Energy Singularity: A future closer than we would have imagined
Over the last few years, peer to peer energy trading has been slowly gathering more and more attention. Initially, it was only the research community which was fascinated by the potential of having end customers trading energy among themselves, but the commercial viability of such a service is slowly being realised as well. LO3Energy launched one of the first Proof of Concept (PoC) for Blockchain based Local Energy Markets (LEMs) in the form of Brooklyn Microgrid back in 2016 and sparked the interest from different fields. Since then a lot of PoCs and start-ups have popped up around the world wanting to showcase the cellular grid structure (where community self-sufficiency is prioritized) compared to standard uni-directional (from large power plants to end customers) system. One of the start-ups launched during the same time is GridSingularity. Since their establishment, this Berlin-based start-up has been working on creating an open-source platform that can provide the end customer high degrees of freedom in their energy-related interactions as well as the grid operator a clear view of the energy flows in their controlled network. After testing their approach using simulated household consumption and generation data and testing the limits of their software in the Chaos Theory Experiment, they launched their first Canary Network during Odyssey Momentum, one of Europe’s leading hackathon.
The story behind the ‘Canary Network’ goes back to the early mining days where canaries served as air toxicity measurement tool and thus reduced the chances of unexpected complications for the miners. If you are thinking of the famous phrase ‘canary in the gold mine’, you might be on the right track. The Canary Network is the testing process of connecting electricity meters (both digital and smart meters) from household and businesses to GridSingularity’s market platform i.e D3A and letting the autonomous intelligent agents trade electricity on the behalf of their owners. In a way, digital twins of the real energy assets were created. Connecting the end customers to D3A was carried out by OLI Systems using Smart Meter Gateway (SMGW) Interface and OLI boxes (embedded hardware from OLI Systems with multi layer functionality) using the knowledge gained in projects such as Allgäu Microgrid, DOSE and C/Sells.
During the hackathon, the ‘EnergySingularity’ challenge was divided into two streams. Stream 1 required the participants to create and operate the autonomous intelligent agents that traded on behalf of the end customers in the simulated energy communities. In Stream 2, the participants had to tackle the social aspects of the problem i.e. how to incentivize the end customers to get involved in a system which on one hand can seem complicated and adds to the things they need to care about, and on the other hand, provides them with an opportunity to fulfill the needs they are yet to realize. The results of Stream 2 have been discussed by GridSingularity in Energy Singularity Challenge 2020: Social Energy Networks.
Our team i.e. OLI Team (The combination of OLI Systems+Siemens Smart Infrastructure and InSignal) took part in Stream 1 where we developed and deployed our autonomous intelligent agent strategy that optimized itself based on the needs and requirements of every individual household that was part of the community.
Over the weekend of the Odyssey Momentum, there was a lot of interesting discussion based on the results obtained from different grid fee experiments. The discussions about the future energy systems along with the results of the hackathon challenges have been covered in detail by GridSingularity in Energy Singularity Challenge 2020: Testing Novel Grid Fee Models and Intelligent Peer-to-Peer Trading Strategies. But I would still like to talk about some of the key takeaways and questions that remain open for me.
The end customers need to be in the driving seat
It is a common conclusion drawn in the field of Smart Grids that the end customer of energy can not be sitting on the sidelines if we want to do a transition to cleaner and sustainable sources of energy. But the question always comes down to what kind of system design can be implemented to create welfare for all actors involved. Since the motivation of participation of every end customer can be a combination of intrinsic (being part of a community, helping in reducing carbon emission, etc.) and extrinsic (reduction in electricity bills, showcasing your good deeds, etc.) values, the regulations and market need to facilitate fulfillment of their individual motivations. In short, for LEM solution that is to be adopted by the masses, end customer should be able to see the fulfillment of their motivations by including their energy assets (solar panels, batteries, heat pumps, etc) in the local market. In the discussions, it was also realized how there was a huge need for transparency in asset decision making from the customer end.
Trust is an important key for the mass adoption of a solution. People need to feel comfortable with the knowledge or information they get out of the interface they use and also have the security taking over control if that is something they desire. The autonomous agent needs to be created as a complete representation of their believes and values while keeping it simple as well. In order to also comply with the data protection regulations (GDPR in case of EU countries), it would be mandatory to obtain clear consent from the end customers before an autonomous agent makes a decision on their behalf.
The questions that still remain unanswered are:
- Should the end customer be obliged to provide access to their energy asset (for say grid balancing) if the incentives provided do not match their expectations?
- How much information is enough for the end customer to feel comfortable with their energy asset interaction to the LEM? And what might be considered as an overkill of information?
Role of grid fees
In the status quo, wholesale energy trading (actual trading of energy in kWhs) is completely separated from congestion the power flow would be causing in the grid. The balancing of the congestion comes down to the grid operators and the cost is ultimately divided and passed to the end customers in form of grid fees. But when you imagine buying electricity from your neighbour, the application of the grid fee in the current form does not make sense as only a fraction of the public grid is used for the trade. Using IoT and blockchain-based solutions, generation and consumption can be tracked in real-time. This enables the possibility of a hierarchical grid fee system (as considered in D3A market approach) to be developed where the assigned grid fee corresponds to the grid usage and congestion created for the trade to take place. Information about the grid fee applied at every hierarchical level can help the end customers better understand the cost of trading electricity in an LEM and provide the trading preferences for the autonomous agent.
Additionally, the grid operators can be active participants in an LEM structure. By actively changing the grid fee at a level, the cost of congestion can be included and hence energy trading at that level can be influenced. This is one of the many market incentives that can be used to activate flexible assets such as batteries, electric vehicles (EVs), heat pumps, etc for demand-side management (DSM).
Some of the questions that still need to be investigated are:
- Should the grid fee be calculated and reported to the end customer before the trading starts or post-delivery?
- If beforehand, how long before should the grid fee schedule be communicated?
- Should there be a separate market for trading flexibility for last-minute grid balancing via DSM?
Regulations for LEMs
For a fully functioning LEM, a lot of different actors need to be involved to fulfill very specific roles. Some roles can be combined, while some need to be completely separated. Therefore, the requirements and objectives of every actor need to be understood and their responsibility clearly defined with regulations. The actors and the roles defined under the current research of LEMs are shown below.
Since a significant share of the target audience of the LEMs are households, customer protection needs to be at the forefront of the regulation. There have been issues such as abuse of market power in the wholesale energy markets, that need to be avoided in order to maintain the trust of end customers in the system. Therefore, detailed studies need to be done to understand the scope of such issues and create regulations for mitigation and to provide a guideline in case of conflicts. The regulation would also decide the margins available for trading in the local market and the business model of all the actors involved would also be dependent on it. Additionally, the current grid fee structure would disproportionally burden the non-LEM end customers to support the larger grid in case of mass adoption of LEMs. Therefore, the grid fee structure would also need to be evolved. Some changes in the regulations have already been seen in countries like Spain and Italy over the last months based on the Energy Community aspects mentioned in EU level directives and Clean Energy for All Europeans Package (2019). But in countries like Germany, the directives are yet to be transposed in national law and therefore there aren’t many good options available for small renewable assets that stop receiving the fixed feed-in tariff under the Renewable Energy Source Act from 2021.
Risk of trading
The introduction of LEMs have the potential to reduce the electricity bills of the end customers but also can cause them to increase based on the trading strategy used and the liquidity in the LEM. The variation in prices of electricity was also observed during the hackathon in a different grid fee experiment (as also shown in the community trading price graph below). So far it has been assumed that the end customer participating in the market understands these risks and is onboard. But for many end customers which are not satisfied with these financial risks but are excited about the community aspects, this might be a deal-breaker.
Therefore, for mass adoption of LEMs the end customer should have the option to avoid financial risks and still be an active trader in the local market. The service package provided by an aggregator (as defined in actors and their role above) should include the possibility to switch the preferences based on the financial risk the end customer is willing to take and focus more on trading locally generated green electricity.
Need for new performance indicators
As discussed earlier end customers participating in the LEM have very specific motivations. Similarly, other actors involved for functioning of an LEM can have motivations that are a combination of various factors such as reducing the cost of investment in grid infrastructure, promotion of clean sources of electricity, creating solidarity within local communities, implementation of new technology, offering exciting new products, etc. Quantifiable performance indicators are required for reflecting these complex motivations in order for an LEM to be working in benefit for all the actors.
After multiple rounds of lengthy discussions, our team came up with one such performance indicator for the end customer during the lunch break at a Mexican fast food place (would highly recommend to try out if you are in Stuttgart), the ‘Burrito Factor’. Burrito Factor (BF) considers the financial state of a end customer and reflects the performance of the strategy compared to the average performance of the community in a given time period. In case of a machine learning based algorithm such as Q-learning, increase in BF value can indicate drop in performance and hence a re-learning process can be triggered.
As part of the OLI team, we designed an autonomous agent for the hackathon with the idea of an end customer who understands their needs and desires well and has communicated them to the agent. Throughout the hackathon, we realized where our model was lacking and what other aspects we should be considered as well. We were able to implement some of the changes in the final optimization round and saw it bearing fruits already. Another thing worth mentioning is the openness among the teams to share their knowledge and experience while working on their agents. This is something I personally have not seen before among teams which are competing with each other.
My time in the hackathon has made me quite excited about the possibilities and the challenges we have in store for us to bring LEMs to the masses. Additionally at OLI Systems, we are working on projects such as BMIL and Flexchain and BEST with partners from utilities, grid operators, hardware manufacturers as well as research institutes to develop decentralized smart grid solutions for Germany. These solutions would hopefully help in answering some of the open questions I raised regarding the LEMs. The design of GridSingilarity ecosystem is transparent and comprehensible which should help in easy on-boarding of all the actors. With larger utilities such as E.ON and Engie being the challenge partners for the hackathon, it seems somewhat inevitable that LEM will move out of the PoC phase to the commercial product phase.
