Reference case

The problem: demand growth and complexity

Today’s distribution grids are serving an unprecedented demand increase, driven by electrification and data center load growth. In addition, demand patterns are becoming more complex and dynamic, and so increasingly difficult for grid operators to plan for and manage. Across the network, increasing volume and complexity of demand is outpacing current network capacity. Mismanaged demand risks network losses, reduced efficiency, damage to the network, and even damage or malfunction of consumer devices. This is driving network operators to start planning for — and deploying — billions of dollars of network infrastructure upgrades (poles, wires, transformers, substations) to keep pace.

One major driver of load growth today is the proliferation of customer-owned distributed energy resources (DERs) such as EVs, batteries, and heat pumps. Left unmanaged, without insight into how the timing or location of their usage impacts the network, these devices could strain local grids. However, many of these devices can be flexibly managed, shifting usage away from times when the network is near capacity.

The cost to serve an incremental unit of demand on the distribution network varies significantly by time and location, yet operators often continue to use fairly rigid pricing. Some European countries use time-of-use double tariffs (high and low), indicating higher grid price during the day and evening and lower price at night. In the UK, distribution grids use a rigid, three stage, time-of-use tariff (RAG: Red, Amber, Green).

The solution: dynamic distribution grid pricing

Kraken was recently involved in a UK-based trial to explore the consequences of pricing the network to more accurately reflect cost variability of serving demand. Distribution system operator Scottish and Southern Electricity Networks (SSEN), alongside supplier Octopus Energy (leveraging Kraken’s grid and flexibility capabilities) implemented a trial, which looked at grid congestion at a specific point in the distribution network (around Dundee, in Scotland), and reflected this in a day-ahead dynamic grid tariff to encourage the shifting of flexible loads.

How exactly dynamic grid pricing might be implemented varies, but this reference case holds valuable learnings for energy systems of various shapes and sizes, benefiting energy retailers and “distribution system operators” in deregulated markets such as those across Europe, as well as vertically integrated distribution utilities in the US.

How did the trial work?

• First, we used available grid topology and telemetry data to create a real-time visibility around four substations in Dundee, Scotland.

• We trained a forecasting model using Kraken’s grid capabilities, using grid visibility data to forecast power flow and congestion around those substations.

• We then converted forecasted loading into a price, using a loading-price curve. The specifics of this price curve can be defined by the grid operator. This trial used a curve with three parts:

  • A low price for loading below 80%;

  • A variable price, which increased in line with loading between 80-120%; and

  • A high price for loading above 120% corresponding to the supply interruption and value of lost load.

The loading-price curve

• To simulate a change in grid pricing, this variable price was added on top of the existing tariff structure to form the “dynamic distribution grid price.”

• The supplier, Octopus Energy, then put EV drivers around these substations on a bespoke EV charging tariff, which automated and optimized charging in line with the dynamic distribution grid price. Octopus leveraged Kraken’s platform to rapidly and robustly build the pricing mechanism, and design bespoke rates.

• We measured the impact of having EVs respond to the dynamic distribution grid price.

Network load around substations with dynamic price signals (top) vs control group (bottom)

Responding to dynamic prices: optimizing around customer preferences. Network loading across “Dundee 5” versus control group. The bottom graph in the image above shows the control group: vehicles would have charged at this time if not exposed to this dynamic price.

The graph above shows we were able to reduce load roughly in half, while only charging the cars that needed to be charged at this time to meet pre-specified customer preferences.

• Based on this limited trial, we calculated projected savings across the distribution grid (see CAPEX methodology below).

  • +10 years of CAPEX infrastructure deferral

  • +25% of peak load reduction

  • >100GWh per year of maximum energy shift

  • ~£1 Billion cost savings.

Implications for integrated utilities

In British or European-style deregulated systems, we might assume that the distribution grid operator would pass variable grid tariff pricing on to retailers, who would then choose how to pass these prices on to end-customers.

For vertically integrated systems, such as in the US, dynamic grid pricing could unlock similar benefits, though the process could be more direct. The utility, presiding over both the distribution network and customer interactions, would both determine the grid tariff, and how to expose network users. For example, they could choose to expose all customers to those rates directly, effectively putting all customers on time-of-use rates, or they might choose to expose customers with flexible DERs, and use a platform like Kraken to automate overnight EV charging, for example, in line with those rates.

This trial — and our subsequent projection — suggests that it is possible to reduce congestion by using dynamic grid pricing to encourage, or nudge, flexible consumers with DERs (shown here with EVs) to increase or decrease load. Critically, Kraken’s role proves that the necessary technology is available today to pass dynamic grid pricing on to customers. This can help utilities maximize current distribution network capacity, avoiding or deferring the need for network infrastructure investments.