Big Data will fuel the EV revolution

Data will give us a true understanding of EV performance, smoothing the road to widespread commercial adoption

Judging from the collective excitement of journalists, analysts, and policy makers, it can start to feel inevitable that the 2020s will be the decade when electric vehicles (EVs) achieve ubiquity.

This year, the number of electric car models available to consumers is expected to jump from fewer than 100 to 175. In January, the British Government announced a doubling of available funding for on-street chargepoints, having already announced £400m for high-speed charging infrastructure the previous September. What’s more, the International Energy Agency forecasts that the number of electric vehicles on the road globally will grow from a little over 15 million today to more than 250 million in 2030.

For business leaders, however, hype is not enough. Alongside the 30+ million consumer passenger cars which slowly but surely will transition to electric with the spate of new battery powered options, Britain’s roads play host to millions of commercial vehicles.

From heavy goods vehicles connecting fulfilment centres, to last mile delivery vans bringing parcels to offices and homes, to taxis moving people around urban areas, commercial fleets play a significant role in greenhouse gas and air quality emissions – and are a huge opportunity for electrification.

Electrification for the nation

However, as is the case when moving to any new technology, the costs and risks involved mean that businesses demand certainty on a range of questions. Some, such as whether the necessary charging infrastructure will be available, are being addressed by policy initiatives. Others demand up front analysis on a case by case basis: economic questions such as calculating total cost of ownership and return on investment, operational questions such as verifying that real world performance will match advertised capabilities, or seemingly simple questions such as how long the battery will last before it needs to be replaced.

Happily, big data is here to help. With every EV component from chassis to battery linkages being designed and built with connectivity in mind, data can be aggregated from large numbers of vehicles to get a true picture of how they operate in the real world. Combining this with knowledge of a fleet’s requirements, we have the opportunity to give businesses insight and confidence about making the move to electric while ameliorating the high time and energy investment involved in capability assessment.

How batteries degrade

Consider that simple question: how long do batteries last? Battery degradation is a normal process for lithium-ion battery packs. Over time, the materials forming the battery cells gradually deteriorate, leading to a lower overall energy capacity. This, together with factors such as temperature, vehicle weight, and driving habits dictate how far a vehicle can travel on a single charge. If the battery capacity deteriorates substantially, the battery may need to be replaced.

While manufacturers are generally happy to discuss how far a vehicle will go on a single charge, it can be much tougher to get information about battery longevity. Instead, carmakers will typically offer warranty coverage on the order of 8 years/100,000 miles – but warranty promises do not help fleets to understand how range will decrease over the vehicle’s time in service, to anticipate downtime or to plan for lifecycles which extend beyond warranty periods.

Real world performance

Lab research shows that degradation is caused by factors including age, usage temperature, time spent at very high or low charge levels, high current levels to charge the battery, and total number of charge cycles. However, little research has been done on how these factors play out in the real world – until now.

In the interest of understanding more about how battery capacity degrades, at Geotab we recently conducted a study pooling data from 6,300 fleet and consumer electric vehicles using our platform to assess which factors impact degradation, and how.

Taken together, these vehicles provide over 1.8 million days of data which give us, for the first time, a deep, practical outlook on battery degradation. The headline news is good: across all of the models included in the study, batteries are maintaining sufficiently high levels of health that – if current degradation rates are maintained – means that the vast majority will outlast the vehicles they are installed in.

The study also confirmed the unsurprising fact that battery health declines with age, but with an average capacity decline of 2.3 per cent per year this is unlikely to impact day to day needs. Digging deeper, we find some interesting evidence for how different design and usage considerations impact degradation.

Use it

First, in response to the fact that extreme temperatures can affect performance, some models, such as the Tesla Model S, are designed with liquid cooling systems, while others, such as the Nissan Leaf, opt for cheaper air cooling.

We can quantify this effect by comparing these two models and see that, at the four year mark, the Tesla retains on average 92 per cent of its initial capacity, while the Nissan averages 86 per cent. Climate, of course, affects operating temperature, and we can also see this in the battery data. Looking at the weather data vehicles record, we can see that amongst vehicles with otherwise optimal conditions, those used in hot climates average 89 per cent capacity after four years, while those in temperate climates average 97 per cent.

Second, charging methods do make a difference to battery health. We can categorise charging into three types: 120V, such as a regular North American power outlet; 240V, such as a regular European power outlet; and direct-current fast chargers (DCFCs), which are specialised EV outlets rated higher than 240V.

Because data on charging current is available for analysis, we can see that while there is little difference between using 120V or 240V charging, DCFC causes significantly faster degradation. Amongst vehicles in hot climates, those that never use DCFC retain 89 per cent of their capacity after four years, while those that use it more than three times a month retain 82 per cent.

Third, and in a potentially surprising piece of good news for prospective fleet users of EVs, usage levels seem to have little impact on battery health. Amongst vehicles with otherwise similar usage patterns, those driven less than 8,000km per year retained around 92 per cent of their capacity after four years, compared to 89 per cent for vehicles driven over 20,000km per year – which is a difference within the margin of error for this sample size.

Getting the big picture from big data

If you love EV data like we do, these findings are interesting in themselves – but what do they really mean for users (or potential users) of commercial electric vehicles?

One key takeaway is that we can use it to provide broad, straightforward advice about how best to use EVs. While some duty cycles will demand DCFC usage, keeping it to a minimum – such as switching down to 240V when the vehicle is idle overnight – will help preserve battery health. You can’t control the weather, but you may be able to prioritise cooler days for heavy usage or construct shaded parking areas. And, importantly, heavy-duty usage will not significantly degrade the battery, so plan for high use when calculating return on investment.

The larger point, though, is that it shows how data can take the guesswork out of transforming commercial fleets to run on battery power. Taken together, this data gives us an overview of how batteries perform in different conditions; when connected with data about how a business’s current fleet operates, it can be used to outline precisely which vehicles will meet requirements, what level of charging infrastructure will be required, and even provide return on investment calculations.

Of course, this is just the start for telematics analysis. This study, after all, was performed using data from 6,300 vehicles out of a global stock of around 15 million. As the number of EVs in use trends towards 250 million, whole new ways of working with the data they generate will become possible.