After After Diversity

Responding to a recent article in Utility Week, Adam Bell, the government's former head of energy strategy, discusses market-wide half-hourly and its impact on ‘demand diversity’ – one of the key assumptions around which local electricity networks are built. Bell says this is no reason to delay the move but does mean the industry needs to get serious about “hyper-local” markets for network capacity.

There are a host of voices pushing back against Ofgem’s recent announcement of a delay to market-wide half-hourly settlement, including the CEO of the Energy Systems Catapult amongst others. But one group of voices very specifically not pushing back are those quoted in Utility Week’s recent article pushing back against the end of ‘demand diversity’. This refers to the concept of ‘After Diversity Maximum Demand’ (ADMD).

I want to unpack what in layman’s terms what ADMD is, why it’s important, and why it both is and is not a stumbling block for market-wide half hourly settlement (MWHHS). Strap in for some really gourmet electric geekery.

ADMD is a statistical analysis of consumer electrical demand used by distribution networks to right-size connections and transformers in the low voltage network. Each distribution network has its own approach to this that it uses for, say, designing networks for new-build estates. You can see an example from SP Energy Networks here.

What it’s statistically analysing is how likely it is that all of us are going to use our devices at the same time. The peak demand from my house might be when I switch my kettle and my electric oven on at the same time; let’s call it 1.5kw. The theoretical peak demand from two houses like mine would be both houses running their kettles and ovens at the same time, so 3kw. You can immediately see that it’s unlikely – but not impossible – that I am boiling my kettle in the same few minutes as my neighbour.

What is the most likely peak demand for both houses? Clearly, it’s some fraction of 3kw. We could find out a reasonable number for this just by watching the behaviour of both households over a period of time. What happens if we add a third house? Manifestly, it’s even less likely all three households will be boiling their kettles at the same time, meaning that the most likely peak demand for three households is lower per house than it is for just two. This is After Diversity Maximum Demand.

It means that, for a given electrical network covering X homes and commercial buildings, you don’t need to build enough cables and transformers to ensure that the network is capable of handling everyone boiling their kettle at once, but typically a significantly lower fraction of that.

In the framework given above, for example, the ADMD for a five-bed property is 2kw, or a reasonably sized electric kettle. This may seem very low, but this approach has been proven to work over decades and is a reliable tool for network design.

Why are the network companies worried that market-wide half hourly settlement will interrupt this? They’re less worried about the switch and more what it will enable, specifically time of use tariffs and anything that involves controlling devices in line with market signals. The reason for this is that households with a heat pump and an EV and a smart tariff set at a national or even nodal wholesale price eliminate variability in domestic demand.

Imagine a big uptick in wind output across the country, causing prices to fall. A smart tariff tells the heat pump and EV in a household to switch on, charging a heat store, a hot water tank or a battery. But because all households are run remotely all of these devices switch on at once. This is great from a national perspective, as all that cheap wind is being used. But at the low voltage network level, there’s no ADMD so the low peak capacity almost all our networks are designed around is burst through.

This won’t actually cause anything to explode, because – I assume – we’ll have obliged network companies to massively upgrade the grid. But what’s at stake is billions of pounds of additional investment that we’ll need if we don’t find a way of efficiently managing local voltage networks.

This is about who gets to run their assets and when, and what size of network accommodates a reasonable level of peak demand against its cost. You can either do this through a local operator dispatching assets individually against a centrally determined peak demand, or through developing a new price mechanism to efficiently manage hyperlocal constraints.

The latter is obviously preferable, as the former involves some consumers losing out through no fault of their own. But either route involves knowing what’s happening on very low voltage networks and being able to send signals appropriately. We do not presently have the data for that; most transformers remain analogue. You would need a way of knowing what’s happening in each household connected to an 11kv transformer and the load on that transformer.

Enabling market effects to operate over a period would give us an understanding of how households assets, optimised algorithmically, operate in response to different price signals. This would give a figure for a ‘real’ Non Diverse Demand peak load.

The alternative is monitoring demand from fully electric households with automatic asset dispatch over an extended period to discover whether any ADMD effects remain. If different people have different suppliers with different trading arrangements, this might be the case.

Either route would give DNOs a proxy for network planning, and both require HHS to actually happen to begin to collect the data, or looking at something like the Catapult’s Living Lab work.

The loss of ADMD is not a reason to delay HHS, but it is a reason to get serious about novel hyper-local markets for network capacity and building a data spine to enable them. This could take the form of extending nodal pricing down well below the GSP to the 11kv transformer level or developing specific network charges for each transformer. This, compared to the present market, feels a long way off.  But without a new way of right-sizing network design, we are likely to end up over-building our energy networks, with all the costs for the consumer that entails.

Adam Bell is head of policy at Stonehaven, having previously spent four years at the Department for Business, Energy & Industrial Strategy (BEIS), latterly as head of energy strategy.