Electric Vehicle differentiation comes down to the battery, and the complexity surrounding it. Here’s how data science can help.

    Richard Layne


    Data Science Transportation Data Driven Engineering

    EV data can improve battery life – in design and in use. But disentangling the key engineering insights from noisy data can be hard.

    Benefits of EVs include cheaper fuel, smoother driving, and a sense of doing the right thing. Challenges include concerns about the practicalities and cost of ownership – including range anxiety, longer refuelling times and battery degradation.

    Mainstream EV adoption will come down to convincing buyers of the advantages, and making the challenges easy to overcome or accept.

    The higher the battery performs, the longer it lasts (between charges, and over its lifetime), and the more accurately the range to the next recharge can be predicted, the more appealing and cost-effective the vehicle.

    This comes down to a mix of battery design and use. By working together and sharing data, vehicle and battery manufacturers can deliver better experiences for their end customers. For both, untangling the complexities of that data is a challenge.

    What can we do with battery data?

    Collecting reliable in use battery and vehicle data can generate hugely valuable insight, not just for battery design, but for the entire infrastructure that goes around it. Benefits from data include:

    • Using vehicle data (driving styles, road conditions, routes) to predict demand on batteries over their lifetime for different use cases
    • Modelling optimal battery chemistry, thermodynamics, and engineering, for different lifetime needs or available materials
    • Designing intelligent charging approaches to optimise battery life, aligned to battery chemistry and driver behaviour
    • Creating digital services for drivers or fleet managers which empower sustainable driving and charging aligned to prolonging battery life
    • Modelling battery lifecycle to inform end of life management

    As we see, this is a complex system involving different players. Whilst each stage can be done individually, truly optimising batteries means bringing these all together to create benefits for battery makers, vehicle manufacturers, and end users, greater than the sum of their parts

    How is industry using battery data intelligently?

    To understand how we can use data to attack these multiple complex challenges, we can look at what is already happening.

    For instance, Tessella’s EV customers include a number of projects for automotive OEMs on their larger electric vehicles – such as trucks and buses.

    These vehicles are building up telematics data on things like distance, terrain, stop-start times, ambient temperature etc, which is supporting their fleet management as a service offer. Now, our clients are starting to think about how this data could be used to improve battery design and lifespan.

    By combining this data with measurements of batteries, such as the battery temperature profile, we are building a detailed picture of not just a good design, but an optimised design for different journeys.

    For example how does acceleration, or regenerative breaking, affect the current leaving or entering the battery. How does that change the temperature profile? How does that differ between light and heavy vehicles? How does it differ between vehicles making 100 deliveries per day, vs a lorry driving 12 hours on a motorway, vs a truck roving around a quarry?

    Short periods of high temperature can degrade batteries. But it’s not as simple as measuring the temperature, we need the temperature profile across the whole battery – alongside other important measurements – to understand what is really going on at a physical systems level, and what vehicle behaviour is driving it. Understanding that is important to understanding how batteries behave in different circumstances, and optimising them for different environments.

    Getting the data in the first place

    One of the big challenges is actually getting the right data. Many battery designs focus on the battery itself, and overlook the role of sensors for ongoing monitoring, so real-world data is limited. We may have a temperature measurement, but we could do much more with multiple temperature measurements across the battery’s subsystems. Even where data exists, data ownership and privacy issues can limit the ability to exploit it.

    This is part of the reason Tesla is so far ahead in consumer EVs – it recognised early on how valuable real world data would be for its R&D, and stuck in the sensors to get it. But others can catch up with a smart approach to data.

    Much can be still done with limited data. If you truly understand the physical systems underlying battery design, it’s still possible to build good models by inferring data points on things like temperature and degradation from the measurement and systems models that you have. But to really get transformational insights with complete confidence we need to identify what data points we need, and add in sensors to gather that.

    Combining battery and vehicle data to deliver better EVs

    By combining good battery measurements with good vehicle telematics data, we can model optimal battery design for different uses.

    Those battery models can then be used to optimise the driving and charging systems around it.

    We can combine it with driver data to design fleet management or driver intelligence systems to optimise battery life.

    This must be done by understanding the context of how drivers operate. Pure efficiency may be offset against other needs, such as when and where stops need to be made. This may in turn inform new battery designs.

    Longer term we can also use our knowledge of battery degradation and its effect on journey and charging times to make lifecycle decisions – eg once a long distance vehicle battery degrades to the point it can’t sustain a full journey, can it be switched into vehicles which make shorter journeys, and what are the implications of this?

    The challenges of battery data

    A major challenge will be collaboration. Where batteries are not developed in house, vehicle manufacturers have an interest in sharing their vehicle data with their battery manufacturers to allow them to design batteries for their specific use cases. But they will be wary of doing so if the manufacturer also sells to their competitors. Agreements, and the boundaries of in house R&D, will need to be carefully considered, but there are big rewards from getting it right.

    If these issues can be resolved, the next challenge is the complexity of the data science.

    This goes beyond looking at data on inputs and outputs and seeing how one affects the other. It involves looking at disparate data sets - physical and chemical battery data, the physics and economics of energy systems, models of human behaviour - and understanding how these all interact. These may be owned by different companies with different data management conventions.

    It involves disentangling variables and confounding factors. For example we may see a particular driving style increases degradation of a cell. We then need to know whether that is a direct effect of that driving style, or whether it is really driving a third factor, say temperature between cells, which has a knock on effect. Which one it is affects how you should address the problem. To reach the correct insight, you need to understand engineering systems, data, and models.

    How Tessella can help

    Dealing with this type of complexity is Tessella’s core expertise. We understand complex engineering challenges and the data they produce. We design data projects aligned to business needs, identify what data is needed, and perform modelling to deliver trusted insights. And we specialise in bringing together disparate teams of experts to identify where the data sits, and how to bring it together securely.

    A few projects we have delivered include:

    • Examining vehicle data, including where battery issues occurred, to provide a month’s warning of impending issues. Our client now has visibility across their fleet and can intervene before a battery fails.
    • Using temperature data to understand how different driving styles affect battery degradation to advise on targeted battery design for different vehicles and use cases.
    • Worked with company operating a large battery farm which they use to flatten energy demand, to capture data and analyse battery performance across the farm and support energy management decisions.

    Contact us to discuss how we can help you to radically improve the end-user experience of electric vehicles by using data to transform the design and lifetime management of batteries.