2017 in review: breakthrough tech and tumbling renewable records demand greater flexibility

2017 was a year of dramatic change in the UK electricity market. Overall, total UK electricity consumption fell 2.8% compared to the previous year: 264 TWh compared to 272 TWh in 2016[1]. This follows the long-term trend of decreasing peak and total yearly energy use, while the proportion of renewable generation continued to rise: 2017 smashed 13 clean energy records, low carbon generation exceeded fossil fuels, and the resulting trend for negative prices (as recent as last week in Germany thanks to high wind) looks set to continue.

Last year also saw a fall in the strike price for new offshore wind power to £57.50/MWh. Considering the government’s guaranteed price for Hinkley Point C is £92.50/MWh, it highlights just how competitive renewables, and particularly offshore wind, now are. However, the system must be able to cope with the intermittency that all this cheap, carbon-free power brings.

Figure 1 shows the huge variation in demand over the year: from peaks of nearly 50GW on winter evenings, to troughs of around 17GW on summer nights. Figure 2 shows the average daily profile of consumption. In 2017, the prize for peak demand goes to January 26th, which came in at 49.76 GW at 6pm. Compare this to the profile of June 11th, the day that the UK used the least energy: at 5am, it was 16.57 GW. This swing of over 30 GW presents many challenges for the system operator as more and more of the generation becomes intermittent and demand patterns shift: there is value in being flexible with one’s electricity consumption.

 

Figure 1. Daily demand over the year, and smoothed trend over the year.
Figure 1. Daily demand over the year, and smoothed trend over the year.

 

Figure 2. Peak and lowest demand of 2017, compared to the average daily profile.
Figure 2. Peak and lowest demand of 2017, compared to the average daily profile.

Historically, our electricity system has been built to cope with the peaks; and paying for this network accounts for around 30% of your electricity bill (and rising). What if, by being a bit smarter about when we use our electricity, we could flatten the swing out a little? Or better still, align it to renewable generation?

This is where demand flexibility comes in, empowering consumers and playing a vital role in providing the responsiveness needed to cope with huge swings in renewable generation as it makes up more and more of the UK’s generation mix.

Transforming the network

2017 will be known as the break-through year of batteries and electric vehicles (EVs). With the dramatic fall in battery prices we’ve seen a rush of parties buying up battery capacity, hoping to profit from what were lucrative flexibility markets. National Grid have seen batteries flooding into the Firm Frequency Response (FFR) market, attempting to secure profitable long-term contracts to satisfy investors. Market dynamics mean this is increasingly challenging. In a rapidly changing marketplace, a variety of revenue streams must be considered. Battery operation must encapsulate multiple markets to insure against future movements and maximise profits, while ensuring safe and careful operation of the asset such that state of charge, warranty, and connection limits are respected, an area where Open Energi has significant expertise.

EV take-up is accelerating more quickly than many estimated – UK sales of EVs and plug-in hybrids were up 27% in 2017 – and the need for managing this additional demand in a smart, automated way is crucial to alleviate strain on local networks. We have explored the enormous potential of EVs to provide flexible grid capacity and are working with a consortium to deliver the UK’s first domestic V2G trial.

While in the short term, as the big electricity players try to keep up with the changing needs of the system, flexibility markets present a degree of uncertainty, the long term need for demand-side response (DSR) and frequency regulation cannot be underestimated.

Grid Frequency and FFR

As the System Operator, National Grid must maintain a stable grid frequency of 50Hz. Generation and demand on the system must be balanced on a second-by-second basis to ensure power suppliers are maintained. Traditional thermal plant operates with physically rotating turbines, which carry physical inertia and act to stabilise the frequency. With the increase in generation from non-inertial sources (e.g. wind turbines, which don’t carry inertia in the same way, and PV cells), this stability is reduced. Larger deviations in frequency can result in the event of a power station, or interconnector trip, for example.

During 2017, the largest low frequency event (demand greater than supply) occurred on 13th July, when it dropped to 49.57Hz. Given that National Grid’s mandate is to keep it within 0.5Hz of 50Hz, this was rather close! Figure 3 shows the period, and we see a sudden drop in frequency which typically indicates the trip of a significant generator. In this case, the fault was at the French interconnector. Here, what usually functions to improve energy continuity and smoothen geographical variations in supply was the culprit for the biggest second-by-second imbalance in 2017!

The largest high frequency event (supply greater than demand), during which frequency reached 50.41Hz, occurred at the end of October, was much more gradual and seems to have been due to a combination of several effects. Demand typically drops quite steeply this late in the day, so large CCGT plants are reducing their output and on this occasion a sudden drop in wind-generation seemed to have been over-compensated by pumped storage.

Figure 3. Lowest and Highest frequency extremes in 2017.
Figure 3. Lowest and Highest frequency extremes in 2017.

As well as these relatively rare large frequency events, there are excursions that can last for several hours. Figure 4 shows two periods where the frequency deviated from 50 Hz. In general, the average frequency is 50Hz, and therefore any response to frequency regulation averages out to zero. However, over these medium-term time periods the average frequency is not 50Hz. For flexible assets like batteries, that are dynamically responding to correct grid frequency during such periods (performing FFR) the state of charge is affected.

For this reason, the state of charge of the battery must be actively, and automatically, managed – so that optimal state of charge is quickly recovered after such events. The battery is then able to continue to perform FFR, or other services such as peak price avoidance or price arbitrage in wholesale markets. The state of charge (bottom panels in Figure 4) can also have strict warranty limits set by the manufacturer.

Figure 4: Extended frequency events and impact on battery state of charge
Figure 4: Extended frequency events and impact on battery state of charge

Interestingly, 2017 saw an increase in both the number of frequency events (usually defined as frequency excursions larger than 0.2Hz away from 50Hz), and frequency mileage (defined as the cumulative deviation of the grid frequency away from 50 Hz), shown in Figure 5, particularly during the spring and autumn.

Could this be due to the large, somewhat unknown amount of PV on the system? It is distributed, meaning National Grid see PV generation as a fall in demand; they also have no control over it (unlike most other generation). PV efficiency is high in cold weather, so perhaps unexpectedly high and erratic solar generation on cold, sunny days in the Spring and Autumn led to a more unstable system this year, compared to 2016.

Figure 5. The grid has experienced more mileage and more events in 2016 than 2017, especially in March and October. Frequency “event” here is defined as a deviation of 0.1 Hz around 50Hz.
Figure 5. The grid has experienced more mileage and more events in 2016 than 2017, especially in March and October. Frequency “event” here is defined as a deviation of 0.1 Hz around 50Hz.

Figure 5. The grid has experienced more mileage and more events in 2016 than 2017, especially in March and October. Frequency “event” here is defined as a deviation of 0.1 Hz around 50Hz.

The rise of distributed generation, accelerating EV uptake, and plunging battery storage costs, are all driving a rapid transformation in the UK’s electricity system.  Managing these changes requires new approaches.  Demand-side response technologies, like Open Energi’s Dynamic Demand 2.0 platform, mean patterns of demand can be shifted in a completely carbon neutral way; enabling electricity to be consumed when it’s being generated: as the wind blows, or the sun shines. Rather than inefficiently changing the output of a gas fired power station to meet demand, we can make smart changes in demand up and down the country to meet generation, deliver local flexibility, and put consumers in control of their energy bills: delivering completely invisible, completely automated, intelligent DSR which paves the way for a more sustainable energy future.

By Wouter Kimman, Data Scientist, Open Energi

[1] For demand here and throughout this post we use INDO values as reported by ELEXON Ltd.

The future of flexibility: making the most of behind-the-meter storage

Camborne Energy Storage

The slowdown in the Firm Frequency Response (FFR) market, where prices are falling and volumes have been capped – in large part thanks to the decreasing cost of battery storage – has been a big talking point of late.

National Grid has now procured much of the low response it needs for the next two years, and average prices of accepted bids have decreased 20% over the last 7 months. National Grid has also announced proposed changes intended to rationalise and reform the current suite of balancing services it procures, including a streamlined FFR-type product.

These rapidly changing market conditions may leave some battery companies going back to the drawing board and asking how to guarantee income over a long period and satisfy investors that it is worth investing hundreds of thousands in the latest battery storage system, without the relatively high revenues that providing FFR has historically guaranteed?

Fig: The distribution of FFR accepted tenders for the period January - October 2017. Source: National Grid
Fig: The distribution of FFR accepted tenders for the period January – October 2017. Source: National Grid

But it is possible to turn to other markets for flexibility. FFR will remain a part of the revenue stream for batteries, and for the fast-acting Demand Side Response (DSR) that makes up most of Open Energi’s FFR portfolio; but being a bit smarter about where, and how, you employ the flexibility at hand can open up a wealth of other revenue streams. These, stacked up, provide a stronger business case than FFR alone.

Open Energi’s new platform, Dynamic Demand 2.0, is designed with exactly this in mind. By connecting to distributed energy assets which have inherent flexibility in their electricity consumption or generation, including industrial equipment, electric vehicle charging stations, and of course battery storage systems, it enables businesses to dispatch this flexibility in the most valuable place at any one time.

Tailored to the particular operational or site constraints of an asset, balancing services, energy trading, the capacity market, network constraint management, peak price management such as DUoS red-band and TRIAD avoidance are all opened up as stacked revenue streams, as well as operational energy efficiencies automated with machine learning.

For behind-the-meter batteries, this means considering any on-site generation or demand, the import and export limits of the site, the warranty particulars of the battery, and evaluating all the potential revenue streams of the asset. By taking a holistic view of the energy market, using state of the art machine learning techniques and cloud based technology, we remotely operate the asset to maximise its return on investment: FFR is just one slice of the cake.

In many ways, a battery is the perfect asset for flexibility. It has a defined storage capacity; it can discharge power up to a maximum well-defined rate; and it has a known state of charge (SoC) at any one time. However, to operate a battery across multiple markets, careful management of the SoC of the system is necessary.

While batteries very naturally perform frequency response, charging up when frequency is high – removing excess electricity from the grid, and discharging when frequency is low – when more electricity is required, the operation is not quite that simple. To have the capacity, or availability, to both charge up and discharge in line with grid frequency, the ideal SoC of a battery is 50%.

Periods of high or low grid frequency can rapidly take the SoC of the system far from 50%, and batteries have an inherent efficiency (around 90% in the best systems): active state of charge management is required to maintain the availability for frequency response.

The figure below shows the same 500kW, 800kWh battery over the same 52-hour period, with and without state of charge management. The availability to fully charge or discharge at the capacity of the battery for 30 mins (equivalent to a frequency 50.5Hz or 49.5Hz for half an hour) – is quickly impeded on when no SoC management is present, and SoC will eventually tail off to zero due to the non-perfect efficiency of the battery. Battery throughput, defined as the cumulative discharge of energy through the battery, is also higher without SoC management, eating into that allowed by the warranty.

Fig: The state of charge (%, top), availability for FFR (kW, middle) and throughput (kWh, bottom) of a 500kW, 800kWh system are shown across 2 days of frequency data. With no SoC management in action, state of charge is routinely out of the dotted lines which signify ½ hour of storage capacity being available to charge and discharge at full power; here, state of charge is low. As a result, low availability for FFR is diminished. Throughput of the battery, defined as the cumulative sum of the battery discharge, is also higher when no SoC management is used.
Fig: The state of charge (%, top), availability for FFR (kW, middle) and throughput (kWh, bottom) of a 500kW, 800kWh system are shown across 2 days of frequency data. With no SoC management in action, state of charge is routinely out of the dotted lines which signify ½ hour of storage capacity being available to charge and discharge at full power; here, state of charge is low. As a result, low availability for FFR is diminished. Throughput of the battery, defined as the cumulative sum of the battery discharge, is also higher when no SoC management is used.

When introducing revenue streams in addition to FFR, as we do with Dynamic Demand 2.0, battery operation gets rather more complex. For a behind-the-meter battery, electricity bills can be significantly reduced by discharging over peak periods to either avoid importing electricity when it is most expensive (if the site has demand), or exporting to grid and taking advantage of the various network charging revenue streams (DUoS, TRIAD, CM Levy).

Electricity generated earlier in the day, from on-site PV generation, or stored when it was cheap to import, can be exported to grid at peak times. Energy arbitrage on the wholesale market: charging up when the N2EX price is low and discharging when it is high, or hedging against pre-bought volumes, can give a day-ahead price signal, as in the figure below; playing on the intraday market can similarly offer additional uplift. If the asset owner is exposed to the Balancing Mechanism through their supplier, short notice calls to discharge in line with a large system imbalance offers another revenue stream.

Battery Storage Energy Arbitrage
Fig: Energy arbitrage using the N2EX day-ahead market is shown for a 60kW, 300kWh system over one day. The stored capacity of the battery is varied to minimise the cost of electricity over the day, such that the battery charges when electricity is cheap, and discharges when it is more expensive. By layering on additional price signals closer to time and re-optimising the power profile, additional revenues from energy arbitrage across a range of markets can be achieved.

These requests to charge and discharge in line with ahead-of-time and real-time price signals must be dispatched depending on the import and export constraints of the site and of the capacity of the battery. It needs enough stored energy to be able to discharge for the duration of the price spike, while maintaining the availability to do FFR in accordance with any pre-contracted volumes required by National Grid, and such that recovering the SoC afterwards does not cost more than earnings during the call itself. It also needs to do so without overly increasing the throughput of the battery such that the valuable warranty is violated.

Open Energi has experience of doing this with assets already in operation behind-the-meter. At South Mimms Welcome Break services, we are managing a Tesla Powerpack alongside an EV charging station, to displace site demand from EV charging throughout the day, while providing FFR, and providing energy savings by doing a full discharge of the battery during peak periods.

In a Somerset field, we operate a battery integrated with a solar farm, using PV generation to charge up the battery for free during the day, discharging at peak times, and managing the SoC so that FFR availability is maximised the rest of the time. By stacking these additional revenue streams alongside FFR we increase the income of the asset, and with our Dynamic Demand 2.0 platform, will be able to extend these services across a host of other assets and services.

This includes managing fleets of battery storage systems, where the optimal control strategy goes beyond revenue stacking each asset individually. A central intelligence that is aware of the state of each battery can determine how hard to work different assets to maximise revenue or minimise the number of charge-discharge cycles performed, extending each asset’s life. Dynamic Demand 2.0 can combine revenue stacking with fleet management strategies to meet these objectives.

As we look to the future of energy, it is clear that behind-the-meter battery storage has a huge role to play in creating a more sustainable system. As renewable energy becomes more prevalent, value is emerging in new places, and the key to optimising this value, without negatively impacting battery performance, is technology.

Robyn Lucas is Head of Data Science at Open Energi

Making a success of batteries

Tech image

The surge in interest in battery storage projects has highlighted a fundamental change in the energy market, as commercially viable systems become progressively more available. We explore critical success factors, from choosing the right battery to managing state of charge.

The deployment of physical energy storage assets can broadly be separated into two project categories. The first kind of project consists of grid-scale assets in “front of the meter”, which are usually implemented by industry partners on large grid connections. The second type is “behind the meter” batteries which provide an added layer of flexibility to energy consumption patterns of sites already connected to the electricity network – and offer tremendous potential to unlock previously inaccessible revenue streams for industrial and commercial customers.

Both project types require different approaches to select the best battery type and optimise operational strategy and performance over time.

Selecting the optimal battery operating strategy

Battery flexibility has the ability to unlock several non-mutually exclusive revenue streams. For example, a battery can be used to reduce site demand (for “behind the meter” projects), or export to Grid (for “front of the meter” opportunities) during peak price periods, reducing costs associated with wholesale, Duos, Triads and Capacity Market levy charges. Outside periods of peak tariffs, batteries can participate in the frequency response market and earn a revenue from National Grid for helping to dynamically balance electricity supply and demand.

The characteristics of Battery Energy Storage Systems (BESS) differ widely between manufacturers, with important factors to consider including capital and operating costs, power rating, energy storage capacity, energy density, cell chemistry, operating temperature, round-trip efficiency, self-discharge, degradation profile and tolerance to various depth of discharge. All these parameters have an influence on the economic viability of the project, so it is important to select the appropriate technical solution for a given project.

Once the different parameters are known, the determination of the most economical operating strategy becomes an optimisation problem in response to an aggregated electricity price signal and a potential frequency response revenue, under several constraints such as the battery technical characteristics and the site operational constraints (existing demand/generation on site if any, and import and export capacity).

The operating strategy might change over time, for example because one component of the price signal has changed, or if there is a new opportunity for flexibility that is more financially viable than current revenue streams. In that case the optimisation process will be performed again and the operating strategy modified accordingly.

Battery State of Charge profile
State of charge profile of a BESS doing peak price avoidance from 4PM to 7PM and participating in the frequency response market the rest of the time. The energy stored in the system is maximised before 4PM in order to optimise arbitrage revenues.

Choosing the right battery

The next crucial decision is choosing a battery that is optimal for a given project and operating strategy. The goal here is to select the battery that will be commercially viable under the constraints of a given project. For a “front of the meter” BESS the main factors driving the battery characteristics are the Authorised Supply Capacity (ASC) for importing and exporting, the capital and operational costs and the electricity tariffs for import and export.

There are additional parameters for a “behind the meter” battery. As most of these projects are implemented in sites with no or a small export capacity, the battery would respond to a low frequency event by discharging power into the site, reducing its overall energy consumption. It is therefore crucial to forecast the demand on site to choose the optimal battery size and tender an accurate power availability in the frequency response market.

The same approach can be used for generating sites (like wind or solar farms) where there must be sufficient potential for export in addition to the generating activity on site. The potential energy savings are also dependent on the demand and the site constraints, which might in return drive the optimal power/energy ratio of the BESS.

Managing battery state of charge and maintaining performance

Once installed, the challenge is to manage batteries while ensuring high performance following the operating strategy selected. A requirement of entering the frequency response market is to be able to provide the power tendered for 30 minutes at a time, which highlights the need for a performant state of charge management.

There is an inherent efficiency in BESS, with average efficiency ranging from 75% to 90 % for conventional systems. When used in the frequency response market, successive cycles of charge and discharge will progressively cause a net discharge of the battery, and ultimately cause the battery to be fully discharged if no corrective actions are taken. Similarly, if several large high frequency events happen in close succession, a frequency-responsive BESS might reach a high state of charge at which it will not be able to respond to high frequency events anymore.

Battery charge management graph
State of charge of a 1MW/2MW.h frequency responsive battery. An appropriate state of charge management helps keep the energy stored in the battery at an optimal level over time.

A control strategy should ensure that the battery state of charge always stays within appropriate boundaries in order to meet its contracted obligations at any given point in time. It should also ensure that the total throughput of the battery (which is the cumulative sum of discharge processes over time) is minimised while in operation. A reduced throughput decreases the wear and tear of the battery, enhancing the BESS lifetime.

At Open Energi we are working with several customers to successfully operate batteries in the frequency response market, optimising their operating profile to maximise revenues, applying designed state of charge management techniques, while limiting the degradation of the battery lifetime to the lowest value possible.