Energy Transition Data

Wind and solar: curtailments over time?

Wind and solar curtailments average 5% across different grids that we have evaluated in this data-file, and have generally been rising over time, especially in the last half-decade. The key reason is grid bottlenecks. Grid expansions are crucial for wind and solar to continue expanding.

Curtailments occur when wind and solar are capable of generating electricity, but operators cannot dispatch that electricity into the grid.

This data-file tabulates curtailment rates in California, Australia, the UK, Germany, Spain, Chile and Ireland, averaging c5% in 2022.

The main reason for curtailment is bottlenecks in the grid — i.e., moving renewables from points of generation, to points of unmet demand — rather than renewables having saturated total grid demand.

In a recent research report into the ultimate share of renewables in power grids, we calculated that based on their statistical distributions, solar would only start meeting 100% of a grid’s total demand around 1% of the time when solar was providing c30% of the total grid, while wind would only start meeting 100% of a grid’s total demand around 1% of the time when wind was providing 40% of the total grid (see below). We are not at these levels yet, in the countries in this sample.

Power grids: when will wind and solar peak?

The bottleneck is power grids. You might have a 100MW grid, composed of 10 x 10MW inter-connected nodes, and the issue causing curtailment is trying to flow 20MW through a 10MW node.

This confirms that meeting the theoretical potential of renewables (per the note above) requires vast grid expansion, and indeed, countries that have seen YoY reductions in curtailment rates have often achieved this by building new interconnectors.

Renewables: the levelized cost paradox?

Second, it gives a new lens on energy storage. Battery deployments can absorb low-cost renewables and prevent curtailment, by circumventing grid bottlenecks, especially for renewables developers who fear bottlenecks in the grid will be persistent.

Renewable-heavy grids: dividing the pie?

California’s grid: wind and solar statistical distributions?

This data-file aggregates wind and solar statistical distributions, plotting solar generation and wind generation, every 5-minutes, across California, for the entirety of 2022, in order to understand their volatility and curtailment rates. The data suggest that wind and solar will most likely peak at 50-55% of renewables-heavy grids.

How will ramping up wind and solar increase volatility of other generation sources, lower the utilization rates of the overall grid, and where will wind and solar naturally peak?

We can answer these questions by evaluating 100,000 x 5-minute intervals of power generation, from California’s power grid in 2022, then calculating the statistical distributions of wind, solar and total demand.

Our key findings are in our recent research note, assessing where will wind and solar peak? (see below).

Power grids: when will wind and solar peak?

As always, we publish the data behind our analysis, to give transparency on the methodology, and assumptions can be stress-tested.

Tabs ending _Averages plot the average load across different five-minute intervals, across the year, showing wind and solar statistical distributions, and how they vary throughout the day, in different months.

Tabs ending _DitsributionFlex plot the distribution of each time series, by percentile, across the entire year.

The analysis then asks: if the future distribution matches today’s distribution, how much of a hypothetical ‘constant’ grid load could wind, solar and wind+solar supply, at different curtailment thresholds.

If we assume that supply will stop growing when marginal capacity additions are likely to incur 30-60% curtailment rates, this suggests that solar would most likely peak out at 35-40% of a grid, wind would most likely peak out at 60-70% of a grid, while a mixture of wind and solar would most likely peak out at 50-55% of a grid.

Please see our broader research into power grids, into California’s grid and for similar statistical analysis of wind generation elsewhere, we have run similar analysis for the UK power grid.

California power generation over time?

California’s power grid ranges from 15-61GW of demand. Utility scale solar has almost quadrupled in the past decade, rising from 5% to almost 20% of the grid. Yet it has not displaced thermal generation, which rose from 28% to 36% of the grid. We even wonder whether wind and solar are entrenching natural gas generators that can backstop their daily, weekly and even seasonal volatility.

This data-file aggregates descriptive statistics into California’s power grid, plotting California power generation over time, looking across 150MB of CAISO data into solar, wind, nuclear, hydro, imports and thermal generation, every five minutes, for almost a decade.

Over the past decade, California’s power grid has averaged 26GW of load, but the demand is highly variable, troughing at 15GW in April-2022 and peaking at over 61GW in July-2021. Summer demand is almost 50% higher than winter demand due to air conditioning.

Utility-scale solar is the biggest change in the generation mix, rising from 5% of the grid to almost 20% of the grid over the past decade. Solar generation is volatile. The average load from solar was 4GW in the trailing twelve months, but c40% of the time, solar generation is zero, while 25% of the time it is above 10GW and 5% of the time it is above 13GW.

Seasonality also matters for solar. Solar generates around 2x higher output during the summer months than the winter months. Similar charts for wind are in the data-file.

Thermal generation has not necessarily been displaced by the rise of solar and wind in California’s grid, but arguably, entrenched, as gas-fired power plants can rapidly ramp up production, after the sun sets, or in non-sunny months.

The increasing volatility of California’s power prices rewards natural gas generators, but is not high enough to bring longer-duration battery storage into the money.

Total thermal generation has actually risen from 28% of the grid in 2017 to 36% of the grid in 2023, albeit the picture is somewhat distorted by annual fluctuations in hydro output and the change in imports, tabulated in the data-file.

Is California’s thermal generation base backstopping its renewables? There is also evidence of thermal generation plants being run more flexibly, to ‘backstop’ solar and wind in California’s grid, with rising interquartile ranges, absolute ranges and deviations between upper-decile and lower-decile generation.

Full descriptive statistics are in the data-file, covering all of California’s main generation sources — solar, wind, nuclear, hydro, imports and thermal — and the proportionate share of each one. For each month, we plot each generation source’s minimum output, 10th percentile, lower quartile, median, mean, upper quartile, 90th percentile and maximum output.

Thermal energy storage: cost model?

This data-file captures the costs of thermal energy storage, buying renewable electricity, heating up a storage media, then releasing the heat for industrial, commercial or residential use. Our base case requires 13.5 c/kWh-th for a 10% IRR, however 5-10 c/kWh-th heat could be achieved with lower capex costs.

Thermal energy storage solutions aim to help integrate solar and wind into power grids, by absorbing excess generation that would otherwise be curtailed, and then re-releasing the heat later when renewables are not generating.

Different storage media are compared in one of the back-up tabs of the model. However, one-third of the companies in our thermal energy storage company screen are pursuing molten salt systems, hence our thermal energy storage model focuses on this option.

In our base case, the cost of thermal energy storage requires a storage spread of 13.5 c/kWh for a 10MW-scale molten salt system to achieve a 10% IRR, off of $350/kWh of capex costs. Costs are sensitive to capex, utilization rates, opex, electricity prices and round trip losses. The sensitivities can be stress tested in the data-file.

Capex costs of thermal energy storage may be reduced below our base case estimate, which has been built-up using the same input assumptions as our broader battery cost models. Larger systems require proportionately more storage material, larger tanks, and more insulation. But other lines in the capex build up do not change, and hence these costs deflate in MWH-terms.

The round-trip efficiency of thermal energy systems can also be higher than we might have naively expected, possibly in the range of 85-95%. The physics is modeled from first principles in other back-up tabs of the data-file. As a generalization, a large and well-insulated thermal energy storage system loses 1-2% of its stored heat over the course of 24-hours.

The full data-file contains the workings behind our recent deep-dive into thermal energy storage. We have also included similar estimates for residential-scale storage, adding an electrically heated hot water tank to absorb excess renewables, which looks simple and can be highly economical. Please download the data-file to stress test all of our numbers.

Renewable grids: solar, wind and grid-scale battery sizing?

How much wind, solar and/or batteries are required to supply a stable power output, 24-hours per day, 7-days per week, or at even longer durations? This data-file stress-tests grid-scale battery sizing, with each 1MW of average load requiring at least 3.5MW of solar and 3.5MW of lithium ion batteries, for a total system cost of at least 18c/kWh.

Start by modelling a power demand curve. Then model how much wind or solar would need to be installed to provide this electricity demand across a comparable timeframe. Then model how big a battery is required to move the renewables to align with the timing of the power demand curve. This data-file works through the maths, for different batteries, including their round trip efficiencies, and their costs.

The minimum possible requirement for a fully solar-powered electricity grid is that each 1MW of load requires 3.5MW of solar modules and 3.5MW of lithium ion batteries with daily charging-discharging, in a location where every day is perfectly sunny, with no clouds, and no seasonality, for a total levelized cost (LCOTE) of 18c/kWh.

Introduce volatility into the weather pattern, and the requirement for a fully solar-powered grid is that each 1MW of average load requires 5MW of solar modules and 9MW of lithium ion batteries with full charging-discharging every 1.5 days on average, and a total levelized cost (LCOTE) of 35 c/kWh. For more detail, please see our data-file into the volatility of solar generation.

Introduce seasonality in the weather pattern, with 50% lower solar output in winter versus the summer, and the requirement for a fully solar-powered grid is that each 1MW of average load requires 6MW of solar modules and a somewhat insane 235MW of lithium ion batteries with full charging-discharging every 70-days on average, for a total levelized cost (LCOTE) of 800c/kWh. Which is also somewhat insane.

Wind numbers are more demanding than solar numbers, all else equal, because the sun rises and sets daily (helping the utilization rate of the batteries), while wind can incur 2-3 windy days followed by 2-3 non-windy days (hurting the utilization rate of batteries). For more detail, please see our data into the volatility profile of wind generation.

Redox flow batteries are particularly helpful for integrating larger shares of renewables, and are modelled to result in total system costs that are c50% lower than using lithium ion batteries at grid scale. Please see our deep-dive research note into redox flow batteries.

This data-file provides underlying workings into renewable asset sizing, grid-scale battery sizing and total system costs for our recent research into renewables’ true levelized cost of electricity (LCOTE).

Top Public Companies In Energy Transition

Some of the top public companies in energy transition are aggregated in this data-file, looking across over 1,000 items of research into the energy transition published to date by Thunder Said Energy.

The data file should be useful for subscription clients of Thunder Said Energy, if you are looking for a helpful summary of all of our research to-date, how it reflects upon public companies, and links to explore those companies in more detail, across our other research.

Specifically, the file allows you to filter different companies according to (a) listing country (b) size — i,e., small-cap, mid-cap, large-cap, mega-cap (c) Sector — e.g., energy, materials, capital goods, OEMs (d) TSE research — and whether the work we had done made us incrementally more optimistic, or cautious, on this company’s role generating economic returns while advancing the energy transition.

A back-up tab then reviews all of our research to date, going back to 2019, and how we think that specific research conclusion might impact upon specific companies. This exercise is not entirely perfect, due to the large number of themes, criss-crossing a large number of companies, at a large number of different points in time. Hence the observations in this data-file should not be interpreted as investment recommendations.

The screen is updated monthly. At the latest update, in January-2023, it contains 324 differentiated views on 162 top public companies in energy transition.

Energy economics: an overview?

This data-file provides an overview of energy economics: 150 different economic models constructed by Thunder Said Energy, in order to help you put numbers in context. This helps to compare marginal costs, capex costs, opex costs and other key parameters of technologies and materials that matter in the energy transition.

Specifically, the model provides summary economic ratios from our different economic models across conventional power, renewables, conventional fuels, lower-carbon fuels, manufacturing processes, infrastructure, transportation and nature-based solutions.

For example, EBIT margins range from 3-70%, cash margins range from 4-85% and net margins range from 2-50%, hence you can use the data-file to ballpark what constitutes a “good” margin, sub-sector by sub-sector.

Likewise capital intensity ranges from $300-9,000kWe, $5-7,500/Tpa and $4-125M/kboed. So again, if you are trying to ballpark a cost estimate you can compare it with the estimated costs of other processes.

Renewables stand out. Despite high capital intensity (34% of revenues, 2x the average), once constructed, they also have the highest cash margins (76%, also 2x the average).

Low-carbon fuels and manufacturing/materials are similar. Both tend to have c20% average EBIT margins, after deducting 70-75% opex and c5-10% capex shares. This makes sense, as low-carbon fuels are effectively “manufactured” energy products.

The most exciting opportunities can also be picked out. They are clustered in the top-left of the chart, with high EBIT margins, low capital intensity and low costs once they are up-and-running.

Full data are available in the data-file below. To read the overview of energy economics send to our distribution list, please see our article here. All of the underlying economic models that feed into this data-file are available here.

Hydrogen: overview and conclusions?

The best opportunities for hydrogen in the energy transition will be to decarbonize gas at source via blue and turquoise hydrogen, displacing ‘black hydrogen’ that currently comes from coal, and to produce small-scale feedstock on site via electrolysis for select industries. Some see green hydrogen becoming widespread in the future energy system. We think there may be options elsewhere, to drive more decarbonization, with lower costs, lower losses and higher practicality.

(1) Green hydrogen economy? Our main question mark is over “economy”. Costs are modeled at $7/kg, equivalent to $70/mcf natural gas, after generating renewable electricity, electrolysing water into hydrogen and storing the hydrogen. Levelized costs of electricity then reach 60-80c/kWh, for generating clean electricity in a fuel cell power plant, yielding a CO2 abatement cost of $600-1,200/ton (note here). We think costs matter in the energy transition and the entire world can be decarbonized via other means, for an average cost of $40/ton in the TSE roadmap to net zero.

(2) Fuels derived from green hydrogen are by definition going to be more expensive than the hydrogen itself. We have evaluated electro-fuels, green methanol, sustainable aviation fuels, hydrogen trucks, again finding CO2 abatement costs above $1,000/ton. Again, we think transportation can be decarbonized cost-effectively via other means.

(3) How much can capex costs come down? There is an aspiration for electrolyser costs (presently around $1,000/kW on a full, installed basis) to deflate by over 75%. However, we have reviewed electrolyser costs line by line and wonder whether 15-25% deflation is more realistic (note here). Alkaline electrolysers vs PEMs are contrasted here. We have recently screened NEL’s patents to explore future cost deflation in electrolysers.

(4) Efficiency: the second law of thermodynamics. The absolute magic of renewables and electrification is their thermodynamics. These technologies can be 85-95% efficient end-to-end, precisely controlled, and ultra-powerful. A world-changing improvement on heat engines and an energy mega-trend for the 21st century. However, the thermodynamics of hydrogen depart from the trend, converting high-quality electricity back into a fuel. The maximum theoretical efficiency of water electrolysis is 83% (entropy increases). Real world electrolysers will be c65% efficient. End-to-end hydrogen value chains will be c30-50% efficient. We want to decarbonize the global energy system. It therefore seems strange to take 100MWH of usable, high-grade, low-carbon electricity, and convert it into 40MWH of hydrogen energy, when you could have displaced 100MWH of high-carbon electricity directly (e.g., from coal). And all the more so, amidst painful energy shortages.

(5) Backing up renewables? It is often argued that renewables will eventually become so abundant, especially during windy/sunny moments, that the inputs to hydrogen electrolysers will become free. We think this is a fantasy. Instead, industrial facilities and consumers will demand shift. Conversely, we are not even sure an electrolyser can run off of a volatile renewables input feed without incurring 5-10% pa degradation, or worse (if you read one TSE note on green hydrogen, we recommend this one).

(6) Operations, transport, logistics all feel strangely challenging. Our studies of patents suggest that electrolysers and fuel cells can be the Goldilocks of energy equipment. Past installations have declined at over 5% per year. Due to its small molecular size, 35-75% of hydrogen produced in today’s reformers can be lost. Some vehicles seek to store hydrogen fuel at 10,000 psi, which is 1.5x the pressure of hydraulic fracturing. Even in the space industry, rocket makers have been de-prioritizing hydrogen in favor of LNG (!) because of logistical issues. The costs of hydrogen transport will be 2-10x higher than comparable gas value chains, while up to 50% of the embedded energy may be lost in transportation: our overview into hydrogen transport is here, covering cryogenic trucks, hydrogen pipelines, pipeline blending, ammonia and toluene. Is a hydrogen truck really comparable with a diesel truck? (note here, models here). Finally, the gas industry is bending over backwards to stem methane leaks, due to methane’s GWP of 25x CO2, but hydrogen itself may have a GWP as high as 13x CO2.

(7) Will policy help? We are not sure. We are tempted to draw analogies to the Synthetic Fuels Corporation, bequeathed $88bn of US government money in 1980 amidst the oil shocks, which in today’s money is similar to the $325bn Inflation Reduction Act. It completely missed its targets of unleashing 2Mbpd of synfuels by 1992, due to challenging economics, thermodynamics, technical issues, logistical issues. What evidence can we find that green hydrogen will prove different to this historical case study?

(8) Niche applications can however be very interesting, where clean hydrogen is used as an industrial feedstock. An overview of today’s 110MTpa hydrogen market is here and underlying data are here. At large scale, we are currently most excited by using clean hydrogen in ammonia value chains and steel value chains, as the technology is fully mature and looking highly economical. It is also booming in the US. Elsewhere, an excellent large-scale application is to displace black hydrogen (made from coal), which is 20% of today’s hydrogen market and has a staggering CO2 intensity of 25 tons/ton. At smaller scale, there is also a weird and wonderful industrial landscape, using hydrogen to make products such as margarine or automotive glass. Putting an electrolyser on site beats shipping in hydrogen via cryogenic trucks. But these are also quite niche applications.

(9) Blue hydrogen is the most economical, low-carbon hydrogen concept we have found. Effectively this is decarbonizing natural gas at source, by reforming the methane molecule into H2 and CO2, the latter of which is sent directly for CCS. Our best overview of the topic is linked here. There are still c15% energy penalties. Costs are $1-1.5/kg in our models, to eliminate c90% of natural gas CO2.

(10) Turquoise hydrogen is also among the more interesting concepts, pyrolysing the methane molecule at 600-1,200◦C into H2 and carbon black. Our base case cost is $2/kg, with a $500/kg price for carbon black. But if you can realize $1,000/kg for the carbon black, you could give the hydrogen away for free. We have screened patents from Monolith and expect others to come to market with technologies and projects.

Around 40 reports and data-files into hydrogen have led us to these conclusions above; listed in chronological order on our hydrogen category page. The best way to access our PDF reports and data-files is through a subscription to TSE research.

LNG: top conclusions in the energy transition?

Thunder Said Energy is a research firm focused on economic opportunities that drive the energy transition. Our top ten conclusions into LNG are summarized below, looking across all of our research.

(1) LNG markets treble in our energy transition roadmap, rising from 400MTpa today to 1,100MTpa by 2050, for a c4% CAGR. The main reason is to displace coal, which is 2x more CO2 intensive. This LNG growth rate is 1.5x faster than total global natural gas supply growth, which “merely doubles” from 400bcfd to 800bcfd, for a 2.5% CAGR. The world needs $20bn of new liquefaction capex per year. Our LNG outlook through 2050 is modeled here.

(2) Marginal cost is $10/mcf as a rule-of-thumb for the 2020s. This is summing up the economics across the entire value chain for gas production, gas processing, pipeline transportation, LNG liquefaction, LNG shipping and LNG regasification. The best projects work at $7/mcf. But prices will run well above marginal cost amidst under-supply.

(3) Under-supply in 2023-28 in our supply model augurs for $15-40/mcf spot global LNG prices. After adding +20MTpa of new LNG supplies each year from 2015 to 2022, we think the world will be lucky to add +10MTpa in 2023 and 2024. There is always a further risk of supply disruptions. Meanwhile, Europe’s 15bcfd of Russian gas imports, volumetrically equivalent to 110MTpa of LNG, are shifting. The best note covering our gas outlook is linked here and our European gas models are linked here.

(4) The key challenge is CO2. Liquefying natural gas at -160C requires 300-400kWh/ton of energy, depending on the LNG plant design. This results in 3-4 kg/mcf of Scope 1+2 CO2. Across the value chain, LNG will have 7-10kg/mcf of Scope 1+2 CO2. Adding the Scope 3 from combustion, we reach total CO2 intensity of 60-65kg/mcf. Coal is 130kg/mcfe. Yet it feels like we could die of energy shortages before gas critics listen to “relative CO2” reasoning and countenance long-term LNG contracts.

(5) Rising to the challenge. The LNG industry can satisfy its skeptics. This is earnestly happening. It includes measuring CO2 in LNG supply chains. Then offsetting it via nature-based CO2 removals. Or capturing CO2 from combustion, then sharing regas terminal infrastructure to liquefy it, and ship it away for disposal. We have written a full note on back-carrying CO2 here. CO2 abatement costs range from $50-125/ton, or $3.0-7.5/mcfe. This scores well on our cost curves.

(6) 2020s supply growth will be dominated by the US, which is particularly well placed to assuage gas shortages in Europe. US LNG can treble from 70MTpa in 2021 to 200MTpa by 2030. It requires an extra 17bcfd of gas (c18% total US gas supply growth), which in turn pulls on E&P activity in the Haynesville, Permian and Marcellus.

(7) Longer term supply growth will be dominated by the Middle East, which is particularly well placed to phase out China’s coal. These numbers are mind-blowing. As an idea, if China directly substituted all 4GTpa of its coal (10GTpa of CO2 emissions!), this would require 1,600 MTpa of LNG, i.e., 4x more than today’s entire global LNG market. If you read one note, to understand this topic, we would recommend this one.

(8) Smaller-scale LNG and transport upside? We have reviewed opportunities in LNG in transport, smaller-scale LNG, LNG-fueled trucks, LNG-fueled ships, eliminating methane slip, LNG fuelling stations, small fixed LNG plants, floating LNG plants. There are some interesting concepts, especially for specific applications. But we have not materially de-risked smaller-scale LNG upside in our numbers yet.

(9) Cyclical industries reward counter-cyclical behaviours, and LNG is deeply cyclical. The title chart above shows this nicely, with spurts of growth, punctuated by plateaus, once per decade. It always feels uncomfortable to sanction projects when others are not. But our view is that bravery gets rewarded. “If you build it, the demand will come”.

(10) Companies. Incumbents benefit most from under-supply in the 2020s. Upcoming projects and their sponsors are summarized in our LNG supply model. We have also screened LNG shipping companies. But the question that fascinates us most is whether upcoming project sponsors can avoid the cost inflation that marred the past cycle, with some interesting evidence from patents in our note here.

Around 45 reports and data-files into LNG have led us to these conclusions above; listed in chronological order on our LNG category page. The best way to access our PDF reports and data-files is through a subscription to TSE research.

TSE Patent Assessments: a summary?

New technologies for the energy transition range across renewables, next-gen nuclear (fission and fusion), next-gen materials, EV charging, battery designs, CCS technologies, electronics, recycling, vehicles, hydrogen technologies and advanced bio-fuels. But which companies and technologies can we de-risk?

One way to appraise new technologies for the energy transition is to lock yourself in a room with a stack of patents from publicly available patent databases, read the patents, and then score them all on an apples-to-apples framework.

Our technology assessment framework is derived from 15-years experience evaluating energy technologies, from the best of the best world-changing technologies, to companies that ultimately turned out to have over-promised. The framework includes five areas:

(1) Specific problems. We find it easier to de-risk patents that pinpoint specific problems that have hampered others, and set about to solve these problems.

(2) Specific solutions. We find it easier to de-risk patents that pose specific solutions, whereas it is harder to de-risk technologies that are more vague.

(3) Intelligibility. We find it easier to de-risk patents that explain why their inventions work, often including empirical data and underlying scientific theory.

(4) Focused. We find it easier to de-risk patents that all point towards commercializing a common invention, and different aspects of that invention. Conversely, patenting 10 totally different solutions might suggest that a company has not yet honed in upon a final product.

(5) Manufacturing details. We find it easier to de-risk patents that explain how they plan to manufacture the inventions in question. Sometimes, very specific details can be given here. Otherwise, it may suggest the invention is still at the ‘science stage’.

The purpose of this data-file is to aggregate all of our patent assessments in a single reference file, so different companies’ scores can be compared and contrasted. The average score in our patent assessment framework is 3.5 out of 5.0, although there is wide variability in each category.

In each case, we have tabulated the scores we ascribed each company on our five different screening criteria, metrics on the companies’ size and technical readiness and a short descripton of our conclusion. You can also view all of our individual patent assessments chronologically.

Cookies?