9/2020

Energy Transition Outlook 2020

Your questions answered

The DNV GL Energy Transition Outlook team

DNV GL's Energy Transition Outlook 2020 was launched at a virtual event in September. Attendees had the opportunity to ask questions via the broadcast platform. The Energy Transition team has reviewed the recurring themes and answered the most frequently asked questions.

Q: What is the ‘Energy Transition Outlook’?

A: In the Energy Transition Outlook, DNV GL provides its best estimate to 2050 for the entire world’s energy system, and for ten world regions. In addition to the main publication on the global energy transition, DNV GL publishes three companion reports – Oil and Gas, Maritime, and Power Supply & Use – focusing on implications for each of the main industries we serve.  

The Outlook, published annually, is based on DNV GL’s own independent energy model, which models and forecasts the world’s energy demand and supply, and the exchange or trade of energy between ten world regions. The forecast data can be accessed on DNV GL’s industry data platform, Veracity.com

Q: Why does DNV GL publish an Energy Transition Outlook?

A: As a world-leading provider of risk management and assurance services, we have a strong footing in both the fossil fuel and renewable energy industries. This, coupled with being fully owned by an autonomous, independent foundation, allows us to take an impartial view of the energy future. 

Approximately 70% of DNV GL’s business is related to energy in one form or another. For us, and for many of our customers, the energy transition itself is a great source of risk – and opportunity. Understanding the nature, timing and industry implications of the transition is therefore a critical strategic exercise, both for us and our customers.  

Our Outlook draws on DNV GL’s broad involvement across entire energy-supply chains, spanning complex offshore infrastructure, onshore oil and gas installations, large- and small-scale wind, solar, storage, and energy-efficiency projects, electricity transmission and distribution grids, and the seaborne trade in fossil fuels.  

Q: What do we mean when we claim to present ‘a best estimate’ for the energy transition through to 2050?

A: DNV GL has produced a model-based forecast, not a scenario. Other commentators on the energy future have developed many scenarios that contrast possible futures and outcomes. But too many scenarios cause confusion, and – understandably – our customers seek our expert view on the likely energy future.  

Given our extensive research capacity and our broad exposure to the energy industry across entire energy supply chains, we are well placed to make an independent, technology-centric forecast of the energy future. 

Our aim is to produce an objective and balanced view of the future that could come to be regarded by planners as a base or central case. The Outlook presents our ‘best estimate’ of the energy transition and energy future through to 2050, given expected developments in policies, technologies and associated costs.  

That we present one forecast does not mean we are certain of how the future will unfold. In the Outlook, we discuss sensitivities to key parameters. 

Key findings - 2020

Q: What are the key findings in your 2020 forecast?

A: Highlights from our 2020 report include: 

COVID-19 reduces energy demand by 8% and places peak emissions behind us

The impact of COVID-19 is large, both short and long-term. This year, global energy demand will reduce by 8%, CO2 emissions also reduce by 8%, and global CO2 emissions are never likely to be as high as they were in 2019.

Rapid electrification, dominated by solar PV and wind, transforms the energy mix

  • The global energy system is electrifying rapidly in all sectors, and global electricity production more than doubles over the next 30 years. 
  • Almost 1/3 of all electricity in 2050 will be generated by wind, and almost 1/3 by solar PV. 
  • 62% variable renewable share by 2050 

Decarbonization of hard-to-abate sectors remains too slow; we are set to miss the Paris Agreement targets 

Sectors than cannot be electrified are decarbonizing too slowly, which is the main reason why humanity will miss by a clear margin the Paris agreement target of “well below 2°C, striving for 1.5°C” 

Existing technologies can deliver the 1.5°C ambition, but stronger policies are needed to scale uptake

The technologies that already exist are capable of delivering a future within the bounds of the Paris Agreement target – no new breakthrough technologies are needed, although promising solutions may come to help. Much stronger policies and regulations are needed to scale solutions in all sectors, particularly hard-to-abate sectors that have yet to be subjected to targeted policy measures.

Q: What is the effect of COVID-19?

A: COVID-19 has had a significant effect on the energy system. As the global economy shrinks by around 6% this year (compared to a normal 3% growth), demand for energy will reduce accordingly, 8% in our forecast. And although the economy will likely rebound somewhat in 2021, the size of the global economy will stay “lower forever” than in a non-COVID case, and the same will be the case for global energy use, fluctuating 6-8% lower than a non-COVID case through to mid-century. 

On top of the changes caused by a smaller economy come behavioural changes in terms of less air travel, less commuting and office work. Some of these changes are likely to last beyond the COVID crisis. 

Lower energy use also lowers emissions, and we notably find that peak CO2 emissions is likely to have passed in 2019. The same is true for peak oil. The difference, however, is limited, and COVID has a limited long-term effect on emissions reduction. 

We find today that the pace of the transition is largely unchanged due to COVID, but the direction of future stimulus packages has the potential to slow or speed up the transition. 

Finally, everything related to COVID-19 has high uncertainty as we are still in the middle of the pandemic at the time of launching the 2020 Outlook. When we update our forecast in 2021, we will know much more. 

Q: What is meant by peaking energy demand?

A: We forecast a significant (8%) drop in 2020 in global energy demand due to COVID-19, and it will take just over a decade to reach the same level as the pre-pandemic energy demand. Final energy demand* thus peaks at 440 EJ/yr around 2032, and then declines slightly towards mid-century to 420 EJ/yr. 

*Final energy demand is the sum of energy supplied to the final consumer for all energy uses. In our Outlook it is disaggregated into transport, buildings, manufacturing, non-energy (e.g. lubricants and plastics) and ‘other’ (e.g. agriculture, forestry, military and other small categories).

Q: Why does energy demand peak?

A: Demand peaks at the point where the energy intensity (unit of energy per unit of GDP growth) of the world economy starts to decline at a faster rate than economic growth. In our forecast period, energy demand initially slows in line with slowing rates of population and economic growth. After 2035, efficiency gains outpace economic growth and demand declines. Efficiency gains are related to technology innovation but are mainly the result of pervasive electrification. 

Q: Why does global primary energy supply peak before final energy demand?

A: The difference between primary energy supply* and final energy demand is energy losses, and the largest category is conversion losses in fossil-fueled power plants. Primary energy supply peaks in 2032. It peaks before demand as there is a steady reduction in energy losses when the electricity sector gradually moves to renewables. As losses reduce, the difference between primary and final energy reduces, and primary energy supply therefore peaks before final demand. 

*Primary energy supply is the amount of energy supplied in the form that has not been subjected to any conversion or transformation process. For fossil fuels, this corresponds to the quantity of production. For renewables, the electricity produced is the primary energy. For details see the Energy Transition Outlook main report pp68-69.

Q: Energy efficiency plays a key role in the DNV GL forecast – why?

A: Energy efficiency plays a decisive key role in our forecast and in the energy transition (even more than the shifts in the mix of energy sources). The world’s energy intensity (units of energy per unit of GDP) has been declining by, on average, 1.1% per year for two decades. We calculate this will double, to an average annual decrease of 2.3%. 

  • The main reason for this is accelerating electrification of the energy system. Electricity’s share of final energy demand grows from 19% in 2018 to 41% in 2050. 
  • In a more electrified world energy system, efficiency is higher and energy losses lower, because electric processes have smaller losses than their non-electric alternatives, i.e., using electricity rather than fossil fuels have lower heat losses. 
  • With the renewable share in electricity growing, energy intensity benefits from there being lower losses in power generation from renewables than from fossil fuels. The effect of greatly expanding electrification will be superimposed on the ongoing gradual effect (from insulation, improvements of combustion engines, replacement of light bulbs etc), more than doubling the overall improvements in energy intensity. 

Electrification of demand sectors:

  • The efficiency trend is boosted by electric vehicles (EVs) becoming mainstream in the transport sector. They are greater than 3 times more efficient than internal combustion engine vehicles. With electrification, the transport sector efficiency improvement is 1.4%/yr over the forecast period. Adding the efficiency improvements of fossil-fuel engines, the overall efficiency improvement of the transport sector becomes 2.6%/yr. 
  • The other demand sectors (buildings and manufacturing) do not electrify that quickly. We expect the annual efficiency improvements to be 1.1%/yr for buildings and 1.2%/yr for manufacturing. 

Q: How many electric vehicles will there be in the future?

  • Despite a growth in car sharing and semi-automated driving, the world’s car fleet will expand by 75% by 2050. 
  • All the three road-transport subsectors are electrifying rapidly, first in the two and three-wheelers segment, then passenger segment and, finally, in the commercial vehicle segment. By 2050 there will be: 
    • Passenger combustion: 420Mn 
    • Passenger electric: 1.3Bn 
    • Commercial combustion: 220Mn 
    • Commercial electric: 320Mn 
    • Two-wheeler combustion: 10Mn 
    • Two-wheeler electric: 1.3Bn 
  • By 2032, electric vehicles (EVs) will represent half of all passenger new car sales globally, but with large regional variations. Already in the late 2020s, we will see the 50% sales mark passed in Europe, North America and Greater China. The fleet 50% share (i.e. share of vehicles on the road) will follow 8-9 years later. 
  • By 2037 sales of EVs in the heavy vehicle category will be 50%. This is somewhat slower than for passenger cars, owing to the bigger batteries and higher costs involved, although we do expect a rapid electrification of municipal bus fleets and local delivery and garbage trucks. 
  • The two and three-wheeler category – which will account for more than a billion vehicles in Asia by 2030 – is electrifying fast. The 50% share of sales is as early as 2024 for the 2&3-wheeler segment, and five years later, half of the fleet will be electric. 

Q: Your forecast indicates a surge in renewables – how will that affect future energy investments?

A: We forecast investment levels to shift and to increase dramatically. 

  • Capital expenditure (capex) on both renewable generation and grids is increasing. 
  • Fossil capex has already peaked and is on a downward trend towards 2050. 
  • By 2036, capex on non-fossil and grids will be larger than fossil capex. 
  • By 2050, 55% of the global energy expenditures will be capex for renewables and grids, up from 22% in 2018. 

Q: How high is the uncertainty in the results?

A: We acknowledge the significant uncertainty that exists when trying to forecast the energy future in a 2050-time horizon. 

  • We have analysed how our forecast deviates from the base-case prediction under different parameter assumptions (sensitivity analysis). 
  • Although some changes in our assumptions could slow down the pace of transition, none of the sensitivities that we discuss in the Main Energy Transition Outlook report alters the main conclusion that a rapid energy transition is underway with electrification and decarbonization as key pillars. 
  • Another robust conclusion is that we are not on track to meet Paris Agreement ambition on “well below 2 degrees, striving towards 1.5 degrees”. 
  • The uncertainty is highest on the policy side; it is indeed a challenging task to forecast what will be the future energy policies. 
  • We do not quantify the uncertainty in our forecast, giving e.g. standard deviations, or confidence intervals of the results we forecast. 

Q: Do you have results at country-level?

A: No. Our analysis focuses on a global forecast and 10 regional forecasts. To calibrate our model, we use country-level historical data which is weighted and aggregated to produce regional figures. Our non-linear system dynamics model then produces our future forecast. 

Q: What has changed since last year – Outlook 2020 vs. 2019?

A: The Energy Transition Outlook model has been refined for 2020 with more detail as well as updated input data and assumptions. The main conclusions from 2019 are largely unchanged in the long term, however the effect of COVID-19 will have both short, as well as, long-term effects (see also question #5 on COVID above). 

Other changes include policy updates, notably the EU Green Deal which seems quite robust even to the COVID-19 economic downturn, and this has made us increase CO2 price expectations in Europe, with subsequent effects on higher hydrogen and CCS forecasts. 

Further, we have updated technology assumptions on a number of key technologies, following the latest developments. 

The model also includes further details on certain areas in 2020, e.g. a split of offshore wind into fixed and floating offshore wind. 

Q: How come energy demand peaks while population and economy are still growing?

A: The key explanation is falling energy intensity of the global economy, i.e. the energy used per unit of GDP (economic output) will improve more quickly than the rate of global economic growth in the next three decades. That is why we are approaching an era with peak energy. 

This should not be confused with a lack of energy or decline in any way. It simply means that we are getting more out the energy that is produced – that the work done by the energy produced is greater. 

  • Example: the average person in Europe currently uses 139 gigajoules of energy per year. 83 GJ is forecast in 2050 
  • Example: the average person in Sub-Saharan Africa consumes 27 gigajoules of energy. In 2050 that number will be close to the same (25) gigajoules per capita but with joules consumed much more efficiently. This also explains why it is possible to increase the GDP per person (from USD 3,900 in 2018 to USD 8,100 per person in 2050) without an ever-growing energy demand. 
  • Efficiency improvements: replacing traditional cooking methods, burning biomass with efficiencies around 5-10%, and efficiency further boosted with switch from coal to gas or gas to electricity. For example, 1% of the primary energy mix in Africa in 2050 will be off-grid. However, that will enable almost all households in Sub-Saharan Africa to have electricity access by 2050. 

Q: What is your climate change forecast?

A: In spite of decarbonization in all regions and a decrease in the carbon intensity of economic growth (tonnes of carbon dioxide per terajoule of primary energy consumption) in all regions, and most rapidly in China; this is not happening at a scale and speed fast enough to meet the Paris Agreement ambitions. 

  • The 1.5°C budget will be exhausted already in 2028. The 2°C budget will be exhausted in 2051, and emissions continue far into the second half of the century. 
  • This implies a 2.3 degree warming compared to the Paris Agreement goal of staying ‘well below’ 2 and 1.5-degree targets. 
  • ‘Closing the Climate Gap’ will require a simultaneous mix of extraordinary measures, such as more energy efficiency, more renewables and more carbon capture and storage (CCS). 

Q: Why is the uptake of hydrogen relatively low?

A: Hydrogen uptake is negligible today and starts to grow from 2030s onwards, reaching 5.5% of global energy use in 2050, but with large regional variations (e.g. 15% in Europe). We model both blue hydrogen (from natural gas with CCS) and green hydrogen (from electrolysis) in our analysis. 

The main reason for relatively low hydrogen uptake is high costs; and the main reason for high costs is lack of support such that deployment is low and technology costs therefore remain high. Although the costs will start to come down over the coming years, they are still at a level where uptake based on competitive cost remains low for another decade. Uptake based on mandate and governmental incentives is emerging but is yet quite scattered. 

Q: What is your finding on CCS?

A: Carbon capture and storage (CCS) is a technology that has existed for many decades but has not received sufficient policy support for technology costs to come down to a competitive level. 

Uptake of CCS remains low until the mid-2030s in our results before it starts to grow relatively sharply, reaching just above 2Gt in 2050, equalling almost 15% of energy-related emissions at that time. Around half of the CCS is from blue hydrogen and half from post-combustion CCS in power and industry. 

CCS uptake in the first decade is based on scattered plants financed by economic incentives. Towards 2040 we will see technology costs falling to a level that starts to be comparable with carbon prices, particularly in Europe. As technology costs continue to fall and carbon prices increase, CCS in the 2040s starts to be competitive with unabated emissions in more regions. 

Methodology

Q: Which methodology has been applied?

A: Our core Energy Transition Outlook model is a system dynamics feedback model, implemented in Stella software. It is non-linear and contrasts with simpler econometrics/statistics forecasting typically used. It models the world’s energy system covering energy demand and supply globally, and the use and exchange of energy within and between ten world regions. The model incorporates the entire energy system — from source to end use — and simulates how its components interact. 

Our Outlook model includes all the main consumers of energy (buildings, industry, transportation and feedstock) and all sources supplying the energy. Population and productivity are the two main drivers of the demand side of the energy system in the model. 

The model uses a merit-order cost-based algorithm, based on forecasted decreasing production costs in its energy sectors (power, oil and gas), to drive the selection of energy sources / production technology over another through time. 

The evolution of the cost of each energy source over time is therefore critical and learning-curve effects are taken into account. We have complete dynamics of power costs, determining which power stations in a region will increase/decrease their market shares. Cost learning, driven by accumulated capacity, is a key input here (because the cost of a technology decreases by a constant fraction with every doubling of installed capacity). Similarly, oil and gas costs reflect learning and resource availability and help determine regions’ and technologies’ market shares.

Q: What are your sources?

A: There are multiple data sources and methods used for the model, and the Outlook report. Historical data on variables like energy use is obtained from primary data sources like IEA, IRENA, UN, Platts, Clarkson, Rystad and national statistics. Data on parameters such as cost and efficiencies were mostly estimated by BA colleagues based on similarly available data, other reports, and literature. 

On topics that are not core to DNV GL, we use outside research (e.g. on population). For productivity growth, we have our own expert models. We also have a global Energy Transition Collaboration Network with around 40 experts globally. More than 100 colleagues at DNV GL are actively involved globally. 

Q: Are details on your model available?

A: We communicate our assumptions, inputs and models in our main report; and we wish to be clear about the parameters used and how they are related. Our aim is to present a transparent model, not a black box. As a complement to our report, we have issued a ‘model documentation report’ on the DNV GL Outlook model. Though primarily designed for our own team to document what we have done, we also make this documentation report available to those stakeholder communities interested, upon request

Q: How is policy incorporated in the modelling and in the forecast?

A: While technology improvements with efficiency gains and the falling costs of renewables are central to the forecast energy transition; the energy future will also be influenced by political and public issues. 

In general, we expect an important kick-start from policies, such as world-wide emission reduction targets and policy mechanisms for their achievement. 

Policy considerations influence our Outlook in various ways: a) supporting technology and activating markets that close the profitability gap for renewable/low-carbon energy technologies competing with existing technologies; b) restricting the use of inefficient or polluting products/technologies by means of technology requirements or standards; or c) providing economic signals – such as price incentives to reduce carbon-intensive behaviours. 

Examples of policies explicitly reflected in our model include renewable support, carbon prices, and fuel/efficiency standards.

  • Carbon prices will increase and reach an average of between USD 20 and 80/t CO2 by 2050, depending on region. 
  • Renewable energy support will decline as the gap between the expected profitability of renewables (expected received price - levelized cost of energy) and the profitability of the most profitable conventional technology in the same region shrinks. 
  • Energy efficiency improvements: Existing and planned standards and regulation for energy use and efficiency improvements in buildings, transport, and industry sectors are incorporated. The energy mix in maritime transport also reflects the expected implementation of environmental regulations, e.g. IMO 2050 carbon reduction strategy of at least 50% by 2050. 
  • Air-pollution interventions are reflected by a cost proxy for control measures. In general: Policy incentives, mandates, and subsidies accelerate the speed of technology uptake, in turn affecting technology cost-learning rates (CLRs). General approach: country-level data are translated into expected policy impacts, then weighted and aggregated to produce regional figures for inclusion in our analysis. 

Q: What are your GDP assumptions?

A: Future GDP is driven by population and productivity and is the key driver of energy demand. On future GDP - there is no central or main source for global GDP forecasts. 

We produce our own GDP forecast based on productivity figures (GDP/capita = productivity, the output achieved per capita). In the relationship between GDP per capita and growth rates, we acknowledge that high GDP per capita is correlated with lower growth as countries enter the third sector of the economy. Productivity growth slows down as economies mature (improvements increase quality rather than the amount of output). 

Using this methodology and multiplying regional productivity dynamics with respective population forecasts, we see a 100% global increase in global GDP from 2018 to 2050 reaching 269 trillion by mid-century. Average annual growth rate is 2.2%/yr, which includes COVID-19 effects, with large regional variations. Our forecast is in line with projections from McKinsey and PwC. 

Q: What are your population assumptions?

A: On population growth, we use the IIASA/Wittgenstein Centre for Demography and Global Human Capital in Austria models, because they seem to better consider how urbanization and rising female education levels are linked to declining fertility rates. 

Using the IIASA models, but adjusting for a lower education update and faster population growth in Sub-Saharan Africa (which lags other regions in the expansion of education and socio-economic development), gives us a global population in 2050 of 9.4 billion. 

This is some 3.5% lower than the UN median forecasts (The UN has been criticized for not taking country-specific education levels into sufficient consideration).  

DNV GL's outlook compared to other energy forecasts

Q: Why does oil demand peak earlier in DNV GL’s Outlook compared to other central scenarios?

A: The main reason for this is that we forecast a faster electrification of the road transport sector, which in turn will replace oil use with electricity use, and the electricity is not based on oil. 

Other reasons include smaller growth in feedstock, and somewhat lower overall GDP growth. 

This year our COVID-19 oil demand reduction is somewhat larger, and rebound somewhat slower, than some other forecasts. 

Q: How is your forecast different from other central scenarios and why?

  • Other energy forecasters use higher GDP growth figures, resulting in higher energy consumption/demand estimates. The difference is partly due to them using the UN median population forecast, which is 9.8 billion for 2050, 3.5% higher than our 9.4 billion population forecast (IIASA/Wittgenstein Centre). 
  • The strong electrification of the transport sector we forecast leads to a higher electrification share. As the number of electric vehicles grows exponentially, and as these vehicles have higher energy efficiency than internal combustion engines, it contributes to peak energy - and lower oil use. 
  • Our Outlook typically has more gas and less oil than central scenarios. Projections coming out of the energy industry, e.g. the International Energy Agency (IEA) and BP, see continued energy demand growth under all scenarios and no peak in oil demand until the 2040s. They also see a much slower uptake of electric vehicles: DNV GL expects 50% of all new passenger vehicles sold to be zero-emission from 2032. 
  • Our Outlook typically has low nuclear share, with us forecasting uncompetitive nuclear costs. 
  • Our Outlook finds that the shift from fossil fuels to renewable energy will be quicker than anticipated by other industry projections, leading to a huge shift in investment flows. 
  • Many forecasts have very high amounts of CCS uptake, which we don’t. As the technology is expensive, it only starts to scale from 2035 onwards in our forecast. Still we have increased CCS this year compared to previous years, due to higher carbon price expectations, particularly in Europe. 
  • Our Outlook does not include negative emission technologies, such as BECCS and DACCS. These CO2 removal technologies are part of almost every Paris-compliant scenario but are largely unproven. We do comment that it is indeed a high-risk approach to push solutions into the future, but we also acknowledge that if we are to close the gap to 2°C or 1.5°C, negative emission technologies will have to be part of the solution. Not having quantified this is also one of the reasons we typically have lower amounts of CCS that many other forecasters. 

Subscribe to the latest Energy Transition insights from DNV GL

Subscribe here