A new frontier in Dynamic Frequency – does it stack up?

National Grid ESO’s new dynamic frequency regulation services suite needs fast-responding assets, such as batteries, to enhance grid frequency stability. Operators will be able to pick and choose the services they provide. However, managing the state of charge ahead of time and keeping systems within warranty constraints by stacking these new products to achieve desirable utilisation and revenues adds complexity.

As the National Grid ESO procures Dynamic Regulation with a minimum unit size of 1MW, it could mean significant underutilisation of most existing systems below a 2-hour duration. So what does this mean for the state of charge (SoC), and how we can create the optimum ‘stack’?

In this blog, we explore a theoretical battery’s response to the new Dynamic Response (DR) and Dynamic Moderation (DM) services, modelling the expected impacts to the utilisation, cycling, and State of Charge (SoC) management.

An Increasing Appetite for Energy Storage

The UK’s electricity system is increasingly experiencing lower inertia, and there is a growing risk of more considerable and frequent generation losses. Last year the publication of the Future Energy Scenarios (FES) from National Grid ESO (NGESO) and the Government’s Smart Systems and Flexibility Plan 2021 outlined the need and role for flexibility in a net-zero UK energy system. As variable renewable power replaces conventional generation sources, flexibility not only unlocks increasing potential for renewables and ensures energy security but could also reduce system costs by between £30-70bn by 2050. At least 15GW of shorter duration assets (less than 4 hours) will be needed by 2050 to help balance the system. According to the FES, by 2030, 18GW of shorter duration assets could be built in the fastest credible decarbonisation scenario.  

New Frequency Services for a More Resilient, Lower Carbon System

In recent years we have seen a growing number of low-frequency events, where grid frequency has deviated by more than 0.3Hz (Figure 1). For example, on the 9th August 2019, we saw a blackout; 3% of system demand disconnected, and in London, 60% of the Tube network ground to a halt.

Figure 1. Number of lower than 49.7 Grid frequency events since 2014.

There are different strategies to combat dropping inertia:

  • procuring it directly from a conventional plant or synchronous generators;
  • reducing the size of the most significant single loss on the system (which reduces the risk of instability if that load is lost) or;
  • the procurement of faster frequency response.

Batteries enable the procurement of faster frequency response; they can react to significant frequency disturbances and restore stability in under a second by delivering ‘synthetic inertia’ to the system.

By April 2022, NGESO will have launched the full suite of new faster-acting frequency response services, eventually replacing all legacy frequency products, including Firm Frequency Response (FFR). 

Introduced in October 2020, Dynamic Containment (DC) was the first of these new services, and it now has over 850MW of participating volume. DC operates post-fault, providing a less than 1-second response to frequency deviations of greater than 0.2Hz. DC helps the system recover from sudden significant losses of generation or load.

In March and April 2022, we expect the soft launch of Dynamic Moderation (DM) and Dynamic Regulation (DR). Both DM and DR are pre-fault services. DR acts with a constant and proportional power response across the ‘operational’ frequency range and reaches full power at a +/- 0.2Hz frequency deviation. DM kicks in between +/- 0.1Hz-0.2Hz, giving an additional boost to stabilise frequency as it moves towards the edge of operational limits. Figure 2 compares the droop curves (the power response according to grid frequency) for all three services with the existing FFR product.

Graph showing Droop curves for Dynamic services (left) and FFR (right), showing the power response as % of nominal asset power across the frequency range in Hz.
Figure 2. Droop curves for Dynamic services (left) and FFR (right), showing the power response as % of nominal asset power across the frequency range in Hz.

The New Dynamic Services

Figure 2 shows that DR requires the highest comparative response due to its steep droop curve. We model the impact of performing this service for a 10MW system with infinite storage (i.e. with no energy limit to charging or discharging) using historic frequency data, accounting for efficiency losses. 

DR has twice the utilisation of FFR, at 10% (Table 1), and subsequently double the throughput (cumulative discharge energy). DM and DC have much lower utilisation. A 1-hour system would perform 0.5 and 0.12 cycles per day, respectively, compared to 2.4 for DR.

Table-showing-the-average-utilisation-and-average-throughput-for-Fast-Frequency-Response-and-dynamic-services
Table 1: Average Utilisation (%) and average throughput (MWh/MW) for FFR and new dynamic services. Calculated from historic Grid frequency.

Figure 3 shows the expected power response to a large low frequency event (0.3Hz deviation). DR and DM act to restore grid frequency faster than FFR. While DR and DM services reach their maximum response power levels at or below 49.8 Hz, the DC response stays below 50% of its contracted level.

Figure 2. Droop curves for Dynamic services (left) and FFR (right), showing the power response as % of nominal asset power across the frequency range in Hz.
Figure 3. Response to >0.3Hz low grid frequency deviation for all Dynamic Services and FFR. Here, negative power indicates discharging of a battery.

DC acts in reserve should an additional loss push frequency further away from 50Hz. Theoretically, grid frequency should become more robust to system trips even as inertia continues to fall with these three services acting in concert – vital for the UK’s security of supply. Depending on how much volume NGESO procure, we may therefore anticipate grid frequency to depart from what we see in historic frequency data. The utilisation and throughput of each service may then look different, but using historic frequency still gives us a reasonable idea of what to expect.

The Impact on State of Charge

With the higher utilisation of DR, SoC management becomes increasingly crucial for batteries, ensuring a ‘real’ battery of finite capacity can deliver the service and consider the service’s cost.

In Figure 4, the first plot shows how much the SoC would drift in each of 20 days for the same 10MW system. The bottom histogram shows the SoC drift over EFA blocks 1, 2 or 3 for a more considerable period of historical data. With no actions taken to rebalance the stored energy of a system providing 10MW of DR for 8 hours—or 2 EFA blocks. SoC would have drifted by more than +/-5MWh, 35% of the time.

This scenario suggests that a 1-hour battery would fully charge or discharge to empty after just 2 EFA blocks if it begins delivering the service at 50% SoC, ignoring availability requirements.

Graph-showing-the-state-of-charge-drift-of-a-battery-providing-Dynamic-Regulation
Graph-showing-the-state-of-charge-drift-of-a-battery-providing-Dynamic-Regulation_4Hr_8Hr_12Hr
Figure 4. State of charge drift of a battery providing Dynamic Regulation only.

Participants must provide the service continuously for the periods they have tendered for or risk penalties. Furthermore, NGESO has stipulated a 60-minute minimum energy requirement for energy-limited assets. That means even a 2-hour battery would need a rebalancing action within 8 hours (2 EFA blocks) 34% of the time, and within 12 hours (3 EFA blocks), 42% of the time. Alternatively, providers could de-rate the tendered volume (i.e. MW) offered, which would also be favourable to reduce cycling and preserve the system warranty.

Managing State of Charge

Baselining actions must be taken to manage SoC while a battery performs DR. We simulate a 10MW/10MWh battery, de-rated to 4MW, delivering DR over three days continuously. The SoC is managed with spare MW using an automated control algorithm that attempts to maintain a 4MWh of headroom and footroom, equivalent to a 60-minute energy requirement.

Following BM requirements, units performing DC baselining actions are planned 1-hour in advance. Under these conditions, the battery requires, on average, three rebalancing actions per day, with 3.8MWh of energy needed for baselining. Figure 5 shows simulated battery power and energy while responding to historical frequency deviations.

Graph-showing-modelled-Energy-management-with-stacked-dynamic-services-2MW-Dynamic-Containment-to-7MW-Dynamic-Regulation
Figure 5. Simulated State of Charge with a 10MW, 1-hour battery responding to Grid frequency with a Dynamic Regulation response curve and a de-rated 4MW DR tender. Automated baselining is applied to maintain system energy around 50%.

Given the ahead-of-time baselining requirement, the rebalancing algorithm fails and leads to 5 hours per day outside the 60-minute minimum energy requirement for a 4MW tender on a 10MWh asset. So a 1C system de-rated to 40% of its nameplate capacity–in effect becomes a 0.4C (or 2.5h) system that could still cycle over 400 times per year and would still not meet a 60-minute minimum availability requirement.

Stacking Dynamic Services

National Grid ESO has stated that the three services can be stacked together (albeit not in the initial soft-launch phase). A single battery can deliver different services simultaneously but with each MW partitioned to provide a single service.

Since DR has high utilisation, we limit the DR tender to much less than the battery’s power rating. This prevents exceeding warranty limits on cycling and mitigates against the cost of a substantial SoC rebalancing regime, which may require peak-time charging to maintain availability. 

To monetise the remaining MWs, we stack lower utilisation services on top. Figure 6 below shows a low grid frequency event under two scenarios: the first with a complete 10MW DR response; the second with a stacked tender of 5MW DR and 5MW DC. In the latter case, during a frequency event below 49.8Hz, utilisation averages 60% of the total tendered volume during the period, rather than 100% for the DR response alone.

 Graph showing the resulting power response of Regulation service vs stacked Regulation and Containment services during a low-frequency event. NB negative power indicates discharging
Figure 6. Resulting power response of Regulation service vs stacked Regulation and Containment services during a low frequency event. N.B. negative power indicates discharging.

Towards an Optimal Stack

We can assess the ‘best’ stack for battery size by calculating utilisation and cycling for a range of tender stacking combinations. For example, figure 7 shows the utilisation and mean daily cycles when we combine DC and DR. The x-axis shows the share of DC, such that 10 represents 10MW of Containment and 0MW Regulation. 

If we must also operate below a “Warranty Cycles” level set to 400 cycles per year, a 6MW/4MW split is best for a 1-hour system (LHS of 7). If we must not exceed 5% utilisation, we also require 6MW/4MW. However, note that since a 1-hour system will likely not provide more than 5MW DR because of the energy requirement limits, the trend below 5 is purely theoretical.

The right plot shows how a 2-hour battery (10MW/20MWh) could significantly loosen constraints on delivering the Regulation service – reaching 8MW of Regulation to 2MW of Containment before reaching the Cycles limit. But, of course, the optimal split should also consider revenue potential. The revenue potential depends on the available auction prices, which are also influenced by the National Grid ESO requirements and, otherwise, the technical barriers to participation.

Graph-showing-the-utilisation-and-number-of-daily-cycles-for-a-10MW-battery-when-different-ratios-of-stacked-dynamic-services-are-selected
Figure 7. Utilisation and number of daily cycles for a 10MW battery when different ratios of stacked dynamic services are chosen.

A Real-World Stacking Strategy

We simulate a potential ‘real-world’ scenario, stacking DR and DC. We choose a 2-hour battery with an 8-2 tender split between DR and DC to meet the cycling constraint of approximately 1.1 cycles per day (400/year). However, reserving capacity as headroom for managing energy, we need to de-rate the tender to 7MW-2MW, preferring the lower-utilisation service. Services are presumed to be symmetrical (both high and low response). Our SoC management aims to keep the system close to 50% and maintain the minimum energy requirement.

The optimum solution compromises between minimising the number of actions taken to balance SoC, which could be expensive, and taking sufficient measures to limit the chances of violating minimum energy constraints and protecting the battery’s life through minimising time spent at extremely high or low energies. Figure 8 shows the modelled response over three days.

With a 2MW/7MW DC/DR split, for a 10MW 2-hour system, we see ca. 7.3% utilisation and 0.83 cycles per day. Balancing SoC requires a mean daily baseline volume of 6.2MWh. Despite the current cost uncertainties associated with providing grid services (due to recent high electricity prices), the simulated case would, if exposed to only cash-out prices, cost more than £3/MW/H.

Whilst achievable compared to current FFR prices, we also have to consider the penalty of higher utilisation and the opportunity cost of going elsewhere for traded revenue.

Graph-showing-modelled-Energy-management-with-stacked-dynamic-services-2MW-Dynamic-Containment-to-7MW-Dynamic-Regulation
Figure 8. Modelled Energy management with stacked dynamic services: 2MW Dynamic Containment to 7MW Dynamic Regulation. Mean daily cycles are under the target of 1.1.

In conclusion

Modelling the forecasted utilisation of 1C batteries with historical grid frequency can deliver DC and DM. However, the pre-fault DR service requires a significant MW de-rating or a longer duration battery to provide a sustained service energy requirement. Therefore, longer duration storage and higher cycling capability could offer good market commercialisation opportunities.

We can generate a greater return from batteries while keeping cycling within a ‘standard’ warranty by stacking high and low utilisation services. Modelling the delivery of this type of stacked service with historic grid frequency demonstrates a trade-off between efficient state of charge management and cost minimisation. Imposing operational constraints such as ahead-of-time rebalancing notification for a high utilisation service presents a challenge even for 2-hour systems. Taking baselining actions ahead of time and offering multiple services at once is operationally complex within system warranty constraints.

Furthermore, if National Grid ESO is to procure these services (particularly DR) with a minimum unit size of 1MW, it could mean significant underutilisation of battery capacity at the sub-10MW level. Ultimately, the prices of these services should reflect this complexity and potential inefficiency. Given the difference in utilisation and technical requirements, it will be fascinating to see where price levels transpire between the three services.

Unlocking the investment necessary to meet the ambitious storage goals of the Smart Systems and Flexibility Plan 2021 requires confidence in the revenue streams for these assets. The existing ‘stack’ of revenue streams open to storage assets can be uncertain and high risk, and barriers to entry remain. Economies of scale, volume caps, licencing, and changes to demand charges (namely the Targeted Charging Review) favour large standalone storage assets. More planning relaxation will further incentivise the building of systems greater than 50MWh.

These new markets will undoubtedly form an essential part of the flexibility revenue stack and should provide investors with confidence in storage revenues’ long-term security. Even without operational examples for Dynamic Regulation and Dynamic Moderation, as with Dynamic Containment, battery storage appears favourable to deliver these services at the lowest cost.

However, with wholesale market volatility, trading opportunities elsewhere may yet present a more attractive option for battery operators looking to maximise revenue stacking if the cost-of-service provision, combined with high technical barriers in grid services, outstrip the benefits.

To find out more about our products and services, email us; info@openenergi.com.

How the rise of ‘Energy as a Service’ can power decarbonisation

open energi wind farm

Energy as a Service is the latest business model innovation to arrive in the energy supply industry. In short it is all about moving away from buying energy on a per unit (p/kWh) basis and moving towards a fixed fee per month within certain volume thresholds; akin to how we pay for mobile phone contracts. Energy as a Service has emerged off the back of disruption to the way we supply, consume and now ultimately buy energy, which has fundamentally changed energy market economics.

This disruption is the result of four major technology-driven trends:

  • Decarbonisation – The growth of energy supply from zero marginal cost renewable resources
  • Decentralisation – The growth in energy generated from smaller scale low carbon resources either on customer sites (Behind-the-Meter) or at the Distribution Level (Distributed Energy)
  • Digitisation – The ability to measure and monitor machine behaviour in real-time and automate how we use and supply energy
  • Democratisation – The rise in consumer participation, control and choice which is increasingly determining how energy is bought and used

Traditional per unit models work where the dominant cost in delivery of the product or service scales according to the volume used. This was true when the majority of power supplied came from sources that required a fuel input e.g. coal and gas. The more energy consumed the greater the proportional cost of buying and burning that fuel to generate more kWhs of power.  Other components which make up the total ‘at-the-meter’ price have also been charged on a per unit basis to ensure those who use more of the electricity network pay more for it; government taxes, utility profit margins and network charges (with some time-of-use element).

However, when you start to use zero marginal cost power the economics get flipped on their head. Renewable ‘fuel’ is free, so the dominant cost in consuming energy becomes the infrastructure needed to deliver it. Wind turbines, PV panels, transmission and distribution cables have low operational costs once built, so the initial capital expenditure is where the dominant cost lies.

Across Europe average wholesale prices now reflect wind and sun patterns more than the cost of coal and gas, and at periods of low demand and high renewable output we consistently see negative prices. Clearly change is needed as consuming more energy at these times is beneficial to the whole system but a per unit charging mechanism disincentivises users from doing that.

Enter, Energy as a Service. Already we are seeing a shift in network charging towards capacity-based charges instead of use-of-system charges. Wholesale prices are not far behind; the task becomes providing the flexibility to firm up renewable output. Thanks to the digital revolution described above this flexibility can come from consumers’ demand, cost-effectively tapping into flexibility inherent in distributed energy resources behind-the-meter.

Take a given offshore wind site, with known capacity factors of about 50%. It is possible to quantify the amount of flexible energy needed to ensure 99% of customer demand is met at all times. Using existing business assets means it is possible to take advantage of zero marginal cost flexibility in everyday processes (such as heating, cooling, pumping, battery storage and CHPs), avoid unnecessary infrastructure upgrades and minimise efficiency losses in transporting power. Once it is understood how much flexible power is needed to firm up the output of renewable generation the next task is what technologies do you use to meet that flexibility requirement.

Artificial intelligence-powered flexibility platforms – like Open Energi’s Dynamic Demand 2.0 technology – which can manage distributed energy resources in real-time, are critical. They can evaluate the amount of flexibility in existing power-consuming assets and processes – in addition to any battery storage and/or flexible generation (such as CHPs) – and map demand to supply. This then becomes a constant, real-time scheduling problem for the platform to manage; invisibly ramping processes up when wind is abundant and storing as much power as possible, or turning processes down to a stable minimum and discharging batteries or using a CHP when wind output is low.  If real-time scheduling isn’t maintained, the cost structure breaks down, so the reliability of these platforms is critical.

What is important to recognise here is that below a certain demand threshold the marginal cost of putting in place this service is the cost of operating the wind and the software required to schedule behind-the-meter flexibility. This is why Europe’s utilities are making huge investments and acquisitions in virtual power plant technology.

By doing so the costs of delivering energy become fixed and predictable and scale with size of connection instead of actual usage. Exactly like the mobile phone industry where the marginal cost of sending a packet of data is immaterial in comparison to network costs of all infrastructure.

For Open Energi Energy as a Service has always been the natural end-game in maximising the value of Demand Response. It shelters consumers from the continuously changing and complex incentives of the existing Demand Response markets, and instead offers a simple proposition: “By installing demand response software across a range of assets you can pay a lower fixed monthly fee for your energy”.

The clarity and certainty offered by Energy as a Service makes it easy to structure simple, long-term financing solutions for different technologies – e.g. solar PV, energy storage, CHP – and allows businesses to concentrate on what they do best.  All the complexities of power procurement and demand response markets are removed in place of a known fixed fee per month that ensures reliable, clean and affordable energy. 

David Hill, Commercial Director, Open Energi

This blog was originally posted on Current News.

Battery storage project a ‘blueprint’ for EV charging infrastructure globally

Tesla South Mimms Supercharger and PowerPack

Pairing batteries with EV charging stations can help to align sustainable transport and energy needs for the future.

At South Mimms Welcome Break Motorway Services, we have installed a 250kW/500kWh Powerpack alongside one of Tesla’s largest and busiest UK charging locations. The Supercharger site can charge up to 12 cars at one time, and since popular charging periods often coincide with peak periods of grid demand – between 4pm and 7pm, when electricity prices are at their highest – flexible solutions are needed to ease the strain on local grids and control electricity costs.

Integrating a Powerpack at the location has meant that during peak periods, vehicles can charge from Powerpack instead of drawing power from the grid. Throughout the remainder of the day, the Powerpack system charges from and discharges to the grid, providing a Firm Frequency Response (FFR) service to National Grid and earning revenue for balancing grid electricity supply and demand on a second-by-second basis.

Open Energi own and operate the Powerpack, which is part of our portfolio of assets that help maintain the frequency of the grid. Combining batteries and electric vehicles makes vehicle charging part of the solution to integrating more renewables without affecting drivers, unlocking vital flexibility to help build a smarter, more sustainable system.

The project at South Mimms Welcome Break Motorway Services provides a blueprint for the development of electric vehicle charging infrastructure globally. Moreover, by reducing National Grid’s reliance on fossil fuelled power stations as a means of balancing electricity supply and demand, the Powerpack helps to reduce UK CO2 emissions by approximately 1,138 tonnes per year.

How Artificial Intelligence is shaping the future of energy

Artificial Intelligence can unlock demand side flexibility for end users

Across the globe, energy systems are changing, creating unprecedented challenges for the organisations tasked with ensuring the lights stay on. In the UK, large fossil fuelled power stations are being replaced by increasing levels of widely distributed wind and solar generation. This renewable power is clean and free at the point of use but it cannot always be relied upon. To date National Grid has managed this intermittency by keeping polluting power stations online to make up the difference but Artificial Intelligence offers an alternative approach.

What’s needed is a smart grid which can integrate renewable energy efficiently at scale without having to keep polluting power stations online to manage intermittency. This requires energy storage to act as a buffer, reducing demand when supply is too low or increasing it when it is too high. Most people associate energy storage with batteries, but the cheapest and cleanest type of energy storage comes from flexibility in our demand for energy.

This demand-side flexibility takes advantage of thermal or pumped energy stored in everyday equipment and processes, from an office air-con unit, supermarket fridge or industrial furnace through to water pumped and stored in a local reservoir. The electricity consumption patterns of these types of devices are not necessarily time-critical. Provided they operate within certain parameters – such as room temperature or water levels – they can be flexible about when they use energy.

This means that when electricity demand outstrips supply, instead of ramping up a fossil fuelled power station, certain types of equipment can defer their electricity use temporarily. And if the wind blows and too much electricity is being supplied instead of paying wind farms to turn off we can ask equipment to use more now instead of later.

Making our demand for electricity “intelligent” in this way means we can provide vital capacity when and where it is most needed and pave the way for a cleaner, more affordable, and more secure energy system. The key lies in unlocking and using demand-side flexibility so that consumers are a) not impacted and b) appropriately rewarded.

At Open Energi, we’ve been exploring how artificial intelligence and machine learning techniques can be leveraged to orchestrate massive amounts of demand-side flexibility – from industrial equipment, co-generation and battery storage systems – towards the one goal of creating a smarter grid.

We have spent the last 6 years working with some of the UK’s leading companies to manage their flexible demand in real-time and help balance electricity supply and demand UK-wide.  In this time, we have connected to over 3,500 assets at over 350 sites, operating invisibly deep with business processes, to enable equipment to switch on and off in response to fluctuations in supply and demand.

Already, we are well on the way to realising a smarter grid, but to unlock the full potential of demand-side flexibility, we need to adopt a portfolio level approach. Artifical intelligence and machine learning techniques are making this possible, enabling us to look across multiple assets on a customer site, and given all the operational parameters in place, make intelligent, real-time decisions to maximise their total flexibility and deliver the greatest value at any given moment in time.

For example, a supermarket may have solar panels on its roof and a battery installed on site, as well as flexibility inherent in its air-con and refrigeration systems. Using artificial intelligence and machine learning means we can find creative ways to reschedule the power consumption of many assets in synchrony, helping National Grid to balance the system while minimising the cost of consuming that power for energy users.

Lack of data is often an obstacle to progress but we collect between 10,000 and 25,000 messages per second relating to 30 different data points and perform tens of millions of switches per year. This data is forming the basis of a model which can look at a sequence of actions leading to the rescheduling of power consumption and make grid-scale predictions saying “this is what it would cost to take these actions”. The bleeding edge in deep reinforcement learning shows how, even with very large scale problems like this one, there are optimisation techniques we can use to minimise this cost beyond what traditional models would offer.

Artificial Intelligence model learning to control the electricity consumption of a portfolio of assets

Graph of AI model

More rapid progress could be made across the industry if energy companies made more anonymised half-hourly power data available. It would enable companies working on smart grid technologies to validate these ideas quickly and cheaply. In the same vein, it would be a major breakthrough for balancing electricity supply and demand if energy companies made available APIs for reporting and accessing flexibility; it would allow companies like Open Energi to unlock enormous amounts of demand-side flexibility and put it to good use balancing not just the grid but also helping to optimise the market positions of those same energy companies.

In the UK alone, we estimate there is 6 gigawatts of demand-side flexibility which can be shifted during the evening peak without affecting end users. Put into context, this is equivalent to roughly 10% of peak winter demand and larger than the expected output of the planned Hinkley Point C – the UK’s first new nuclear power station in generations.  Artificial Intelligence can help us to unlock this demand-side flexibility and build an electricity system fit for the future; one which cuts consumer bills, integrates renewable energy efficiently, and secures our energy supplies for generations to come.

Michael Bironneau is Technical Director at Open Energi. He graduated from Loughborough University in 2014 with a PhD in Mathematics and has been writing software since the age of 10.