Using telematics data to predict future electric charging networks

6 mins.
Have you ever heard of a Catch 22 dilemma? It is a situation where the solution to a problem requires another solution which depends on the positive outcome of the first. Sounds too abstract? Imagine you are a truck owner, and you are very eager to change to emission-free vehicles. The solution would be rather simple: just go to the truck dealer and buy an electric truck. However, to make your transport missions work with the newly purchased electric truck, you need to be able to charge it. And you need a publicly available charging infrastructure. If no such infrastructure exists, you would probably reconsider your initial idea of buying an electric truck, since what use would it be if you cannot charge it on the road? You get the dilemma?
Battery Electric Trucks
Using telematics data to predict future electric charging networks

Telematics data analysis
So, what can I and my team of data scientists do to overcome this dilemma? We can help decision makers make informed decisions using telematics data analysis. But it is not as easy as it sounds, and we still have to overcome some quite tricky challenges. A lot of great work has been done already. It started by determining the optimal battery size related to the truck’s daily routes when the electric trucks were still under development, continued by analyzing which markets or segments were the best candidates to start the electrification journey, and has now changed to other types of challenges such as establishing a robust network of charging stations. My role as a data scientist involves collecting, processing, and analyzing data to predict the future when it comes to the electrification of trucks. It requires a problem-solving mindset, and sometimes even some detective skills.


Starting point
My starting point is usually the telematics fleet management systems where trucks send data while driving. The vehicles are equipped with numerous different sensors that daily send a lot of data. Most important for me are their positions, the odometer reading and the time the signal was sent. With this very “basic” information, I can find out how many kilometers trucks drive per day (the difference of odometer min/max value per day per vehicle) and what is an average day (comparing all vehicles we have information on, and doing some statistics on their mileage, active hours, etc.) I am usually not interested in the behavior of a single truck but rather how the whole population of trucks behaves. The data we get from the trucks has a fairly high frequency, it can be sent every few minutes.


High precision data
This high precision data enables me to answer even more interesting questions, i.e. where they stop, how long do they stop, how many vehicles stop at a certain location. This allows me to find patterns of how the days can look, short days with many stops (e.g. garbage collection trucks, regional deliveries) vs long high-kilometer days with very few but long breaks (e.g. demanding long haul) just to name a few. Since most trucks on the roads still have diesel engines, most of my analysis is based on data from conventional diesel trucks, but it still enables me to make predictions for the future.

By using telematics data analysis, I try to determine which trucks could be electrified, and for how many is it possible to just exchange the diesel engine with an electric motor and still fulfill their daily routes? How many more would make it through their day if they could have a lunch charging opportunity? And most interestingly, for the electric ones out on the roads: how do they charge? Do they tend to charge fully during the night, how often do they recharge during the day, and at what locations?


Charging event
For this I have to define a ‘charging event’: an event where the truck (i) stands still (which I can find out from the GPS coordinates), (ii) has a low power consumption (which I know from the energy consumption), and (iii) changes its battery level (increase in battery level).

To give some flexibility, e.g. for re-parking or re-adjusting the position to be closer to the charger, some small movements are allowed during the stop. Moreover, we only want to consider an event as a charging event if it registers a minimum number of changes in battery levels.
Whenever all three criteria are fulfilled, we characterize the event as a charging event with a certain location, start time, stop time, incoming and outgoing battery level and the amount of energy charged. The charging events can then be collected, clustered into charging sites based on their location, and visualized on a map. In the analysis process, visualizing the data is key, particularly through maps showing charging stations locations, both existing and potential ones. In the telematics fleet management systems I can view a map of charging stations, alongside data tables and figures, and it provides a holistic perspective.

Charging locations
The results of current charging locations are a bit disappointing, though — a few hundred charging sites in Europe suitable for trucks. And that is — unfortunately — not the result of an incorrect analysis but the status of the charging grid. If we aim to see more electric trucks on European roads, we must predict where future charging stations will be required. A good starting point is to look at the data of where trucks commonly stop (for their lunch breaks, to sleep, to load/unload, etc.). Me and the team can delve into further details, such as the number of daily stops made by the truck, the duration of its journeys before and after a stop, and so on. It will be much easier for energy providers (such as gas stations and electricity grid companies) to add electric chargers on an already existing parking lot than to start building from scratch. Based on the current stop behavior, we can predict promising charging locations for the future. The telematics data analysis of existing data plays a fundamental role if we want to have more electric trucks on the roads in the future, and thereby reducing the climate impact, noise and emissions. It is important to ensure an adequate number of charging stations to cover the trucks’ daily missions and have as good knowledge as possible of their placement so that we can help inform decision makers, such as politicians or energy grid providers, where to focus.


A few other challenges are left to be solved. The first challenge pertains to the privacy of the truck drivers. It is important to find a balance between the insights we can get from the data and the preservation of privacy of the drivers and treat the data with respect. For this purpose, we need to seek permission for privacy responsibilities reasons, which can be time consuming, but it is vital to ensure that we handle the data in the right way. The second challenge is that many sensors in the truck, while working perfectly for the tasks they were intended for, were not developed for large scale telematics data analysis. This might result in unreliable output, which occasionally requires extra effort to ensure accurate results since high-quality data is essential for precise analyses. Yet another challenge lies in the complexity of managing the large amounts of data we collect. Several teams within Volvo Group Connected Solutions, the organization I work for, are involved in collecting and storing the data, and hence getting access is not always as easy as it could be. While it is no problem in some projects to wait a bit for data, in others we would like to have data, in the best case, yesterday. In the fast-changing world of electric vehicles, it is important to be on top of the latest developments.


Our long-term objective is to help the customers to make it easy to own an electric vehicle. The customers data is crucial and helps both us in the development of new services towards a greener future, and the customers to make the transition towards an electric fleet as smooth as possible. Innovative solutions are needed to simplify the path for companies to invest in electric and other zero-emission trucks. I believe this is the best way forward. It is time to invest in sustainable solutions to meet our energy requirements.

Author Bio:
Maja Feierabend
Data Scientist from Berlin; at Volvo Group Connected Solutions since 2021. Ended up in Gothenburg, Sweden, thanks to a professor that offered an exciting mission at Chalmers. Enjoys solving problems at work, in my spare time it is all about renovating the house.


Do you want to connect with Maja? Check out her LinkedIn page here.

Join the community and receive regular updates and invitations to #tecHER events.