Vandebron goes Hackathon: The Smart Charging Story
All you ever wanted to know about hackathons, electric vehicles and the endless opportunities of sunlight.
At sleepless nights, you might wonder about that warm, bright morning sunlight that’s ahead of you, and how it is used to charge your electric vehicle. Or even, how your vehicle is being charged while you’re asleep, or during our not so very sunny or windy days. We think about that too. Quite a lot. We would like to give you some insight in what goes on behind the scenes at Vandebron. Since it’s not all windmills, solar panels and farmers - it’s also (or maybe even more) a lot of technological development; data crunching, creating software, apps and other IT business.
average reading time: 10 minutes
Smart Charging, Vandebron style
A few weeks ago, a small delegation of Vandebron’s Digital team took their sustainable bikes and made their way over to the Smart Charging Challenge: a 24 hour hackathon on how to charge electric vehicles, the smart way. A hackathon (hack+marathon) is basically known as a codefest with the goal to create usable software. For all non-programmers, these IT events remain a mysterious phenomenon. Picture an enormous amount of computers, extremely focussed programmers, coffee mugs, energy drinks and an occasional sleeping bag. In this blog Giuseppe, our Head of Big Data, will share some secrets about the technical mystery called ‘hackathon’ and opens up about 24 hours in the life of a green-hearted software enthusiast.
Giuseppe: “Vandebron always strives for innovation in sustainability. Recently, the Electric Vehicles (EV) technology in the energy market is starting to show its disruptive potential, as also demonstrated by our Slim Laden pilot. As you can imagine, when we heard about the Smart Charging Challenge, we thought it was impossible for us not to be there! So we teamed up with a few colleagues (me, Sergio and Alessandro) and made our way over to the Smart Charging hackathon. In this blog we will share what we’ve learned, experienced, and of course, developed!”
Say what? A Smart Charging Challenge Hackathon?
“The Smart Charging Challenge is a two-day hackathon organized by the Living Lab Smart Charging, a collaborative platform composed of companies and public entities operating in the sector of electric vehicles and the charging poles infrastructure. For those of you not familiar with the 'hackathon' format, this typically consists of a competition between teams of software developers and others involved in software development, such as designers, project managers but also business developers and executives. Hackathons tend to have a specific focus. In this case, the focus was on smart charging: how to exploit and optimize the charging process of electric vehicles to create value for the customer or for society, with a focus on sustainability.”
Here comes the sun, little darling
“So, on a nice Friday morning, we took our sustainable bikes and headed towards the Living Lab Smart Charging (which is surprisingly close to Hemweg, go figure). Upon our arrival, we were greeted by lots of EV charge poles connected to your average bedside lamps. These lamps, we discovered later, were crucial for our challenges. After the opening ceremony, Stefaan Vandist gave us a keynote about the disruptive potential of electric vehicles and smart charging for our cities and the economy. We were presented with four ‘themes’ from which to choose from for the Smart Charging Challenge. We chose ‘Here Comes The Sun’. The objective of this challenge was trying to maximize the use of solar energy when recharging electric vehicles. A perfect challenge for Vandebron! And also, a great song.”
Teaming up, crunching data
“When subscribing to the challenge, our team got two more members: Robert van Gent, an EV driver and enthousiast that previously collaborated with us for the launch of our Elektrisch Rijden products, and John Iseghohi, a UI/UX designer from Booking.com. Together, we were given quite a lot of data, among which the most useful were:
an API where to retrieve the solar forecast by city, on a 15-minute basis;
the OCPI protocol specifications, to communicate with the charge pole;
the charging behavior of EV’s based on charging sessions for public, workplace and private use.
After lunch, which was offered by Instock, an amazing company that tries to reduce food waste by ‘rescuing’ excess food from Albert Heijn, it was finally time to code! We picked a spot, plugged in our lamp and our laptops and started hacking.”
Using the Sun while she’s out
“For our solution, we had to make sure that charging would happen during the peaks of solar production. As you can expect, a solar panel does not produce energy regularly during the day: its production depends on its exposure to, well, sunlight. Hence, solar production is higher when the Sun is high in the sky (at noon). During these periods, a lot of solar energy is available on the electricity grid. However, the energy demand is not so high during these hours: demand is typically higher in the morning (around 9am) and in the late afternoon (around 6-7pm). Since electrical energy cannot be stored efficiently, it would be much better to use this energy "on the spot", right when it's available. We had a few specific goals in mind:
We want to maximize the amount of solar energy used by the charging process
We want to make Smart Charging as seamless and interactive as possible for the user
We want to stimulate the user to smart charge in a sustainable way
This is when our Smart Charging app comes in!"
Data for days
“Having set the goals, we started looking into the data. The first consideration we made was that, in order to optimize for solar, it's very important to know where the user is located when a session is started.
The graph above shows the solar forecast for a typical weekday in the Netherlands (black line, left axis), overlaid with the proportion of charging sessions starting at a specific time. In red you see the sessions starting at home: these mostly start around 6-7pm, which is when a typical EV driver comes back home. These sessions yield the least potential for optimization, as the whole session happens by night and the driver expects to have his car charged in the morning to go to work. A better target are sessions that start at work or in public chargers (in blue and green respectively): these mostly start early in the morning, so the charging process can be deferred to when the solar production is at its peak.”
“Of course, this graph tells us nothing about the duration of a charging sessions: sessions starting at the workplace might last quite long, assuming a typical 9-to-5 schedule, but public sessions are more risky as we have no way to predict how long they will last.
So, our conclusion for this analysis was:
To maximize solar energy output, we need to focus on work/public sessions
Therefore we need to know/infer the location of the user: Work, Home and On the way
We do not know the length of the session in advance, and we don't want to ask the user either. So we made assumptions, knowing that in the future we might learn it from the data
We want to stimulate the user without disrupting the user experience. We decided to gamify our tool a bit. We thought of ‘Zonnetjes’, i.e. tokens given according to the amount of kWh charged during peak solar production. Zonnetjes can be exchanged in return for free ‘energy experiences’ (for example, a free trip to Rotterdam, or a Netflix movie).”
From API to app
“Having set our strategy, we started to work on our implementation. We wanted to have something we could easily demo, so we went for a mobile app. To incorporate the integration with the OCPI services (to send commands to the charging pole) and with the solar forecast API, we went for the following technologies:
Analytics: R. The R language allows to easily build complex statistical models, thanks to a large library of packages, that can run as independent services.”
“We then started working on how to get the data from the charging poles, using the OCPI protocol. This was when the lamps came into play: when sending commands to the pole, we could see the lamps turning on and off, according to the charging schedule we would send. The nice thing of the OCPI protocol is that it allows to use a ‘publisher-subscriber’ mechanism: once subscribed to a specific charging card/contract, the charging pole sends regular updates on the status of the charging session. This allows to detect events such as when a session starts or ends, or if a pole starts or stop charging.
The algorithm we used was quite simple; when a session starts:
Retrieve the solar forecast for the current day
Identify the intervals in the current charging session in which the solar production is highest
Charge the battery during these intervals
If necessary, keep charging until car is full
Of course, numerous improvements are possible, but this simple strategy served our purposes for the demo. In the end, it looked quite good!"
“After 24 hours of coding, it was finally time to show our efforts to the jury. Our team-mate Robert even brought a mini-Tesla that we could use to connect to a charging pole!"
“How our pitch went? Well, judge yourself! We had only 4 minutes, so I had to talk really fast!”
“Our project did not win the challenge (winning teams Co-charge TU/e and Emobility Scout had really cool products as well), but we still had a blast! We brought home a lot of positive energy and ideas for our e-Mobility projects within Vandebron, and for sure we will participate to more events like this.”