This paper from @iriepin.bsky.social and @nworbmot.bsky.social just blew my mind.
Google worked with them earlier, getting them to build an open source energy model to see what getting to 100% hourly fossil-free power by 2030 in diff countries might look like.
www.sciencedirect.com/science/arti...
We are offering a new free, hands-on, two-day workshop on PyPSA-Eur, this time in Copenhagen!
📅 Dates: June 26-27, 2025
📍 Location: DTU - Lyngby
👨🏫 Instructors: @fneum.bsky.social, @iriepin.bsky.social, @martavictoria.bsky.social Aleksander Grochowicz
Registration: forms.gle/4559UPat7NtM...
Thanks John. Excellent text!
I'd acknowledge also @nworbmot.bsky.social who made a pivotal contribution to the work on 24/7 CFE.
My latest @forbes article discusses an under-reported part of the current disruption of federal services: three edicts aimed at removing the foundations of climate-related financial risk management. www.forbes.com/sites/johnko...
24/7 clean procurement catalyses learning for long-duration energy storage and clean dispatchable power - new op-ed in @latitudemedia.bsky.social! 🔌💡
www.latitudemedia.com/news/how-24-...
in this context, an excellent slide deck by Prof. Hannah Daly
www.friendsoftheearth.ie/assets/files...
(i) 24/7 CFE buyers can spread funding across a tech portfolio, with iron-air as one option (here we were rather conservative in the paper) (ii) Iron-air procurement also happens outside the analysis zone, incl. buyers not focused particularly on 24/7 CFE. 2/2
Yes, you get an estimate by proportional scaling; but when communicating this calculation I'd use careful wording, since there are several hard-to-quantify factors affecting demand pull: 1/2
Shifting jobs between hyperscale data centres can help them chase renewables ☀️💨 around the globe, lowering the cost of hourly-matched clean energy commitments.
New commentary together with @iriepin.bsky.social, @gzachmann.bsky.social and Victor Zavala.
www.bruegel.org/first-glance...
5/5 Funny observation -> let's calculate average utilisation of DE data center fleet: 17.9 * 1e6 / (1955 * 8760) = 1.045. Either one of these numbers is off, or someone’s Bitcoin farm got spotted. Honestly, not sure why. 🤔
4/5 For reference, so far in 2025, German total load has varied between 40,009 MW and 76,582 MW.
🔗 www.energy-charts.info/charts/power...
3/5 On power: the same report says there are 1,994 data centers in DE with >50 kW load. Over 309 of these are colocation data centers with 1,300 MW load. The total load is near 2 GW (or exactly 1,955 MW -- as can be dug out from another GDA report).
2/5 On energy: with German C&I annual consumption is at 323 TWh (Eurostat energy balances), that is ~5.5%. If we take the commercial sector only with 117 TWh -> 15.3%.
1/5 Good point, let's take a look. The German Datacenter Association's 2024 report says the total energy consumption of data centers in DE was 17.9×1e9 kWh → 17.9 TWh
🔗 www.germandatacenters.com/fileadmin/do...
Nice illustration how systemically biased energy scenarios could be in envisioning technology success.
Taking average capacity factor for offwind at 0.36 and 0.11 for solar PV (rough values for DE), and ~1.8 GW of PPA capacity for each tech (from dena report), and German C&I demand at 323 TWh (Eurostat), and assuming all PPAs are signed by C&I (seemingly right, see p.6 dena) --> yields roughly 2.3%
The German Energy Agency (dena) reports that the combined capacity procured via PPAs was at 3.6 GW in 2023 (with nearly all energy signed from solar PV and offshore wind). There is no numbers related to energy, but let's make back-of-the-envelope calculation 1/2
www.dena.de/en/infocente...
💰 With tight fiscal budgets & uncertain governmental support, 24/7 CFE offers a demand-driven path to scale new technologies. By creating a demand for long-duration storage & clean firm power, we can help these technologies get off the ground—without waiting for perfect policies 7/7
When early adopters commit to 24/7 carbon-free energy, they create a self-reinforcing “virtuous circle”—driving innovation, attracting investment & making advanced energy tech widely available. A handful of companies & governments can set this in motion 🚀 6/7
3️⃣ As advanced technologies are deployed repeatedly, they become more competitive and are adopted widely in the bulk system, leading to emissions reductions far beyond the original voluntary commitments 5/7
2️⃣ Hourly CFE matching becomes economical by deploying advanced tech—clean firm power & LDES—alongside wind, solar & batteries. This creates an early market, driving learning & cost reductions, lowering the cost of 24/7 CFE, and making it more accessible for others 4/7
1️⃣ Companies aiming for 24/7 CFE directly reduce their emissions by matching a greater share of energy consumption with clean electricity on an hourly basis, which reduces the need for dispatchable fossil generation to firm variable renewable supply 3/7
Matching electricity demand with carbon-free supply every hour isn’t just about reducing your own emissions—it’s a catalyst for the energy transition. Here’s how it works: 🧵⬇️ 2/7
⏳⚡🌍 How do we scale advanced energy technologies before it’s too late? New Joule commentary by
@nworbmot.bsky.social @devonswezey.bsky.social @jessejenkins.bsky.social & me shows how companies doing 24/7 carbon-free energy matching accelerate energy transition for everyone
doi.org/10.1016/j.jo...
🚨 New Article and Policy Brief in Nature Energy 🚨
📄We examine the value of biomass in the European energy system and its interplay with other technologies under stringent emissions scenarios.
🔗 Article: doi.org/10.1038/s415...
🔗 Policy Brief: doi.org/10.1038/s415...
@natureenergy.bsky.social
A common criticism of energy system models is their reliance on single weather years, despite significant year-to-year variations.
But how can we ensure robust infrastructure planning in this case? What happens during rare events like the once-a-decade Dunkelflaute?
www.nature.com/articles/s41...
Our work discusses in detail the former part (additional percent of flexible loads reduces costs by avg. 1.29 ± 0.07 EUR/MWh); and we have no good data about the latter part. AFAIK hyperscalers have often operation headroom, and tend to have increasing share of flexible worksloads (like AI training)
🙌 This work was done with @nworbmot.bsky.social and Victor M. Zavala. Huge thanks my colleagues @fneum.bsky.social , Fabian Hoffman and Elisabeth Zeyen for always being there when I needed them most. And of course, a nod to the whole brilliant #PyPSA community. 8/8
📚 We’ve kept everything open-source — feel free to play with our open model! 🐙💻
🔗 The paper (in Advances in Applied Energy): www.sciencedirect.com/science/arti...
🔗 Explore the open model: github.com/Irieo/space-...
7/8
Illustration source: https://doi.org/10.1016/j.adapen.2024.100202
Signal 3️⃣: Follow the Sun — Earth’s rotation offers an opportunity for a clever solar power shuffle. 6/8