based on the number of cumulative units produced worldwide. Cost functions define various vehicle configurations, including vehicle component efficiency upgrades (e.g. improved tyres or air conditioning controls), material substitution and vehicle downsizing, conventional spark and compression ignition engine improvements, conventional and plug-in hybrid powertrain configurations, batteries, electric motors, and fuel cells. These configurations are added to a basic glider cost. The ratios of differences in vehicle technologies deployed in plug-in hybrids are extrapolated to other road vehicle types (i.e. two- and three-wheelers and freight trucks).

The primary drivers of technological change in transport are assumptions about the cost evolution of the technology, and the policy framework incentivising adoption of the technology. For each scenario, the model supports a comparison of marginal costs of technologies and aggregates to total cost across all modes and regions.

Infrastructure and fuel costs

MoMo estimates future infrastructure costs according to scenario-based projections on modal activity and fuel use. Infrastructure cost estimates include capital, operations and maintenance, and reconstruction costs, split by geography into urban and non‑urban regions according to the location of the investments. Fuel costs are also estimated based on scenario-specific projections of urban and non-urban consumption, and include all fuel types (fossil-derived fuels, biofuels, electricity and hydrogen).

Economic activity (Table A.1) and population (Table A.2) are the two fundamental drivers of demand for energy services in ETP scenarios. These are kept constant across all scenarios as a means of providing a starting point for the analysis and facilitating the interpretation of the results. Following the outbreak of the Covid-19 pandemic in early 2020, the estimates for GDP growth have been lowered. Under the new assumptions, global GDP will more than triple between 2019 and 2070. The strongest growth is expected in Africa, Asia Pacific and the Middle East. The way economic growth plays into energy demand growth depends heavily on the structure of any given economy, on the balance between different types of industry and services, and on policies in areas such as pricing and energy efficiency.

The population data come from the medium variant of the United Nations projections. In this variant, global population growth slows over the coming decades, but the total population nonetheless rises from 7.7 billion in 2019 to 10.4 billion in 2070, an average growth of 0.6% per year. Almost three-quarters of the global increase to 2070 is in Africa, underlining the importance of this continent to the achievement of the world’s Sustainable Development Goals. India accounts for 10% of the rise in population and becomes the world’s most populous country around 2024.

Table A.1: Real GDP growth projections in ETP-2020 (assumed identical across scenarios)

 

Compound average annual growth rate

 

2000-19

2019-70

North America

2.0%

1.9%

   United States

2.0%

1.8%

Central and South America

2.4%

2.0%

Europe

1.8%

1.4%

   European Union

1.5%

1.2%

Africa

4.1%

3.7%

Middle East

3.6%

2.6%

Eurasia

3.9%

1.5%

Asia-Pacific

5.9%

2.7%

   China

8.8%

2.8%

   India

7.1%

3.5%

World

3.6%

2.4%

Note: Growth rates based on GDP in United States dollars (USD) in purchasing power parity (PPP) constant 2019 terms. Sources: International Monetary Fund World Economic Outlook database; World Bank databases; and IEA analysis and databases.

Table A.2: Population projections in ETP-2020 (millions)

 

Compound average annual growth rate

Population (million)

Urbanisation share

 

2000-19

2019-50

2019-70

2019

2050

2070

2019

2050

2070

North America

0.9%

0.5%

0.4%

493

579

609

82%

89%

92%

   United States

0.8%

0.5%

0.4%

329

380

405

82%

89%

92%

Central and South America

1.1%

0.5%

0.3%

521

607

602

81%

88%

91%

Europe

0.3%

0.0%

-0.1%

695

687

652

75%

84%

89%

   European Union

0.3%

-0.1%

-0.2%

514

499

470

76%

85%

89%

Africa

2.6%

2.1%

1.8%

1,309

2,490

3,309

43%

59%

69%

   Middle East

2.2%

1.2%

0.9%

243

348

379

72%

81%

86%

Eurasia

0.4%

0.2%

0.2%

235

253

255

65%

73%

80%

Asia-Pacific

1.0%

0.4%

0.2%

4,177

4,727

4,612

49%

65%

75%

   China

0.5%

-0.1%

-0.3%

1,406

1,376

1,235

61%

80%

87%

   India

1.4%

0.6%

0.3%

1,366

1,639

1,631

34%

53%

68%

World

1.2%

0.8%

0.6%

7,672

9,691

10,417

56%

68%

77%

In the ETP model, the definition of technologies ‘’available and in the innovation pipeline’’ includes around 800 technologies across the whole energy system and all fuels that are commercially available, or are at a relatively advanced stage of development (i.e. at least demonstration or large prototype stage has been reached).

The ETP Clean Energy Technology Guide was developed for a subset of these technologies across all sectors (over 400) that contribute to achieving the goal of net‑zero emissions. The guide is an interactive framework that contains information for each of these individual technology designs and components on the level of maturity (or technology readiness level, TRL2) and a compilation of development and deployment plans, as well as cost and performance improvement targets and leading players in the field. Further information can be found on the IEA website.

Existing commercial best available technology, for example, solar thermal and heat pumping technologies for space and water heating, LEDs for lighting, high‑performance windows (e.g. low-emissivity, double- or triple-glaze), high‑performance insulation, green or cool roofs, thermal energy storage, enhanced catalytic and biomass-based processes for chemical production, onshore wind, offshore wind, solar PV, solar thermal electricity, hydropower, geothermal (direct, flash), nuclear power, large-scale electric heat pumps, and conventional biodiesel and bioethanol.

Technologies which offer incremental improvements in performance compared with today’s best available technologies (may not be available yet but can be envisaged to be available within the time frame of the scenarios), for example, high-performance appliances in buildings, improved controls of cooling and heating (smart thermostats), advanced district energy networks, low-rolling resistance tyres, vehicle design improvements that reduce energy needs, and energy intensity improvements towards best available technology in industrial process technologies.

Technologies in demonstration phase (TRL 7 or 8). For example, high‑performance heat pumping technologies, high-efficacy (e.g. greater than 150 lumens/watt) LED lighting, advanced building insulation (aerogel, vacuum insulated panel, phase-change materials), upgraded smelt reduction for steel making, integration of CO2 capture in cement plants, zero-emission fuels for transport, large battery-electric ships, hydrogen fuel cell electric trucks, coal-fired integrated gasification combined cycle (IGCC), coal-fired IGCC with CO2 capture, coal-fired power plant with post‑combustion CO2 capture, conventional bioethanol with CO2 capture, advanced biodiesel, large-scale hydrogen electrolysis, and hydrogen from natural gas with CO2 capture.

Technologies for which large pilots are being tested (TRL 5 or 6), for example, “smart” building technologies and intelligent controls, dynamic solar control, hybrid heat pumps, fuel cells and hydrogen‐ready equipment, pure hydrogen-based steel production, full oxy-fuelling kilns for clinker production with CO2 capture, ammonia-fuelled engines for shipping, solid-state batteries with Li-metal anodes, oxy-fuelled coal power plants with CO2 capture, gas-fired power plants with CO2 capture, biomass integrated gasification combined cycle, wave energy, tidal stream, tidal lagoon, enhanced geothermal energy systems, advanced biodiesel with CO2 capture, hydrogen from biomass gasification, and biofuels from algae.

Supporting infrastructure to facilitate the uptake of improved and newly demonstrated technologies, for example, low-temperature distribution, high-performance district energy networks, smart grids with intelligent demand-side response, transport and storage infrastructure to support CCUS, and EV charging infrastructure.

Technology options that are not commercially available today are not deployed within the model until later time periods, depending on their current level of readiness, and some have constraints to account for process-specific limitations to deployment. See the sectoral chapters for more detailed discussion of technologies included in the ETP-2020 analysis.

The ETP model includes a set of analytical tools to derive new trends and insights from scenario results. One of them is an emissions decomposition framework allowing to split emissions reductions in a low-carbon scenario relative to a baseline trajectory by key overarching measures. These include:

  • Avoided demand: emission reductions from reductions in sectoral activity with no loss of useful energy service, such as material efficiency, modal shifts from private to public transport, urban planning strategies, active controls or indoor temperature settings in buildings,
  • Technology performance: emissions reductions from efficiency improvements in end-use sectors and in power and heat generation,
  • Electrification: emissions reductions in the end-use sectors caused by shifting from fossil fuel use to electricity, including also the often higher efficiency of electricity-based end-use technologies compared to fossil-based ones (including heat pumps),
  • Bioenergy: emissions reductions from increased bioenergy use in end-use sectors and the power sector; excluding bioenergy with carbon capture and storage (BECCS), which is reported under CCUS,
  • Other renewables: emissions reductions from the increased use of renewable resources, other than bioenergy, and excluding heat pumps
  • Other fuel shifts: emissions reductions from changes in the fuel mix of the end-use sectors and the power sector by switching to less carbon-intensive fuels, including nuclear, but excluding renewables, electricity, hydrogen and synthetic fuels,
  • CCUS: emissions reductions from the use of CCUS in electricity generation, other energy transformation and industry.

The ETP-2020 decomposition framework also enables emissions reductions to be split by technology readiness, depending on whether low-carbon technologies being deployed are mature, at early adoption, under demonstration or still a prototype in the base year.


The Logarithmic-Mean-Divisia Index (LMDI) breaks down the difference in CO2 emissions between two scenarios (in this case, the Stated Policies Scenario and the Sustainable Development Scenario) for a given year across a range of key factors. It provides a comprehensive framework to quantify the effect of low-carbon measures on CO2 emissions reduction, as their effect might not be additive. It also accounts for cross-sectoral interactions. For instance, an increase in the production of a low-carbon fuel might increase supply-side emissions but decrease overall emissions thanks to fuel shifting on the demand side.

The assessment is based on sectoral Kaya identities that single out the effects of activity, technology choice, energy intensity and CO2 emissions factors. In the case of transport:

Modes include aviation, shipping, two- or three-wheelers, light-duty vehicles, buses and minibuses, heavy‑duty trucks, etc. while powertrains include internal combustion engines, hybrid electric vehicles, plug-in hybrid electric vehicles, fuel cell vehicles, etc.

The LMDI decomposition framework compares each of these factors for all model regions (i.e. 25-35 model regions depending on the sector), for each year (from 2021 to 2070) and across the two scenarios. It uses logarithmic operators to transform this product into a sum of factors that are a function of one, and only one, lever used in the Sustainable Development Scenario to reduce CO2 emissions. These are:

  • Avoided demand, corresponding to changes in the activity variables
  • Fuel switching, corresponding to changes in the fuel shares
  • Technology performance, corresponding to changes in the intensity variables
  • CCUS, covering industry, power sector and other energy transformation applications, and accounting for the amount of CO2 captured relative to emissions induced by CCUS-related fuel use.
  • Decarbonisation of fuel supply, corresponding to changes in the carbon intensity variables including electricity, grid gas (e.g. through hydrogen or biomethane blending), hydrogen (e.g. by switching from steam reforming and coal gasification to electrolysis), commercial heat, etc.

One of the underlying principles of the LMDI is that the allocation of emissions reductions is based on how much each lever displaces fossil-fuel use (as opposed to how much of each fuel is used). This principle provides a way to allocate emissions reductions stemming from a technology route that uses more than one technology solution or energy vector (e.g. hybrid systems, combined production of heat and power, increased electricity use for the production of hydrogen, electro-fuels, etc).

Depending on each end-use, other influencing factors might be included in the decomposition, such as the effect of climate on heating and cooling demand, or the number of people with access to clean cooking as it affects liquid petroleum gas demand, electricity, biomass and natural gas use. The implications of demand‑side flexibility on the decarbonisation of power do not, however, form part of this methodological framework.

A major benefit of the LMDI decomposition is that it captures cross-sectoral interlinkages such as the effect of energy efficiency in demand sectors on power sector direct emissions reduction, or the effect of material efficiency strategies in construction on material manufacturing emissions (that are direct industry emissions). To account for these interlinkages, the decomposition is performed on direct and indirect CO2 emissions for each sector. Results at the energy sector level are obtained by summing sector-specific contributions only. In other terms, when results are aggregated, the contributions from other sectors to the reduction of the emissions of a given sector are excluded (e.g. decarbonisation of fuel supply for demand sectors, or reduction of electricity demand for the power sector) and substituted by the specific contribution of the sectors in question.

The decomposition also compares the market and technical characteristics of each of the 800 technology designs of the ETP model for each model region and each year, thereby allowing a detailed assessment of emissions reductions by technology readiness level. The contribution of fuel shifting and the energy efficiency of clean energy technologies are allocated to their respective technology readiness level, while activity effects are allocated to the maturity of their enabling technologies (or if not applicable, of their enabling policies). Overall, the methodology allows the identification of the share of emissions reductions from now to 2070 attributable to technologies that are currently mature, at an early adoption stage, under demonstration or still a prototype.

References
  1. Further details on the added national and municipal policies, the elasticities that are used to model transport activity, stock and mode share responses to these policies, and the demand generation module can be found in Annex F of ETP 2016 at http://www.iea.org/etp/etp2016/annexes

  2. See ETP Clean Energy Technology Guide for more information on technology readiness levels.