E-mobility meets AI with Nuvve

Maria: Hello, everyone. Welcome to the new episode of AI Watch. I am Maria, senior data scientist at 2021.AI, and I’m here with Xavier Moreau. We are going to talk today about e-mobility.

Xavier: Hi, Maria.

Maria: So, would you like to introduce yourself a little bit so the people can know you a bit more?

Xavier: I’m part of Nuvve to help them in their business development and their strategy. And with Nuvve, I’m in charge of developing new business segments and also new technologies, new offers such as around predictive analytics and AI.

Maria: So we would like to know, first of all, if you can provide some overview about what Nuvve is doing in the e-mobility sector.

Xavier: Nuvve is leading the electrification of the planet, starting with transportation through our intelligent energy platform, we combine the most advanced V2G technology, V2G means vehicle to grid, to integrate electric vehicles with the power system. We combine it with batteries and with an ecosystem of electrification partners to accelerate the electrification of users. We do this by combining the charging of electric vehicles and the discharging of the batteries of electric vehicles with batteries that we are building, and with the grid. The benefits of this approach is that we can reduce the total cost of ownership of electric vehicles. We can reduce the total cost of ownership of charging infrastructure, and we can make it even more sustainable to drive electric and ensure that electrification is more affordable for everyone, to accelerate the trend to sustainable energy. We started with some pilots, and since 2016, so for over seven years, we’ve been operating commercially, starting here in Denmark, where we provide the regulation services to the energy operator.

Maria: Why is V2G technology so important in this sector?

Xavier: Today, grids are subject to high variability of renewable energy. Wind and solar energy are fluctuating all day long, and at the same time, you have electric vehicles which are parked 95% of the time. So what we do is we use a tiny portion of the electric vehicle’s battery to absorb excess energy from wind and solar and to backfill it into the grid when there is not enough green electricity available, for example. Of course, to predict what will happen, if there will be too much energy or not enough, if the vehicles will be there or not, is very difficult. And this is where AI can help to have a forecast of what will happen, so that we can plan ahead the charging and discharging to absorb the variability of renewable energy.

Maria: So that will also help the stability of the grid.

Xavier: Exactly. And by helping the stability, we make it more resilient and we avoid oversizing the grid. So it’s a benefit for everyone, even for those that don’t have an electric vehicle, because with the right size grid, the electricity tariff will be cheaper.

Maria: Sounds very interesting. As a data scientist, I have also been working with Nuvve and with the e-mobility sector for a few years. So I would like to ask you, where do you see the most significant value that AI can bring to e-mobility?

Xavier: At Nuvve, we use the batteries of the electric vehicles to provide flexibility to the power system. But to do this, we need to know when the vehicle will be parked and how long it will be parked and how much energy it will need to recharge before leaving again. And this is where AI is providing a lot of value, because thanks to AI, we are able to predict when the vehicles are going to come back to their parking lot, to the garage, to the depot, how long they will stay there before their next trip, and how much energy they will need to recharge. And with this forecast we can place the best bids on energy markets to monetize the flexibility of the batteries of the EVs. And with this information we can really improve the monetization of this capacity.

Maria: Could you share some examples or use cases where AI has brought real value to the company?

Xavier: Yes. For example, last year when we were placing our bids on the market, we used to do that with a trader that was just doing it based on his expectation of the prices on the market. Then, thanks to 2021.AI, we developed a model to better forecast what will be the price of the service we are going to perform on the market. We focused on frequency, regulation, and service. We are, thanks to AI, able to have a forecast of what will be the right price level to place our bids, to have a winning bid. Not only a winning bid, but a bid at the best price that will maximize the value of the capacity of the battery of the vehicle.

Maria: And talking about that, what are also the challenges that you have identified with AI?

Xavier: Our biggest challenge so far has been the structuring of the data, garbage in, garbage out. So we had to do a lot of work to structure our data to get the right level of quality to get a good forecast. In this, we were helped by you and your colleagues, data scientists, in structuring the data for ensuring the right level of quality that will enable us to have a forecast that is effective and usable.

Maria: That was a very interesting point, because we have been working with the data quality for a long time together with Nuvve, because that’s the key. That’s also the way I see it. The key of how good or bad the forecasts are is based on the quality of the data. So it’s not only about having data, but the data has to be sorted and well stored in a different bucket.

Xavier: Yeah, exactly. That’s why I use the word structuring, because the data is what it is. If it’s bad quality data, you have to take it as it is. But starting from what you have, by having this structured approach, we built a specific data warehouse to host this data. We are able to cleanse it or structure it in a way that is usable by the models.

Maria: For us, as a data scientist, having data, of course, is very important, but it is really important, like, even more than having a lot of data, to have that data aggregated in a correct format. So then we can process it, we can use it in a much better way. The models, everything goes faster. The training of the models, all the apps, everything is going to go faster if the data is well processed. So that’s a very important part of AI.

Maria: So now that we are talking about those challenges, I think it will be a good point to also introduce our viewers, like what is the collaboration that we have been doing together for the last, I think, three years already.

Xavier: So Nuvve and 2021.AI are in a joint venture that we call Astrea. And the objective of this joint venture is to bring all the expertise of Nuvve on e-mobility. Our electric vehicles charge and discharge our expertise on energy markets also, and to combine it with 2021.AI’s expertise in data science, in structuring data, in modeling, in AI. With our subject matter experts. Working with your data scientists, we are able to create specific models for the e-mobility sector.

Maria: Here at 2021.AI, we work with an MLOPs platform that is called GRACE. What are the most important tools or applications features that GRACE is bringing into this collaboration?

Xavier: The GRACE platform has been of great help to structure the data. First, we benefit from all the developments you’ve done that are mutualized with other customers. The GRACE platform brings a governance framework. It ensures data security, it ensures the data governance is appropriate. All these topics are not our core business, and you are the experts. So by adopting this platform approach, we benefit from all your expertise and the synergies you can have with other customers.

Xavier: Also, personally, as a business user, I appreciate the way that the data results and outputs are made available to business users, and how we can really play with the models without being a data scientist or an engineer.

Maria: At the end, if we don’t take both expertise all together, it wouldn’t work out. So that’s how it should be.

Maria: So, to wrap up, what are the key takeaways that you have taken from our collaboration?

Xavier: To use AI to integrate e-mobility with the grids, we could not have done it alone, and I don’t think 2021.AI alone could have done it. It’s only by combining our expertise of e-mobility and energy markets with the GRACE platform and your data scientists that we are able to deliver models and AI which is secure and effective. Our customers can trust they are not taking any risk by putting their data in our platform. And the results they get can be used for their business because they are relevant to the e-mobility sector and to the energy markets. So really, the key is that it’s only by combining the strength of the two companies into this joint venture, Astrea, that we are able to have this solution that is scalable and replicable across the five continents where we already have operations today.

Maria: Thank you Xavier, for coming today and for these conversations and all of you. Thank you for watching this video. If you would like to get notifications for all of our AI videos, please subscribe to our newsletter. Thank you.

Watch the previous episodes