PATtech.com’s Post

PATtech.com reposted this

View profile for Paolo Gabrielli, graphic

Energy management algorithms @ Huawei | PhD @ ETH Zurich | Optimization and machine learning for energy storage

Something is moving towards carbon-aware energy management and 24/7 carbon-free electricity (CFE) trading.🌿 The UK's National Energy System Operator is studying this with analysts AFRY, hourly matching enabler Granular Energy and market operator Nord Pool. Here is some insights from the analysis by Killian Daly (EnergyTag). 🌍 24/7 CFE has the potential to significantly benefit transparency, grid decarbonisation, and better investment signals (especially for flexibility). 🔭 The modeling approach is holistic and novel. It answers key questions such as costs, implications for balancing/dispatch, demand effects, and trading. 📜 24/7 CFE drives demand response, as hourly certificate prices will vary during the day: Low/zero when there is plenty of CFE and high when there isn't. 💷 24/7 might foster the deployment of batteries energy storage systems (#BESS), as hourly certificates could contribute to BESS revenues in the future. The link to the webinar is in the comments below ⬇️ #energymanagement #optimization #carbon #trading

24/7 carbon-free energy trading: Results and next steps

24/7 carbon-free energy trading: Results and next steps

afry.com

Paolo Gabrielli

Energy management algorithms @ Huawei | PhD @ ETH Zurich | Optimization and machine learning for energy storage

3w
Like
Reply
Stewart Dodd

Chief Executive Officer @ Empati | Economics Degree. Board member, Sentient Sports

2w

"Interesting exploration of 24/7 CFE trading, but there's a fundamental technical issue that needs addressing: half-hourly or hourly matching of generation and consumption is mathematically insufficient for accurate carbon accounting. Consider what actually happens on the grid: 4kWh consumed in the first 5 minutes of a half-hour cannot physically be matched with 10kWh generated in the last 10 minutes of that same half-hour, yet your block-based system would show them as matched. You need second-by-second matching of generation and consumption to reflect physical reality - electricity is consumed the instant it's generated on a grid. Anything less precise fundamentally misrepresents what's actually happening. This precision becomes even more critical when you consider that, as Jensen Huang says, 'AI doesn't care where it goes to school' - training workloads can shift globally in milliseconds to chase the cleanest, cheapest power, Would be interested to hear how you see your system evolving beyond temporal matching blocks to achieve the second-by-second precision needed to reflect actual grid physics, particularly given the real-time nature of global AI workload optimization."

See more comments

To view or add a comment, sign in

Explore topics