Renewable Energy Certificates: Oracles and Green Supply Chains
Prompt CC1. Many city, state, and national governments and companies are setting ambitious targets to have 100% of their energy needs provided by renewable energy sources. In addition, to help governments and companies reduce their Scope 3 carbon emissions and create demand for new renewable energy projects, there is a need to simplify the process of finding, procuring, and reporting certified renewable energy across suppliers in government/company supply chains and vehicle fleets.
We need to develop solutions that (1) enables suppliers to easily make and prove their purchases of certified renewable energy to the companies and governments that they supply and (2) automates the purchase of certified renewable energy based on EV charge volumes among electric buses, cars, bikes, and scooters to ensure that EVs are 100% powered by renewables rather than the existing electric grid mix.
Please contribute to help accelerate solutions addressing this unmet need in the clean energy sector!
Contributors have an opportunity to support the development of impactful solutions that EWF is developing with its global community of 100+ electric utilities, energy companies, grid operators, and startups that will help governments and companies leverage their supply chains and EV fleets to increase demand for new renewable energy projects. Contributors are welcomed to build solutions focused primarily on greening supply chains or EVs—or build solutions that support both. Contributors should also consider using EWF’s existing open-source software development toolkit for renewable energy and carbon market applications called Energy Web Origin.
Successful contributions may be integrated with real commercial dApps that run on the public, open-source Energy Web Chain by companies and regulators that are currently looking to provide solutions that support the greening of supply chains and EV fleets.
There are various opportunities to contribute impactful solutions to enable green supply chains and EV fleets, including but not limited to the following:
(1) Green supply chains: Imagine a large multinational corporate that has a 100% renewable energy target by 2025. This corporate has figured out how to buy renewable energy that matches the energy consumption of its own facilities. However, this corporate depends on many other companies across its supply chain to create its products and wants to encourage these suppliers to ensure that delivered supplies are also produced with 100% certified renewable energy. In the future, the corporate may even require its suppliers to guarantee this 100% green assurance to remain a supplier. There are several challenges and requirements that suppliers have that should inform any 100% green supply chain solutions, including the following:
- Suppliers that want to achieve and prove this 100% guarantee have specific privacy needs in that they want to prove they bought an equivalent amount of renewable energy as their energy consumption. They don’t want to share their energy consumption because this may reveal their internal cost and associated pricing structure (since energy is the top cost among many suppliers) and/or their expected quarterly profitability in relation to stock prices. Suppliers thus will require a solution that enables them to prove to this multinational corporate that they did in fact procure the necessary amount of renewable energy that matches their actual energy consumption without revealing the specific volume.
- Suppliers are usually very cost sensitive and so will want to avoid anything that adds time or new costs to their business. Any solution must therefore be very simple to use and ideally easy to integrate with any existing supplier reporting processes. Similarly, companies like the multinational corporate at-hand may already be reporting their environmental impacts to regulators and NGOs (namely, CDP) and these companies demand that any supply chain reporting solution integrates with CDP and similar reporting formats/processes.
- Suppliers likely supply goods for many companies (i.e., not just this one large multinational company) and those other companies might not require this 100% guarantee. If a supplier only wants to buy the volume of renewable energy equivalent to the relative amount of energy used for the goods delivered to the multinational corporate, there must be a way for the supplier determine this relative energy use volume and prove this is accurate to the corporate.
(2) Green EV fleets: Adoption of EVs is growing due to consumer demand, government policies, and more EV options being offered by automobile companies. EVs provide a source of carbon mitigation because they pollute less than traditional internal combustion engine vehicles. However, EVs are powered by the electric grid, which in most countries isn’t yet 100% powered by renewables. As individuals increasingly buy EVs for their households, governments deploy electric buses for citizens and EV fleets for government staff, and companies deploy EV fleets for their operations, there is an opportunity to develop streamlined solutions that automate the purchase of certified renewable energy based on the electricity consumed by EV charging. This will help increase demand for new renewable energy projects similar to how large companies are today staking their demand for new projects. There are several challenges and requirements that should inform any 100% green EV charge, including the following:
- This 100% EV green charge guarantee must protect EV owner privacy and not reveal publicly the specific amount of electricity that a given EV owner consumes.
- It is currently difficult to collect accurate data about EV charges, especially given how a significant portion of EV charging among individual EV owners happens at home.
- Payment systems currently vary across different EV charging station operators, which creates pain points when EV riders charge their EVs in different locations.
- It isn’t yet clear in the market as to whether EV charge point operators or EV manufacturers should be the entity that delivers this 100% green charge guarantee to EV owners.
- There may also be opportunities for a 100% green EV charge guarantee to support (and receive carbon credits through) emerging regulations for carbon emissions from transportation, such as California’s Low Carbon Fuel Standard (LCFS) since EVs help displace carbon emissions from transportation.
All solutions must also align with existing industry practices and regulations around certified renewable energy procurement. For example, any renewable energy purchase can only be certified, claimed, and reported with energy attribute certificates (EACs), where 1 EAC represents 1 megawatt-hour (MWh) of electricity produced by a registered renewable energy project. Thus, a company that consumes X MWh in 2019 must buy X amount of EACs to make the claim they are 100% powered by renewables. The main examples of EACs include guarantees of origin (GOs) in Europe, renewable energy certificates (RECs) in the United States, and international renewable energy certificates (I-RECs) in roughly 30 developing economies. These different EAC frameworks are relatively similar and differ mainly in terms of the local/national source of trusted data about actual renewable energy production in the country.
- EWF partnership with PTT to develop new direct renewable energy procurement platform in Thailand and wider ASEAN using EW Origin: https://www.energyweb.org/2019/09/11/ptt-and-energy-web-foundation-launch-blockchain-based-renewables-platform-for-thailand-asean-japan/
- EWF collaboration with PJM EIS in US to integrate blockchain into legacy renewable energy tracking system using EW Origin: https://energyweb.org/2018/10/25/energy-web-foundation-and-pjm-eis-announce-collaboration-to-build-and-evaluate-blockchain-based-tool-for-a-major-u-s-renewable-energy-certificates-market/
See Q&A Video walkthrough here
Renewable Energy Certificates (RECs) are attestations of 1MWh and meaningful for multiple reasons. Use data science and machine learning to build an open oracle that can cross-verify both the energy data used to certify renewable energy credits, and calculations for displaced carbon associated to it. For example, what would be an acceptable approach to turn a REC into a calculation of avoided carbon emissions? This would create a function where one could find the most impactful renewable energy projects in terms of avoided carbon emission and thus drive investment in new meaningful renewable energy projects.
With the advancement of technology, the number of energy generation sources is rapidly growing; from utility scale wind and solar farms, to commercial, industrial and residential households being able to generate their own electricity using solar PV and, increasingly, battery storage too.
Machine Learning (ML), the biggest subfield in AI, strives to recognize patterns, make predictions, and classify data. By leveraging ML models that are particularly suited to recognize outliers (such as support vector machines and k-nearest neighbors) we can detect anomalies in our datasets when an agent within a supply chain is trying to trick the system and use fossil fuels rather than renewable energy
Given such distributed energy resources (DER) centralized tracking of generation and clearance is no longer feasible at scale. At the same time, new technological developments in Blockchain, IoT, remote sensing (such as satellites) and Data Science create the opportunity to achieve consensus, tracking and transparency on the amount of electricity generated and emissions avoided by DER, which is the necessary foundation for trading energy and Renewable Energy Credits from DER, as well as many other applications such as carbon offset tracking.
Data collectors and transmitters such as sensors, humans, and oracles can be implicated in malicious acts and we need to use our ML models not just to detect anomalies in the data but also on the process of data collection and transmission. By monitoring data collectors, we can notice if someone is trying to tamper with our systems.
Background on Machine Learning in supply chains
Supply chains can be complex and intricate processes that span many countries and actors. How can we make sure that renewable energy was used in every step of the chain? How can we detect anomalous data or nefarious actors in this puzzling process?
Supervised learning is the process of teaching machines how to learn based on showing them both the input data and its label. This method of learning is by far the most successful in ML (and AI) and in order for us to leverage this technology in supply chains we must think on what kind of input data and labels we are looking for. Once we do this exercise, we should think if such datasets are readily available or if we would need to pay for them or create them ourselves. If the data we need is not available or the cost of getting it is too high, we should think of going back to the drawing board.
Since supply chains involve different actors in different locations, it is very likely that the data we receive will be in different formats. A significant amount of time in ML projects is spent working to prepare data for a model. From changing formats to resizing and normalizing, data preparation requires considerable time and resources that should be carefully estimated when planning an ML project. Try inviting at least one data scientist to your team to help you with this.
Once our dataset is ready, we need to look for patterns in it in order to choose an appropriate model. Is this data too complex as to require Deep Learning? Will a simple linear regression do the job? Do we even need ML or should we should we build a normal software program to help us with modelling and visualization? Don't look for nails just because you have a hammer!
Once you have a plan to train your ML model, consider how well would it scale, how much training time and computing power will it require, and if you can use renewable energy instead of dirty energy to train it. Once you get your accuracy score, consider tweaking its hyperparemeters to get better results. There are many more things to consider when building ML models but this brief list gives you a basic idea of the process.