Machine learning and edge computing are rather new fields to me. Even more so when applied to the energy sector. In fact, most of what I know about distributed edge-to-cloud AI and the energy grid is encapsulated in my two blog posts around Davos Energy Week in January:
- Debriefing from Davos: Thank you for your questions! (January 26, 2021)
- Let’s meet at Davos Energy Week (January 19, 2021)
My involvement with RAIN’s team as a marketing strategist and storyteller began just a few months ago. That means that I am on a discovery expedition.
One thing I’ve learned is that energy markets and infrastructure are changing at their core, becoming ever more dynamic and data driven. Nuclear and fossil fuel plants are supplemented with solar, wind, and ‘fast reserve’ electricity stored in increasingly efficient batteries. At the same time, customers want to decide what type of electricity they purchase, when, and from whom.
New challenges to the grid
These changes pose completely new challenges to the entire energy value chain.
The frequency of the grid needs to be kept at close to 50 Hz at all times. With more dynamic demand and supply, some renewables depending heavily on weather conditions, and some sources causing lower inertia in the network than traditional power plants, stabilizing the frequency is a real concern.
What is needed is fast frequency response actuation, and this can only function properly if advanced distributed computing is deployed across assets throughout the infrastructure: on the edges, in the cloud, and on computing resources in-between.
The intelligence in the network has to provide real-time current-state as well as predictive data about frequency, demand and supply. Only with the help of artificial intelligence at the edge can the collected data input be sufficiently analyzed and then reduced before it is transmitted to operational data applications in the cloud.
Another challenge is that power traditionally flows through the grid in one direction only, and now ‘prosumers’ have started feeding electricity back into the network, ie. in the opposite direction. This requires hardware upgrades as well as accurate, real-time data processing.
Furthermore, the way electricity is being traded and how its price is being negotiated is ever more dynamic and closer to real-time.
These challenges can only be successfully met if the energy infrastructure and ecosystem become more data-aware end-to-end.
So far the big picture as I understand it. I beg you to correct me where I’m wrong and to complement whatever I’m missing!
Nordic and Baltic integration
The ‘European Green Deal’ is an EU climate policy, put forward by the European Commission in 2015, that aims for Europe to become the first climate neutral continent by 2050. The ‘Energy Union’ is the policy instrument intended to deliver on this transformation. Among other things, it envisages a fully integrated internal European energy market.
As it happens, the Nordic energy network and market is of particular interest due to its advanced technology roadmap, including the availability of 5G connectivity. This is where the mobile telecommunications revolution started (first with NMT, then GSM), and various aspects of that model appear to be emulated in the energy domain today. As one expert put it: “Energy today is where telecom was in the early 1990s.”
Within the European Network of Transmission System Operators for Electricity (ENTSO-E) there are 41 European TSOs (or grid operators) cooperating in 34 countries. The TSOs of Norway, Sweden, Finland and Denmark appear to work particularly closely together, having established common rules and regulations for operating the Nordic transmission system.
At the same time, the Baltic region has now been integrated with the Nordic electricity market as well, so that electricity retailers in the Nordic and Baltic countries can purchase electricity on a common Nordic-Baltic power exchange, Nord Pool. Electricity flows from areas with a lower price to areas with a higher price. The price and direction are determined by demand and supply.
AI and wind
Now, back to the smartness of those grids. What I’d like to learn next is which players in the Nordic energy value chain need that real-time distributed data sophistication most urgently.
Here is an image from a panel discussion at Davos Energy Week dubbed ‘How Machine Learning and AI Can Boost Energy Efficiency and Sustainability’, courtesy of Philippe Vié, Head of Energy, Utilities and Chemicals with Capgemini Group:
It depicts the maturity of certain AI implementations by smart grid operators. What it seems to indicate is that there should be significant demand for higher added-value AI solutions across the ecosystem from production to consumption.
Let’s take a look at wind power. At Davos, Giles Dickson, CEO of WindEurope, an industry organisation representing 400 companies and organisations across the value chain of wind, showed us that within twenty years, wind is projected to become the largest source of energy in Europe.
Currently there is significant political momentum to make this projection a reality. It would be great to see a similar graph zooming in on the Nordic countries…
Players and talking points
Who are the wind power producers, (sub)aggregators, Virtual Power Plant (VPP) operators and grid infrastructure companies that we can help by adding the necessary distributed data processing capabilities to their hardware?
I want to get to the bottom of this. 🙂
My intention is to talk with experts – on and off the record – about future and current needs around edge computing and energy. Starting with wind in the Nordics.
So: whom should I talk with? And what should I ask them?
What do you want to know?