The Energy Onion — a simple conceptual model for a smart system

David Sykes
8 min readJul 8, 2020

The UK is in the process of redesigning its electricity system to be fit for a Net Zero future. It’s a fascinating challenge that requires balancing interactions across many disciplines:

  • Physics and engineering — how do we adapt our networks to transport the energy we need?
  • Economics and markets — how do we do this at the least cost using markets for price discovery?
  • Technology and communications — how do we coordinate and automate all the players in our energy system in a secure way?
  • Behavioural economics — how will customers react to the incentives in this system?
  • Planning — how does this system serve new models for housing, heat and transport?
  • Regulation — who pays for all of this?
  • Social policy — how do we make this system work for the most vulnerable in society?

These questions are being played out on multiple stages in central government, local government, the regulator, the networks and system operators and in numerous industry code bodies.

There’s been an amazing amount of progress, innovation and collective thinking to date with almost unanimous consensus on the need to get to Net Zero. This is a testament to the fantastic group of people and businesses that make up the UK energy industry.

To steal the golden circle analogy from Simon Sinek of “what, how, why”, as an industry we are firmly agreed on the WHY.

WHY: We need to change our system to deliver Net Zero and avert catastrophic climate change.

The challenge the industry faces is it is now reaching for the WHAT without thinking about the HOW. Or in more practical terms, we have buried ourselves in consultations, code modifications, specific policies and implementation detail without stepping back and asking HOW do we want to design the system to deliver Net Zero.

The current, generation driven model

In my opinion, the main blocker to this is that we are still working on an old conceptual model about how to design and operate an energy system.

This is a model where we plan, optimise and operate the system from the top down. This system is characterised by:

  • Unidirectional flow of electricity from high voltage generation to low voltage demand
  • A small number of large generation assets
  • An inflexible, single peak, national demand curve built up of passive off-takers

The design and operating principles are:

  • Treat demand as atomic, non-cooperative, non-interactive entities
  • Rely on the diversity of customers rather than coordination to size network capacity
  • Use generation as the tool to meet demand
  • Balance and optimise the whole system from a central command model

A new demand driven model

The energy system we are hurtling towards doesn’t look like a linear production line. It is characterised by:

  • Multidirectional flow of electricity around the networks
  • A large number of small generation assets often colocated with or near demand
  • A set of local demand curves, built up of active off-takers, with different peaks at different points in the networks

The old model just doesn’t work for this. Central system optimisation will miss out on huge opportunities for local optimisation. Big, lumpy, coordinated loads in the low voltage network will break our diversity assumptions and if we continue to match generation to demand we’ll either miss our Net Zero targets and end up with a fabulously expensive system.

We need a new conceptual model that is completely counter to all the principles of the old model. The new model should start at the network edge, not the centre and be based around:

  • Maximising network utilisation and renewable generation
  • Treating demand as cooperative and interactive
  • Shaping the demand curves in the different parts of the networks to meet generation

I propose a new model for thinking about the system based on hierarchical optimisation from the customer outwards. Conceptually this can be thought of as an onion.

We should design our policies, cost recovery, system operating model, balancing and settlement to reward optimisation from the inside of the onion outwards. Each layer should be designed and operated to optimise two very clear goals:

  1. Maximise the consumption of renewable energy vs non-renewable energy
  2. Minimise the pressure on the outer layers of the onion when they are most constrained — e.g. avoid importing energy from the networks when there is no space on them

The layers

The onion starts with energy efficiency — you don’t need to build generation and networks for energy you don’t use. For an in depth discussion on why energy efficiency is not only good physics but good policy, see Michael Liebreich’s latest piece. Energy efficiency doesn’t feature highly when you take a generation driven approach because if people use more energy the system just finds more generation to meet the extra demand. The Capacity Market is a great illustration of this effect, we spend over £1bn a year on adding enough generation capacity to meet demand when we could spend that same amount on permanently reducing that peak demand through energy efficiency. When you take a bottom up approach, energy efficiency falls at the centre of system design.

The next layer is self consumption — this means we encourage households and businesses to generate their own energy and shift their own demand (or store it) to consume it themselves. This has the benefits of avoiding technical losses in the networks and minimising constraint in the low voltage networks as well as maximising consumption of locally generated green electricity.

The next layer is local balancing and cooperation — this means we encourage the balancing of generation and demand at a local level. In practice this might be communities sharing generation or storage assets or suppliers incentivised to develop portfolios of customers that can intelligently net off within parts of the low voltage networks. To achieve this we need a more granular, hierarchical settlement model. Currently settlement can only be resolved down to the Grid Supply Point. We need to be able to resolve settlement down to the local feeder. This doesn’t mean all cost signals are charged at this level but it gives the flexibility to resolve demand at different levels in the network and therefore reward efficient use of the local networks. Currently the system is very poorly set up to reward local balancing, this failure to realise the value of local demand matching is one of the main reasons why community energy is stalling.

The next layer in the onion is Distribution System Operator (DSO) Balancing — this means that we incentivise the DSO to efficiently use their existing network assets before the Distribution Network Owner (DNO) invests in new ones. The inner layers of the onion should already be working hard to optimise network utilisation. The DSO can further use network charging signals to flatten the locational demand curves within their network, penalising generation or demand when and where the networks are most constrained. Contracted flexibility (ideally delivered through real time markets not tenders) can then be used as the final scalpel-like tool to optimise network utilisation. The DNO/DSO transition would go a long way to start to better define this layer of the onion. In its current state, it is poorly defined with little or no incentives provided to network participants to optimise at this level.

The final, outer skin of the onion is the Electricity System Operator balancing role. The ESO balances and stabilises the transmission system and is incentivised to:

  • Efficiently use the existing network infrastructure
  • Minimise the cost of balancing and stabilising the system
  • Facilitate an increasingly high proportion of renewables in the generation stack

This is perhaps the best developed layer of the onion in the current system because it is closest to the existing top down control and dispatch model. The difference in the onion model is that ESO balancing tools should be the last thing we need to apply once all the inner layers of optimisation have been applied. Equally, the ESO needs to work in concert with the DSOs to achieve a stable, balanced system. Effectively defining the interface between ESO and DSO will be key to operating an efficient system.

I am not suggesting here that the ESO role is minimised. In fact transmission of large scale renewable energy and grid level balancing and stabilisation will be more important than ever. However, the concept of a central organisation coordinating all the assets in the system is like suggesting a general should give individual orders to every soldier. An army or organisation is most effective when split into manageable units with a clear set of duties and targets and reporting lines to the next level in the organisation. The energy system should be no different.

Where does flexibility fit in?

There is a lot of talk around the need for huge amounts of flexibility to enable a renewable based system. However, this is not yet backed up by clear value signals to encourage the uptake of flexible technologies and behaviours at all levels in the system. This is because we are trying to shoehorn flexibility into the existing system design and can only move at the pace at which the networks can design and implement flexibility markets.

We’re also stuck in a Catch 22. Flexibility markets won’t be effective without a critical mass of flexible assets but we won’t achieve a critical mass without the right market incentives. This is why destruction of charge based price signals before the introduction of market based flexibility signals is so disappointing and damaging to the uptake of domestic flexibility.

The demand based model above implicitly values flexibility because it is central to the optimisation goals of each layer. For instance, effective self consumption is impossible without flexibility, as is efficient local balancing. By turning the system model on its head we place flexibility at the heart of the system rather than adding it as a thin layer on top through contracted flexibility markets.

How do we achieve this?

For each layer, and the system as a whole, to achieve it’s dual goals of maximising renewable consumption and minimising overheads on the networks we need three things:

  1. Clear cost signals that reflect carbon intensity and network constraint in real time
  2. A rethink of our settlement model to better reflect the benefits of local settlement — this doesn’t necessarily mean nodal pricing, it means having a better toolset to reward efficient local balancing
  3. An iterative approach — moving to this way of operating doesn’t need big bang industry changes but can be achieved through a combination of the existing reforms and mass in-market, in-system testing, iteration and calibration. We should be thinking about all ongoing reforms like RIIO2 and Access and Forward Looking Charges in the context of an overarching operating model. At the same time we need to develop rapid in-system, in-market testing for cost signals and settlement.

The key is that we don’t delay any longer, the tech is ready, the customers are ready, the industry appetite is there…we just need to start thinking differently.