By Seb Henbest
Lead author, New Energy Outlook
Head of Europe, Middle East and Africa
Bloomberg NEF
Are wind and solar projects just “badly behaved coal plants”? I ask because this perhaps best reflects the argument against the two main renewable energy technologies that critics have put forward in recent years.
Yes, those skeptics admit, wind and solar have become cheaper and will continue to do so. However, they don’t produce electricity at the time people want it, or need it. And that means that the role they can play in the energy system of the future will be strictly limited.
I have news for proponents of the “badly behaved coal plants” theory. Their criticism – when you crunch the numbers to 2050, taking into account the cost dynamics not just of wind and solar but also of batteries – doesn’t stack up.
Every year, Bloomberg NEF runs its New Energy Outlook (NEO) for the future of the world electricity system, taking into account projections for country-by-country demand, technology costs, the transition in transport and many other variables – without assuming any new policy measures. Every year, this modeling exercise seems to produce a big surprise.
This time, with a forecast horizon set for the first time at 2050, NEO’s big surprise is that wind, solar and batteries are set to become far more deeply entrenched in the generation mix of almost every country than anyone has so far thought possible.
Below, I’m going to run you through the dramatic changes that the NEO team of 65 analysts around the world are now projecting. If you have appetite for the full report, then you can find it on the links shown at the end of this article.
Cheap renewables and batteries remake the world’s power systems, with wind and solar producing nearly half of world electricity by 2050
That is the headline message of the NEO 2018. In fact, we claim solar and wind have already won the race for cheap, bulk electricity – it just hasn’t finished playing out yet. In the report, we strip away policy drivers and show how economics alone might deliver a least-cost system filled with clean energy by the middle of the century.
At its heart, NEO is a technology story. It is the story of a shift from resource economics and fuel extraction, to manufacturing, where large-scale production of thousands of small modular components in a competitive environment drives incremental innovation and dramatic cost declines. Perhaps the most iconic example of this cost decline in the energy sector, and some might argue the most important data set in energy economics today is the PV module experience curve. Going back to the U.S. space program in the 1970s, it describes a cost reduction of 28.5% for every doubling of PV capacity – in other words, an exponential decline. Following this curve, the price of PV modules has dropped 83% since 2010.
Wind shows a 10.5% experience curve on a dollar per MW basis. This is less precipitous than PV, but for wind the reduction in unit cost is only part of the story – we are also getting much more energy per megawatt deployed. The average capacity factor of onshore wind has risen from around 20% in 2000, to close to 35% today. This is the result of bigger turbines, taller towers that lift turbines into less turbulent air, computer modeling to better position turbines across the landscape, and more sensors gathering data that can be used to help improve operational performance and reduce maintenance costs, as well as to feed into the development of the next generation of machines.
Solar and wind have already won the race for cheap bulk electricity generation, it just hasn’t finished playing out yet
If we also consider the balance-of-plant and financing costs, and feed these into a cash-flow model, we can calculate the cost of energy in dollar per megawatt-hour terms. Doing this for wind, solar, coal- and gas-fired power shows how competitive these technologies are relative to one another on a levelized-cost-of-electricity (LCOE) basis. Furthermore, if we map each cost component forward, we can construct forecast LCOEs for each technology and we can see two tipping points. The first is when the cost of new wind and PV crosses the cost of new-build coal and gas. The headline here is that, whether we’re talking about coal-fired power in China, or combined-cycle gas turbines spinning in the U.S., well situated and equipped wind farms and solar parks are already as cheap as, or cheaper than, fossil fuel alternatives, almost everywhere. In many places, a wedge opens up as wind and solar costs continue to fall, while fuel prices, weaker-than-expected demand growth and behind-the-meter PV keeps new large thermal plant underutilized.
The second tipping point is when it becomes cheaper to source electricity from new wind and solar than it is to get it from fueling and operating existing coal or gas plants. This takes longer, but is likely to start happening in the mid to late 2020s. What this means is that by 2030 we could see new wind and solar being built without subsidy to undercut the commissioned thermal plants, eroding the latter’s run-hours and revenues.
We conclude that with such steep learning curves, it’s a matter of when and how, not if, wind and solar disrupt electricity systems everywhere.
Batteries…for when the wind isn’t blowing and sun isn’t shining
However, even if renewables get really cheap, there is a problem. Wind and solar are not always available and there are of course times when the wind isn’t blowing and the sun isn’t shining. So how much can wind and solar ultimately do?
To help answer that, we need to think about energy storage, and the analysts at Bloomberg NEF are most excited about lithium-ion batteries. Like PV and wind, these are products of a growing manufacturing industry, economies of scale and incremental innovation. They are being propelled by a burgeoning electric vehicle industry. Our battery pack survey shows an 80% reduction in market price since 2010 – the product of an 18% learning rate. And manufacturing is expanding rapidly. Today we can identify 131GWh of lithium-ion battery manufacturing capacity worldwide. The bulk of that is in Asia, and almost 60% is in China. By 2021, we expect that GWh number to more than triple, with China controlling around 73%. It is clear that China is positioning itself to dominate the global battery market just as it cornered the market for PV technology.
Following the experience curve, we expect batteries to fall another 54% to $96/kWh in 2025, and 67% to $70/kWh by 2030. However there will probably be some bumps along the way – for instance, lithium and cobalt prices have tripled in the past 18 months and are set to slow down cost declines in the near term.
If batteries continue to fall in cost, there will come a point when they too will look competitive against conventional technologies. Comparisons are particularly tricky in the case of batteries, because they have two measurements: their power capacity (MW) and how long they can discharge (MWh). Furthermore, they are most often deployed today to provide fast frequency response in balancing markets. However, if we just consider energy arbitrage – that is, charging the battery when electricity is cheap and abundant, and discharging when electricity is scare and valuable – then we think a standalone, benchmark, utility-scale project will be increasingly competitive by 2025, even if it’s only for a few hours at a time. Rather than charging it from the grid, we can pair a battery with a wind farm or solar park, allowing a fraction of the renewable electricity to be dispatched when the underlying generators are unavailable. In other words, batteries let wind and solar generate when the wind isn’t blowing and the sun isn’t shining. This lets cheap renewables do more and displaces a larger amount of conventional electricity generation.
For gas it’s about value, not volume.
Batteries are an important new source of flexibility for power systems, but we also expect to see time-of-use and turn-down demand response, as well as pumped hydro and new gas peaker plants. We think the latter (open-cycle gas turbines and reciprocating engines) could be an important complement to large amounts of wind and solar – kicking in to help the grid ride out seasonal extremes, when solar resources are poor and the wind dies down for days at a time. Running for so few hours a year, it can make sense to build a peaker gas plant rather than a large, round-the-clock, “baseload” style, combined-cycle gas turbine (CCGT). Today, CCGTs across the world are generally underutilized, with average capacity factors between 20% and 60%. That is well below the 80% or so that their manufacturers and owners would have intended. The trouble is that in many places gas is relatively expensive and gets squeezed between cheap coal and renewables, particularly in OECD markets with weak demand growth. Cumulative CCGT capacity does grow – particularly in the 2020s as cheap gas continues to replace coal in the U.S., and as both Germany and France dial back their nuclear fleets. In the longer term, gas generation becomes critical to meet demand during low-renewables periods, and this sees 730GW of peakers added worldwide between 2025 and 2050.
All this points to the fact that gas-fired power is likely to be very valuable in the future electricity system as a flexible complement to wind and solar. And while this might be a good news story for equipment providers and the owners of gas power plants, it’s a less rosy picture for gas extractors and sellers, with fuel burn up just 14% to 2050 as electricity demand rises 57%.
Businesses drive behind-the-meter solar and batteries
Consumers are likely to play an increasingly central role in the future electricity system, with 7% of global generation in 2050 being done behind-the-meter by PV installed by households and businesses. Some markets, such as Australia (23%), Japan (19%), Mexico (32%) and Germany (15%), are on track to see much more. Consumer uptake is partly driven by cheaper technology, but can grow strongly once penetration reaches a critical point and imitation effects open up the mass market. To date, almost all small-scale PV adoption has been driven by generous upfront subsidies or by net metering policies that allow consumers to get paid for kilowatt-hours fed back into the grid. Strip away these subsidies, and at present PV only achieves “socket parity” – where the value of the system is greater than the cost of the system – in a small number of markets such as Germany and Italy where there is a particularly good solar resource, high consumer tariffs, or both. Over time, however, cheaper PV is set to make it economic for households everywhere even without net metering or other policy support. Rooftop PV is particularly attractive for businesses, which potentially have much higher self-consumption than households and hence reach socket parity much sooner. From around 2025, small-scale battery systems start to get deployed alongside PV as the additional battery capex is paid off by greater self-consumption, in turn allowing people to get more value from their PV systems.
EVs and A/C help renewables go deeper
The demand side is changing too. By 2040, we expect there will be more than 60 million light-duty electric vehicles on the road, making up 33% of all passenger cars. This growth means that electrified transport makes up around 9% of world electricity demand in 2050. The latter figure, however, masks a wide spread in adoption rates, with – on the high side – countries like Germany seeing as much as 24% of electricity demand in 2050 coming from EVs. At this level of penetration, it’s not just important how many EVs there are, but when they charge. We know that consumer uptake is driven by imitation and this makes it clumpy. That means that if EV owners all plugged in when they got home, they’d likely break the distribution grid. Instead, a significant fraction of EVs will need to be plugged in when stationary – which we know is 90-odd percent of the time – and charge when cheap renewables are abundant. We’re already seeing special time-of-use tariffs emerge for EV owners in the U.S. and Europe, and we expect this trend to continue. The relative predictability and super-low marginal cost of electricity from PV suggests that much of this dynamic EV demand might flow toward the middle of the day.
EVs are one major dynamic load, but it’s likely there will be others. We also expect there to be further growth of air-conditioning in emerging markets across Southeast Asia, India and South America. This is particularly pronounced in the commercial sector, where large A/C loads during the day can be supported by cheap PV.
More generally, we think that growing flexibility in demand will fundamentally change the gross load shape. No longer will supply have to dispatch and ramp to meet changes in demand, but demand itself will shift to meet cheap renewables generation, helping wind and solar to achieve higher shares of the electricity mix.
Back-up and curtailment are a feature, not a bug
If we believe that PV, wind energy and battery technology will keep getting cheaper; and if we believe that the households and businesses will play an increasingly central role in a more distributed system; and if we believe that EVs are coming and that demand will get more and more dynamic and responsive to price; how might a system like this fit together?
The first thing to know is that when we allow all this logic to play out in the New Energy Outlook to 2050, we get very high penetration of renewable energy. In Japan and India we get over 60% wind and solar, in France, Mexico and Germany we get over 70%, and in the U.K. and Australia we get over 80%. These are high numbers. Higher than we expected when we started this year’s modeling. And when we looked into the dynamics a simple truth emerged – in a high-renewables system, curtailment and back-up are a feature, not a bug. Perhaps this is best illustrated using seasonal extremes. At one end of the spectrum are the high-renewables days – those with lots of solar and lots of wind. We know we can get strings of these days in a row. By 2050, these periods see way more renewable energy than we have demand. So we need to turn down and curtail output. This is not ideal for asset owners but not a major problem at a system level. At the other end of the spectrum are the low-renewables days. Similarly, we can get strings of these in a row where the solar resource is poor and the wind isn’t blowing. During these periods, it doesn’t matter how many batteries you have, the wind and solar assets simply can’t produce enough electricity to meet demand and you need to ramp up your back-up capacity. The point is that these extremes don’t happen every day. Most days are somewhere in the middle, with low curtailment and small amounts of firm capacity spinning up overnight to meet demand. This system is least-cost because wind and solar get cheap enough that we want to max them out, knowing that on some days they won’t be available and on others there will be too much, but in general they will provide cheap, bulk electricity.
By 2050, we’re painting a picture of an electricity system utterly reshaped around cheap wind, solar and batteries. These technologies provide bulk electricity and are supported by thermal plants that run at low overall capacity factors, but can be dispatched when needed.
Coal is the biggest loser
The other major change in the system is the demise of coal, which falls from 37% of global electricity today to just 11% in 2050. There are several reasons for this. In the U.S. cheap gas continues to displace coal, and in Europe sluggish electricity demand growth, the carbon price and renewables growth forces all but the dirtiest and cheapest lignite out of the mix. In China, coal capacity peaks in 2025 and generation in 2030 as the new-build coal pipeline is exhausted and older plants retire or are pushed out by renewables and batteries. In India, coal generation peaks in 2033 as cheap renewables dominate capacity additions and soak up new demand. In short, coal loses the market for bulk electricity to cheaper renewables, and the market for round-the-clock availability to more flexible gas, which better complements variable wind and solar.
Gas alone can’t get us to 2 degrees
The demise of coal and growth of renewables lowers the carbon intensity of electricity generation all over the world. Power sector emissions peak in 2027, the same year as coal-fired electricity generation, before falling at 2% per year to 2050. However none of this happens fast enough, and the power sector remains on track to outstrip a two-degree emissions trajectory by some margin. Even in the unlikely event the world agreed to a coal moratorium and shuttered all plants by 2035, the additional gas needed to ensure system security would still produce too many emissions.
Final thoughts
The challenge for regulators and policy-makers is twofold. First they need to figure out how to pay for the bulk, cheap renewable energy units that run at close to zero marginal cost and all generate at once, cannibalizing their own energy prices. At the same time they need to figure out how to pay to keep conventional thermal plants online and available, and to build the new facilities needed to provide back-up as the older plant retire. Second – if they are serious about addressing climate change – they need to accelerate the transition to wind, solar and batteries and start direct more money and attention to the next phase of low-carbon technology that can offer a zero-emissions substitute for gas to balance the system and bridge the seasonality gap.
BNEF’s NEO 2018 report, written by Seb and a team led by Elena Giannakopoulou, head of energy economics at BNEF, can be read by clients here. Highlights can be viewed on the BNEF corporate website at https://about.bnef.com/new-energy-outlook/.