AI helps fleet operators plan electric switch

How fleet operators are making the switch to EVs

Image credit: BT

Companies with large fleets of commercial vehicles can save money by making the transition to electric vehicles. Big operators like BT and DHL are using artificial intelligence to get the best from the change.

Switching large commercial fleets from internal-combustion engines (ICEs) to electric vehicles (EVs) could cut the total cost of ownership by between 15 and 25 per cent, according to an estimate by McKinsey and Company.

Big owners are already getting on board, also with an eye on net-zero targets. Amazon and Hertz have placed orders for 100,000 EVs in the US to replace ICEs in, respectively, their delivery and rental fleets. Amazon is also investing in renewable energy to partly power its fleet and wider activities as it transitions.

These players are adopting EVs at pace. Courier DHL aims to have more than 80,000 in its global fleet by 2030. In the UK, it had grown its fleet of electric Transit vans to 270 by March, having added the first 50 in 2021. They now service more than 30 towns and cities including London, Birmingham, Manchester, Glasgow, Leeds, Liverpool and Bristol.

“This next stage of roll-out is a positive step towards achieving net-zero emissions by 2050,” says Richard Crook, director of Fleet at DHL Express.

However, achieving an efficient ICE-to-EV transition is no simple task. There are a lot of infrastructure elements that need to be brought together. To do that companies are turning to artificial intelligence (AI) to best model their rollouts.

When considering those elements, bigger companies have an inherent advantage. They already track deliveries or usage or, in the case of BT, the UK’s largest telecoms company, maintenance. And that tracking has been going on for a good many years. The US took formal signal degradation off GPS in 2000, making civilian tracking much more accurate.

BT is an interesting example because it is also in the process of transitioning one of the UK’s largest fleets – 30,000 or so vehicles for a workforce of about 30,000 field operatives – and already has the data around some of the key questions that need answering because of earlier work it has done on optimising how it schedules appointments. It has also progressively used data analytics and more recent innovations in AI to make that original service more efficient.

This system, launched earlier this century, already allows BT to agree visits within two-hour rather than all-day windows. It has given the company a picture of, for example, what journeys are typically made by different technicians in different places during a working day. It can match that information to geographic data, weather forecasts as well as historic data on its network’s performance to help it carry out both repairs and predictive maintenance.

For its EV transition, this is now helping it make decisions around such factors as where to begin the changeover, where charging points could be put, and how regional variations could affect vehicle performance. On that last point, one reason for the 15-25 per cent range in the McKinsey study is that EV performance can vary according to factors such as geography and temperature – batteries drain faster for vehicles operating across hilly terrain and/or where it is colder.

Getting this right means having a lot of positional and topology data as well as what Paul O’Brien, director of BT’s Service, Security and Operations Lab, calls “a legacy of domain expertise”.

Having data that goes back a long way is also important. Wendy Keyes is a specialist in advanced spatial analytics at Esri, one of the world’s biggest players in analytics made possible by geographic information systems (GIS). She explains the main challenge here for a company looking to map retail activity, though the point applies more widely.

“A company I worked with recently wanted to forecast sales for the next 10 years, but they only had 10 years of past data to train the model. The problem was that the data was collected during a strong economic expansion, which limited how effective the model would be in forecasting sales during times of economic instability or recession,” she says.

Simulations can help get past this, but when faced with the extreme economic conditions looming today, the task is harder if you cannot dig into some of the older data.

There is a further challenge that Keyes has identified when users look to add AI. The resulting combination of GIS and AI, often called Geospatial AI (or GeoAI), is seen as useful for EV transitions as it can bring machine learning to bear on identifying usage patterns and exploit computer vision to further inform map-based analyses.

While both GIS and AI are technologies that have become increasingly digital, there are gaps between them. They have their own terminologies or some that overlap but mean different things. Mirroring the case for almost all types of new AI project, Keyes’ argues: “Business executives with GIS and data science teams in-house should get those teams together, because the magic tends to happen when they start collaborating.”

This is less of a problem for the likes of BT, DHL, Hertz and Amazon that have been promoting that kind of cross-learning across staff for some time. As other, smaller, fleet operators come under pressure to follow the same path, what are they to do? After all, road transport vehicles are estimated to account for a quarter of UK carbon emissions.

In some cases, McKinsey’s cost of ownership estimate may be convincing enough. For the big players, the kind of 2-3 per cent improvement seen from BT’s AI-fuelled efficiencies in forecasting have already become “millions of pounds of savings” according to O’Brien. However, even the vans run by smaller maintenance companies run up a lot of mileage every year.

For those seeking a more granular and holistic justification, the key may be to develop an appropriate geospatial strategy. Companies like Esri, along with competitors Mapbox and Carto, are on the hunt for business – and not just from EVs. GeoAI can be used to help to determine where companies put shops, clinics, warehousing and more.

The good news there is that what the bigger players have learnt and continue to learn is now filtering down. Demonstrating that you are a leader on the path to net-zero can offer a marketing advantage, but the wider recognition is that all players in all sectors need to play a part – EVs are just a very striking example.

Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.

Recent articles