By Robyn Lucas, Head of Data Science, Open Energi

The energy landscape is undergoing an unprecedented change, which is accelerating as market barriers to distributed energy are dismantled. The last 12 months have seen standalone energy trading models emerge, access to the Balancing Mechanism widened and new platforms are promising to create new value streams from localised energy services.

There is now a huge variety of distributed energy assets capable of providing flexible capacity to the system – from energy storage, CHPs, electrolysers and Electric Vehicles, to more traditional demand-side response assets such as industrial pumps, boilers and chillers. What all these assets have in common is they need careful managing to deliver the most benefit with the least disruption.

Battery storage optimisation

For a battery storage system, the cost-benefit of every action has to be weighed in terms of battery degradation and lifetime, whilst continuously managing the state of charge to ensure system availability.

With multiple value streams to stack and optimise across timeframes – from day-ahead to real-time – getting the maximum price per hour of operation requires market insight, automated response, an understanding of the constraints of the battery and the site on which it sits, and an appreciation of the risks involved – with buy-in from all parties.

The limiting constraint on value is typically the number of cycles allowed by the warranty – usually around 400 cycles per year for a lithium ion system. This means that the battery can be completely charged up, and then discharged, just over once per day. Therefore, it is important to make this discharge at the right time to reap the largest rewards. It may, for instance, be more profitable to do two cycles on one day and none on another. Accurate forecasting and regular monitoring ensure the best £/kWh of throughput is achieved.

The necessity to stack multiple revenue streams to achieve an ROI that investors are comfortable with means considering these throughput limitations, akin to strike price setting, in a rapidly changing environment. Some revenue streams introduce a reasonably low utilisation, like Static Frequency Response. Meanwhile, others require higher utilisation. For example, throughput whilst tracking frequency in Dynamic Firm Frequency Response (FFR) accounts for around 1.3 cycles per day for a 1-hour system.

As more of the UK’s aging thermal fleet retire and the renewable generation increases, wholesale and imbalance markets are also expected to become more volatile, particularly when the grid is under stress. If a battery storage system is locked into a dynamic FFR contract during an extreme weather event, it may be unable to benefit from profitable price arbitrage opportunities. Balancing the seasonal risk of this against the reward of assured revenue from FFR needs to be decided between the asset manager, investor, and aggregator.

Seasonal volatility

The graph below shows the throughput, and benefit, over a one-year period for a 1.6 hour battery storage system, modelled for 2016 historical prices. The impact of seasonal price volatility is clear: most energy trading arbitrage opportunities occur over winter when prices are more volatile, so throughput will be high at this time. However, in summer the system can be used to provide reduced throughput capacity-based services to maximise the overall £/kW value.


Behind-the-meter models

Last November saw the unveiling of a 2MW battery installed by Pivot Power at Arsenal’s Emirates Stadium – the first behind-the-meter battery to be aimed primarily at wholesale energy trading – powered by the club’s Official Renewable Energy Partner Octopus Energy.

The system is fully automated and optimised by Open Energi’s Dynamic Demand 2.0 platform.

By using the battery to supply the stadium at the most expensive times of day, Arsenal reduces its electricity bill. At the same time, the system is generating revenue – split between Arsenal, Pivot Power and investor, Downing LLP – from energy arbitrage and imbalance opportunities. Crucially, with a limited number of dispatches, optimisation is about identifying the best opportunities.

To manage this, Open Energi assigns a cost to every MWh of throughput and a limit to the number of cycles for each part of the revenue stack. This ensures the optimum pay-off between throughput and revenues. Given the latest wholesale price forecasts and a full understanding of the other non-commodity costs involved, forecasts of the stadium demand (using Arsenal’s match schedule), and knowledge of the physical attributes of the battery system, Dynamic Demand 2.0 uses machine learning techniques to generate the most optimal profile for the system to follow. This is done at multiple timescales: day ahead, intraday, and real time. The simulations below are designed to illustrate how this process works in practice. Figure 1 shows the price signal, as known day ahead, and the resulting optimisation. This optimisation is then updated within the day, in response to a possible triad call: we deviate from the nominated schedule in order to make the most revenue from the possible triad, shown in Figure 2.

Figure 1: Day ahead price signal and resulting optimisation of battery. This is nominated to the supplier to purchase on wholesale market, day ahead.

Figure 2: Actual dispatch of battery according to day ahead schedule, with intra-day update due to a Triad call, where the price for one of the Triads is shown on a logarithmic axis. The actual Triad won’t be known until after the season.

The technology is helping to maximise benefit from assets on sites across the UK, not just premier league football clubs. Understanding the electricity contract for each of these sites is key to unlocking the most value. Figure 3 below shows one such industrial site which has co-located solar. The battery is charged up using excess solar power during the day, and is then used to take the site offline at the most expensive times. The industrial site operator saves money on their electricity bill as they reduce their imports from Grid, and the system also generates revenue by performing FFR.


Figure 3: Impact of battery on industrial site with co-located solar

Here, understanding the intricacies of the Power Purchase Agreements between the various parties involved has been crucial. Open Energi, acting as the aggregator, must have a full picture of the contracts between the solar system operator, the industrial site, and the asset manager. Will the system earn any revenues from export? Is the import and export held by the same supplier, or are they under providers and exposed to different terms and pricing? Are the parties fully aware of what markets and price components they are exposed to?

As more battery storage projects proceed on a merchant basis, creating innovative, multi-partner business models like this, in a sustainable, asset-centric way, will be vital to ensure momentum is maintained towards a low carbon, decentralised energy economy which reduces costs for consumers and maximises use of clean, cheap, renewable energy.


This blog was originally published at

Recommended Posts