OVO's plans for a zero-carbon energy grid require cutting-edge technology

Group CTO Ed Conolly explains the role of data infrastructure in meeting the demand for clean energy

Finally, electric vehicle sales are starting to take off. In the US the number of new battery-powered vehicles nearly doubled last year; in Europe orders rose 34 per cent, while globally 64 per cent more EVs were driven off the forecourt.

This is positive news for those suffering the effects of urban air pollution, but for EV sales to equate to a significant reduction of greenhouse gases, a much greater proportion of the electricity used to charge their batteries needs to come from renewable sources. More and more wind and solar sources are coming online, but storage has not kept up, meaning we are still reliant on polluting baseload generation (generally gas, oil or coal) to cover times when the sun is not shining and the wind not blowing.

Speaking at the Kafka Summit in London last week, Ed Conolly, group CTO at renewable energy firm OVO, laid out some of the engineering obstacles that governments and suppliers face in working towards a zero-carbon smart energy grid where that baseload supply is no longer required.

Cars as energy stores

After 2025, it's predicted that the vast majority of new vehicles will be at least partially battery powered, meaning that energy that used to be supplied by petrol or diesel will now come from the National Grid. This presents a supply-side challenge.

At present, most charging points provide 7.5 kW or 22 kW of power, the latter high-speed chargers drawing the same power as 12 boiling kettles. With one million new electric vehicles expected on UK roads in the next two-and-a-half years there will be a significant extra demand on the grid - requiring an additional 15 GW of capacity (or up to 15 extra power stations) over that time, said Conolly. And if all cars in the UK were battery powered and used high-speed chargers we'd need a generating capacity of 450 GW to handle the hypothetical peak, or six times the electric power the country produces today.

There are two main approaches to tackling this shortfall as we move towards a zero carbon grid. The first is to build many renewables-powered battery farms that store and feed the energy into the grid as its needed. This would work, but it's hard to make battery farms economic because of the price of the batteries and the huge difference in peak and off-peak transmission and distribution costs, Conolly explained.

The second approach is to encourage home power storage, installing large batteries like Tesla's Powerwall, but despite prices falling batteries are still expensive and for home owners it's tricky to make them better than cost-neutral. Again the economic incentive is lacking.

But there is a third way: use the EVs themselves as energy stores. "We spend tens of thousands of pounds on cars, but they do nothing 95 per cent of the time," Conolly said.

The pool of idle vehicles could be used as a storage lake to be drawn on by the grid at times of peak demand, eliminating the need for extra generating capacity, or to store excess power generated on a sunny day. Importantly, in this case the economic model is much more favourable.

"A car doesn't need to pay for itself, unlike a battery, so there's an economic incentive to go green," he added.

This plan will need "super smart bi-directional chargers" that understand real-time energy prices and power demand and which allow customers to configure them to optimise payback for their unused electrical power. Fortunately, such devices are now just starting to come onto the market, one produced by OVO.

Automated micro-management of power supply also requires infrastructure that can handle the data streams emanating from the millions of IoT devices needed to obtain a detailed picture of supply and demand.

OVO's data platform

For the past four years, OVO has been building an event processing system based on stream-processing software Apache Kafka, in partnership with the cloud supplier Aiven. Kafka is particularly well suited for such massive scale, real-time data management tasks said Conolly, and it has also helped OVO's 300-strong development team become significantly quicker at innovation and problem-solving.

"Kafka's incredibly well suited for huge amounts of writes because of its distributed partitioning nature. We can scale writes horizontally-ish, we can keep adding more nodes and partitions," he said.

It's also well suited for bringing in telemetry data and mixing other data with it such as weather, market data and customer settings, he added. "We can analyse all that data in real-time and make a decision on what to do with our network of charge points, vehicles and battery farms, when to charge, when to discharge and when do nothing."

Other benefits derive from Kafka's ability to compartmentalise workloads, and through its use to democratise data access. Selected teams can work on operational data as it's streamed in, before it's published to a data warehouse.

As a result of creating the platform, the 45 Agile teams at OVO are now more autonomous, and the hands-on nature of the architecture has already had a huge impact on team velocity -the speed at which requirements are dealt with by developers. Conolly estimated that the move to event streaming has already cut the time taken to deal with problems by a factor fifteen in some cases. This is important as time is of the essence.

However, he stressed that the move towards a zero-carbon energy grid is still at an early stage.

"A lot of this stuff is two, three or five years from reaching any scale, but from a climate perspective the problems are very real, so the changes we can make [in that time] are much needed," he concluded.