COP28 kicked off in Dubai this week. For nearly three decades, the UN has been bringing together almost every country on earth for global climate summits – called COPs (which stands for ‘Conference of the Parties’ in case you, like me, had been befuddled by the acronym).
This year the UAE is hosting the 28th. More than 70,000 delegates are expected to attend. Part UN led conference, part solutions showcase and media event, COP28 is split into two parts:
The blue zone
All of the official sessions, meetings, side events and press conferences are taking place in the “blue zone”, the formal conference and negotiation space managed by UN Climate Change.
The green zone
The “green zone”, managed by COP28’s host country of UAE, is a space for youth representatives, artists, businesses, regional and local decision-makers to discuss, present and exchange ideas and solutions for a net-zero future in a more informal setting.
About COP 28, United Nations Climate Change website, Nov 2023
While members of the public can’t just rock up to attend official events, many of the blue zone events including plenary sessions and press conferences will be webcast live and there is a COP28 UAE Youtube channel.
Energy, how we use it and critically how we generate it, is pretty much THE critical topic at COP28.
A cornerstone of COP28 is the conclusion of the global stocktake, a review underway of the world’s efforts to address climate change. It is designed to pinpoint deficiencies and help countries recalibrate their climate strategies.
The global stocktake unambiguously states that, to meet the Paris targets, countries must collectively be more ambitious in cutting greenhouse gas emissions. That includes rapidly reducing carbon emissions from all economic sectors.
It means accelerating adoption of renewable energy such as solar and wind power, … and deploying clean technologies such as heat pumps and electric vehicles on a wide scale.
Fossil fuels currently make up 80% of the world’s total energy consumption. Their use in 2022 resulted in an all-time high of 36.8 gigatons of CO2 from both energy combustion and industrial activities.
UN’s ‘global stocktake’ on climate offers a sober emissions reckoning, The Converation, Nov 2023
Increasing grid complexity
Accelerating the use of electricity generated from renewable sources - the movement of water, the movement of air or directly from the sun’s photons - is central to efforts to curb emissions from generation. It also presents a deeply complex challenge for existing electricity grids - which seem to me to be the energy sector’s equivalent of the mother of all legacy codebases.
When our electrical power grids were built, they were powered by a small number of large power plants, centralised coal-fired supplemented by natural gas and hydro, that created reliable amounts of power on a predictable schedule (more or less). And the energy flowed one way - from centralised producers to end users.
Now we additionally need to work with large renewable energy sources with variable capacity (solar, wind, wave) and an enormous number of small-scale energy resources, (commonly known as DERs or distributed energy resources) e.g home solar panels, home batteries, electric vehicles. And two way energy flow that fluctuates in magnitude and direction throughout the day.
All of this means that what are arguably some of the most complex systems ever built by humans are rapidly becoming even more complex.
Increasing demand, climate change imperatives, increasing grid complexity. It felt like the perfect time to dive into ways in which data and AI, so pervasive in the SaaS tech community, is contributing.
Instrumentation included
When you build consumer SaaS products, you know (or you darn well should!) that you need to instrument your product in order to actually understand it. Where do your users struggle? Where do they drop out of your funnel? What’s engaging? What’s glittery but rapidly abandoned?
With the advent of smart meters, the granularity of ‘energy usage instrumentation’ took a dramatic leap upwards. Smart meters operate within the Advanced Metering Infrastructure (AMI), recording and transmitting your electricity usage on a regular, intra-day basis. In Australia, this means roughly every 30 minutes.
While the above consumer centric picture emphasises the ‘smart consumption’ angle, the granular instrumentation of site by site energy consumption also has implications for grid security, reliability and maintenance.
Smart consumption and behaviour change
I’m a little bit of a quantified self nut so you can imagine that I’m also interested in quantified house. I’m not quite going around and switching off the toaster at the wall when not in use (yes, my father did that, probably still does) but I’m definitely wooed by power company offerings that give you visibility into daily usage.
Currently, I’m a customer of Amber which offers consumers access to wholesale rates (the much discussed ‘dynamic pricing’) AND hour by hour price forecasts
By making the invisible visible, behavioural nudges like these along with price incentives help push me and others to move large variable loads like dishwashers, washing machines and roast meals to times when solar and wind power is abundant. This study from the UK suggests that the visibility of usage that smart meters provide is enough to nudge consumers to reduce electricity consumption by an average of 3.4%.
One step beyond behavioural nudges is Intellihub, a utility services company that provides home sized virtual power plants (VPPs) by taking control, via smart meter, of the pool pump, solar panel, home battery, hot water systems and electric vehicle of a typical affluent Sydney home.
“When you start combining hot water, pools, solar and battery coordination and EVs – suddenly you actually have the hallmarks and the bones of a genuine VPP. All of those [can be] combined in a co-ordinated manner through our smart meter to make a really material impact on the home’s efficiency.”
Wes Ballantine, CEO of Intellihub, AFR June 2023
Amber and Intellihub are both examples of demand response, or shaping the load to shift demand around inside a 24 hour cycle. In the US, Lunar Gridshare is a similar idea, aggregating DERs for homes and for fleets.
Lunar Energy have already used Gridshare to help UPS electrify its fleet of delivery vehicles and to assist ITOCHU to optimize the charging and discharging of home batteries as well as providing disaster resilience by tracking severe weather alerts. They can trigger devices to “automatically charge to and hold at 100% in anticipation of any potential outage. The battery is kept at full charge as long as the weather alerts continue, and then the software releases the battery back to its AI-driven optimization mode after the weather threat has ended.”
Edge computing comes to the energy sector
In consumer SaaS, once you’ve tasted the power of granular usage data you crave more. The same appears to be true for the energy industry.
NVIDIA has partnered with Utilidata to develop a smart grid chip, based on the low power NVIDIA Jetson chip, that is designed as a plug-in augmentation to industry-standard meters.
With the increase in distributed energy resources – especially batteries, solar, and electric vehicles – those technologies can pose a management problem or opportunity to utilities. We’re working with NVIDIA to move that central computation out to the edge of the grid, so we can manage those resources in real time and integrate them into real-time power flow dynamics from the substation to the end of the line.
Rather than measure and transmit data a few times an hour, as is typical with the first generation of smart meters, Utilitdata, Itron and Schneider Electric among others are aiming to add capabilities that measure multiple times a second.
DI-enabled smart panels help utilities manage load throughout the distribution network, including down to specific circuits within a home in strict coordination with distributed energy resources behind the meter.
Itron expands Distributed Intelligence Platform, Itron press release, Jan 2023
This much lower data latency has implications for frequency control and power factor, both important for supply stability and a growing area of focus for energy utilities as home batteries and solar panel installations surge.
As COP28 and the rapidly deepening climate crisis push us towards ever higher percentages of renewable energy sources, leveling up the instrumentation of the energy sector and the ability of all the players, in a industry traditionally conservative when it comes to new software, seems essential. Great news for those with skills in data and an desire to contribute to something more compelling than selling ads 😊
Nowcasting supply
Being able to predict how much sun or wind will be available two hours from now allows utilities to determine how much generated energy they should store and how to tweak generation assets to maximum effect.
Early in November, Google announced MetNet-3 providing high resolution predictions up to 24 hours ahead for a core set of weather variables including temperature, precipitation, wind speed and direction, all relevant to energy generation via wind turbines (which can be impacted by icing in poor conditions).
ECMWF for those not steeped in weather forecasting lingo, stands for the European Center for Medium-Range Weather Forecasts and is one of the existing gold standards.
According to Google, MetNet-3 is used in various Google products and technologies that process weather information and has the potential to "improve the safety and efficiency of many activities, such as transportation, agriculture, and energy production." Super excited to see where this goes and when coverage of MetNet-3 extends outside of North America and Europe.
From wind to sunshine, Solcast, an Australian startup snapped up by Norwegian energy risk and assurance giant DNV earlier in the year, provides real time and forecast irradiance and photovoltaic (PV) power data
Solcast APIs provide “data spanning from 7 days ago to +14 days ahead, at 5 to 60 minute granularity, anywhere on Earth. Includes probabilistic forecast estimates of PV power and all types of irradiance. Actuals available in real-time.”
James Luffman, co-founder and CEO of Solcast, describes this, rather poetically I thought, as the ‘solar pipeline’. Knowledge of short term irradiance helps solar asset owners and grid operators predict how much power they will be able to generate in the next few hours.
By anticipating fluctuations in irradiance, operators can proactively manage energy storage systems and adjust their operations accordingly. Efficient utilisation of storage assets, supported by accurate forecasts, allow solar producers to meet the energy needs of the power grid, helping reduce the need for costly backup power sources, such as fossil fuel generators, which can increase both financial and environmental costs.
James Luffman, co-founder and CEO, Solcast blog, May 2023
Till next time
Having wandered into the energy space on my curiosity-filled semi random walk through real world impacts of data and AI, I find it fascinating! So expect more to come on this topic. Must be my inner physicist stretching her muscles. Enjoy what I hope is a peaceful weekend. I’ll be dodging raindrops again here in not-at-all-sunny Melbourne where summer is stubbornly refusing to arrive.