Manual vs Automated Trading

In essence, optimising flexible assets in traded energy markets means trying to maximise (or minimise) the captured price for whatever energy can be sold (or bought) by the device in question: a gas power station, battery storage or just a single electric vehicle. In practice, it is often a highly complex exercise requiring processing and a combination of information from two distinct sources: the asset characteristics and market intel.

This cross-optimisation involves continuous calculation over different time horizons, as market opportunities (like Day Ahead auctions) and asset limits (such as ramp rates restrictions) must be planned against. A classic example would be a trader managing a gas power plant; assessing changing market conditions and working closely with plant operators to understand variable physical parameters, such as efficiency at different power outputs or the energy required to start up the generator.

This information must then be considered when selling energy into the power markets, and that involves a careful trade-off between physical and price considerations. For example, when low demand causes low prices overnight, plants must choose between shutting down in the evening then starting again in the morning or running through the night, selling power at a loss – whichever is the more economical.

 

Adapting to new technologies

However as storage, demand response and hydrogen production become the dominant tools for balancing the grid, this is also driving change in the methods for optimising assets. Novel technologies are being deployed at the megawatt or kilowatt level, not gigawatt, meaning many more assets will be involved in making up the required level of balancing capacity.

Each device has its own features and characteristics, such as power capacity or response time, as well as dynamically changing parameters impacting optimisation, eg state of charge (SoC) or energy recovery period. So it is clear when dealing with thousands of individual assets the complexity of the problem scales greatly.

Even for larger Front-of-the-Meter batteries, warranties are becoming increasingly complex to manage in real time; specifying rest periods or dynamically limiting depth of discharge, as opposed to simply warrantying a certain number of cycles per year. Also, many projects will be co-located with variable renewable power to exploit the benefits of a shared connection, effectively giving a dynamically varying export connection to factor into optimisation.

 

Pros and cons of automation

In this world, automation is fast becoming essential. Allowing the principles of continuous asset optimisation can be applied at a scale far below what would be economical for a human trader.

Open Energi has been trading fully algorithmically in the Day Ahead markets for over a year and has some key learnings.

On the face of it picking the highest and lowest hour in the day is simple (and fairly predictable). However, to maximise revenue you must also respond to more real time signals which occur within day (eg Triad), which alters your SoC from the planned schedule. This creates a problem when submitting bids in advance for the trading day before the current day has ended, as you do not yet know what SoC the battery will be at at the start of the trading day.

Automated fixes are able to easily correct for misalignments to get round this; however, doing so in the most economical fashion is harder. And the problem is exacerbated when dealing with more dynamically changing markets like the intraday continuous, which require thinking between different trading horizons.

 

Manual plus automated trading – a winning team

Overall, traders still have the upper hand on algorithms in areas like price formation, especially during extreme events like 4th March, and the best solutions will be ones that exploit the strengths of both. This is the principle of our solution Dynamic Demand 2.0 Trader, where full automation capabilities will perform the heavy lifting but oversight from Erova Energy’s 24hr trading desk provides the manual oversight and possibly intervention if greater opportunity is identified.

Ultimately, combining the strengths of power traders and algorithms provides the best optimisation. And ensuring each is capable of running independently provides the built-in resilience that proves its worth as COVID-19 pushes systems to their limits.

 

For a free consultation about automated trading, call +44 (0)20 3051 0600

Batteries in the Balancing Mechanism

The Balancing Mechanism (BM) is the primary flexibility market in the UK. In 2019 over 2TWh of flexibility was procured through the BM with a value worth over £800m. Batteries are only a recent (and small) participant – the vast majority of flexibility is provided by CCGTs and some through pumped storage such as Dinorwig.

Batteries have had over a year in this market and have steadily seen increases in activity, helped by the introduction of the Distributed Resources Desk, while upcoming Project TERRE could also help non-traditional providers receive dispatches. Hence, while batteries remain a niche player in the BM, compared to the dominant technologies of CCGT and pumped storage, there has been a steady increase in activity.

Figure 1 – Activity by batteries in the Balancing Mechanism has been increasing over the last year

Batteries in the BM – The Basics

The Balancing Mechanism is manually dispatched by the ESO Control Room – providers submit prices and volumes but only deliver (and are paid) when selected. Dispatch decisions are made based on a number of operational criteria, of which volume and price are just two. For example, only certain technologies are able to meet certain needs: thermal plant provide inertia but batteries don’t.

However, battery storage does have its own unique benefits as batteries can respond extremely quickly and accurately in either direction. This quality is being exploited by the Control Room with batteries delivering short bursts of power of mostly under 10 minutes duration (see fig 2). Traditional thermal providers cannot do this, given their ramp rate restrictions.

The most obvious difficulty batteries bring is that they are duration limited, whereas a gas power station could increase its power indefinitely. This means, once a battery has discharged completely, it cannot sell any more energy and so must recharge, either through trading or by waiting to be dispatched in the other direction. Batteries can be dispatched in either direction throughout the day – even if the system is long, batteries may be offered up, and vice versa, so leaving the battery empty (or full) would result in missed dispatches and lost revenue.

Figure 2 – A day of BM actions for a battery, light area showing availability and dark bands are dispatches. Most dispatches are under 10 minutes in duration.

 

State of charge – The limiting factor?

State of charge (SoC) is, therefore, a massively important consideration for both operators and National Grid alike. However, unlike the physical restrictions of thermal plant (such as minimum output), SoC is not captured in the BM, given it is a novel issue. We can infer when SoC has drifted significantly, though, as batteries adjust their available power (MEL and MIL) to represent 15 minutes of storage. This means that when less than 15 minutes output is available in one direction, the system can only be dispatched at this reduced level.

For a two-hour system this only has a small impact – state of charge can drift significantly in either direction before this limit is hit. However, for a one-hour system the impact is much more significant, as the battery could potentially be offering up reduced availability 50% of the time.

Overall, taking a much more active role in managing SoC is necessary to maximise benefit, especially for more limited duration batteries.

 

Being Active

Market optimisation of batteries within the BM takes two forms: integration with other trading strategies, and through much more dynamic provision of bid and offer price. Both of these offer solutions to more actively managing SoC to reduce time spent offering reduced availability.

Trading

The most obvious route to managing SoC during the course of the day is through the intraday markets. If one or more offers in a row start to deplete SoC, energy can be bought on the intraday market to recharge the system. However, doing so may not always be the optimal solution – eg if the price is too high, or perhaps it is likely a bid will arrive soon anyway.

An advanced optimisation and forecasting solution combining manual and automated inputs is needed to effectively manage SoC through trading; system warranty is a constant consideration and confidence will be needed that any actions will increase profitability later in time.

For the example below, purchasing just 30 minutes of energy at the time shown would increase total daily returns by 13%.

Figure 3  – A day of BM actions for a battery, with energy purchased through intraday markets around 9am (green), and corresponding increase in availability and dispatches shown in grey

 

Bid/offers

Although operators have no control over whether their assets are dispatched in the BM, they can influence the likelihood of being dispatched in either direction by adjusting their posted prices, or by providing stepped bids and offers.

Increasing bid price in response to low SoC could be provided to increase the chance of dispatch, in order to then capture higher revenues across the whole day. Meanwhile, stepped bids and offers provide the Control Room with two or more prices, which can be paid to access different levels of power output.

However, this route still has a limitation in being dependent upon being dispatched, even if the probability of being so can be influenced.

 

Conclusion

The BM has long been talked about as the holy grail for battery operation but there is still a lot of uncertainty over when (and indeed whether) that point will be reached as system balancing transitions from CCGTs (with infinite duration headroom and footroom) to fixed duration energy storage.

However, recent activity has provided good signs for batteries to be the principle candidate to take over from CCGTs as the UK moves towards net zero, and we expect to see further design aspects of the BM to be updated in favour of storage assets, to enable NG ESO to meet its target of zero carbon operation by 2025.

Meanwhile, value from the BM continues to increase and challenge frequency response revenues – Open Energi and Erova Energy will be launching our Balancing Mechanism offering in the coming months, so watch this space!

 

For a free consultation about trading in the Balancing Mechanism, call +44 (0)20 3051 0600

Solar plus storage – The benefits of co-location

The UK will need 30GW of storage to meet our climate goals, according to Imperial College. That’s 10 times the current capacity. There are many questions this raises, such as ‘what is the appropriate technology mix?’, but if we consider the form predominantly being developed now (lithium-ion batteries of under or around two hours duration), one of the main questions is where best to situate this?

So far the vast majority of volume has been deployed as front-of-the-meter installations – a standalone site with a dedicated connection for the battery alone. However, given that connections can be costly or administratively difficult to obtain, a more cost-effective solution would be co-location.

In this blog we will dig into some of the advantages of co-location of batteries with solar farms (PV installations with a dedicated connection). We will leave the current policy/regulatory framework to the side and focus on the fundamental benefits.

 

The case for co-location

When new generation connects to the grid, it has to pay for the new copper in the ground. This isn’t cheap. For a battery (short duration, lithium-ion) the connection typically represents 10-20% of the total capital expenditure – and can be much higher, depending on the conditions of the local network (eg if you are unlucky enough to be the marginal connection requiring a sub-station to be upgraded).

If the battery was able to share an existing connection with a solar farm of the same size, it could save the 10-20% of upfront cost; it would still need to buy an import connection, which the solar farm doesn’t need, to ensure the battery can draw from the grid to charge, but the economy would be significant.

In the UK, solar PV (without tracking) has an average capacity factor of 10-15%. This means that for the vast majority of the time the connection is an expensive sunk cost, lying there unused. By sharing it with storage, the return on investment from the connection increases several times over.

 

Why solar and storage are perfect bedfellows

Sharing one connection becomes a problem when both energy sources are producing at the same time. Export from one resource will constrain the ability of the other to export. With solar and battery storage, we find that this is rarely the case; ie both assets are very seldom trying to use the connection to export at the same time [see fig 1].

This is due to the price cannibalisation of renewables – where prices are lowered at times of high renewable generation because they produce at zero marginal cost, and so will sell power at any price above zero. There is sufficient solar power in the UK (>12GW) that during the middle of the day market prices will drop lower.

If we are performing price arbitrage with the battery, it is obvious that we would not be looking to discharge at these times when prices are low. We want the highest prices – usually over the evening peak. In fact, we have found that when optimising the battery against Day Ahead prices, a conflict occurs with the solar generation less than 1% of the time.

Figure 1: Generally, the battery looks to discharge at times of high prices which does not align with solar output

What this means is that we can effectively trade the two assets separately from each other, rather than as one. This has great benefits because solar predictions are best made at portfolio level, taking advantage of geographical dispersion to neutralise stochastic cloud impacts on production. Therefore, the solar can be traded as a portfolio, while battery activity can be scheduled separately.

This also highlights one major advantage of pairing batteries with solar rather than wind. Solar has very high diurnal periodicity, so we can guarantee there would be no production over the most lucrative peak time (4-7pm, especially in winter). The same cannot be said for wind; so if it a wind farm was generating during this peak period there could be large entailed opportunity cost for the battery.

Given that the battery is the more controllable resource, and solar produces at zero cost, the battery should accommodate the solar generation and plan activity around it.

 

The need for advanced optimisation

Pairing solar with batteries, then, looks like an obvious and straightforward win. However, there is more nuance.

While Day Ahead prices will rarely incentivise discharging the battery in the middle of the day, there are often later market opportunities that do; for example, if a large generator trips off the network to create a shortage. The most advanced battery trading strategies will cycle far more than once per day by looking to optimise at each time horizon, to stack value across the day.

Given that cloud cover could have a strong impact on PV output, there will still be many instances where a solar farm has the connection unutilised in the middle of the day [fig 2].

Figure 2 – Clouds drive large short term volatility in production

To take advantage of this we need reliable, real time monitoring of the solar farm (we can only dispatch the battery and make use of the connection if we can be certain the solar PV isn’t) and dynamic responsive controls to ensure solar output is tracked closely (to prevent overloading the connection point).

Dynamic Demand 2.0 Trader is powered by our fully automated platform, needed to support this level of sophistication. Control logic is held both locally on the battery, to ensure responsiveness to changing site conditions, and centrally in the cloud, enabling responsiveness to live price opportunities provided by Erova Energy’s advanced market insight.

With this level of automated control and trading insight, pairing solar with storage represents a cost-effective solution for storage location and a potentially lucrative way forward for energy investment.

 

For a free consultation about co-location and your energy investment, call +44 (0) 20 3051 0600

EDIE: UK’s largest Tesla battery to be AI-optimised

According to Bloomberg New Energy Finance (Bloomberg NEF), the global lithium-ion battery market for energy storage will reach at least $239bn by 2040, as renewables come online and as technology costs fall. The early signs of this battery boom can be seen in the UK’s utilities sector, where one in three firms have already invested in onsite storage.

This week Zenobe Energy announced that it will use artificial intelligence (AI) to optimise its front-of-the-meter battery at Hill Farm, Leicestershire. The move will enable the 12MW Tesla unit to better take part in the utility markets of the future and will also extend its lifespan by tracking and managing the battery’s state-of-charge. Additionally, Zenobe Energy claims, it will make the facility more resilient and flexible – enabling it to play a role in maintaining energy security.

Open Energi has developed the AI platform, called Dynamic Demand 2.0. Other firms utilising this technology include United Utilities and Aggregate Industries.

“As our energy market continues to be revolutionised by the growth of renewable sources, optimisation strategies in battery storage systems has never been more complex or critical,”

“We designed our algorithmic trading capability with companies like Zenobe in mind, who recognise the importance of innovation and optimisation to meet the needs of power-intensive businesses and the UK at large.” Open Energi’s director David Hill said.

Read the full article.

Current± News: Open Energi partners Zenobe for AI-enabled battery optimisation

Current News

Open Energi is partnering UK battery storage firm Zenobe Energy for the automated trading and optimisation of the latter’s grid scale Hill Farm battery.

Open Energi’s Dynamic Demand 2.0 energy optimisation platform has been connected to Zenobe’s 12MWh battery at the Hill Farm site, enabling further optimisation of the battery across a wider range of ancillary services and trading markets.

The platform uses AI to increase responsiveness and track key indicators while reducing throughput and degradation, therefore extending battery lifetime, Open Energi said.

Read the full article.

Power Engineering International: Arsenal football club install energy storage system

Power Engineering International

The first energy storage system to be installed at a UK football stadium has been unveiled.

The 3 MW battery system at the ground of Premier League club Arsenal in north London will be able to run the 60,000-seat stadium for the entire 90 minutes of a match. A further 1MW of storage to be added next summer.

Arsenal managing director Vinai Venkatesham said: “This is a big step forwards for us in being efficient with energy usage and it builds on our work in reducing our carbon footprint as an organization.”

The system is also intended to operate as support for the wider UK grid. It be automatically traded and optimized by Open Energi in response to market signals and has already secured a frequency response contract from National Grid.

Read the full article here.

Current News: Arsenal makes UK first with behind the meter battery for wholesale trading

Current News

Arsenal football club’s Emirates Stadium has become home to what could be the first behind the meter battery of its size to be aimed at wholesale energy trading over frequency response.

The 2MW/2.5MWh Tesla system, unveiled yesterday, is the first battery storage system to be installed at a UK football club’s stadium following three years of development between the club, Pivot Power and Downing LLP.

The installation has a firm frequency response (FFR) contract via Open Energi, which will use the battery to meet the needs of a 2MW contract won in September’s tender for delivery from 1 October 2019 for six months, ending 31 March 2020.

Read the full article here.

Utility Week: Anglian Water partners for solar and energy storage project

Anglian Water has partnered with redT and Open Energi to have energy storage facilities installed alongside solar panels at one of its water treatment works.

The water company has purchased a 60kW/300kWh redT energy storage machine to install alongside a 450kWp solar PV system. This will enable it to store excess solar generated during the day and use it at other times, to reduce the site’s reliance on the grid.

As the largest power consumer in the East of England, reducing reliance on “volatile grid electricity” will help optimise a £77 million energy bill, which is one of the company’s “most significant” operational costs.

Read the full article.

edie: Anglian Water to boost onsite generation with AI-powered energy storage technology

Water utility Anglian Water is set to install an energy storage machine controlled by Artificial Intelligence (AI) technology at one of its water treatment facilities, in a move it claims will increase the site’s solar generation by 80%.

The 60kW/300kWh storage device, designed by energy storage firm redT, will be set up at the company’s ‘pathfinder’ site in Norfolk to bolster the performance of its existing photovoltaic (PV) array from 248kWp to 450kWp.

The machine, which can store enough energy to power the facility for at least five hours, will enable Anglian Water to store surplus power generated by the array for use within its own operations. Meanwhile, it will use AI software to provide real-time balancing and energy flexibility services. The machine is expected to have a lifespan of 25 years.

Called Dynamic Demand 2.0 and designed by Open Energi, the AI software will optimise the site’s energy consumption and stack multiple demand-side value streams, enabling Anglian Water to take advantage of wholesale energy price arbitrage. In total, the installation is expected to halve the site’s electricity bills by 2050.

Read the full article.

The Energyst: Anglian Water takes 300kWh of RedT’s flow storage, plans 30MW of solar

Anglian Water is to buy flow storage units from RedT to co-locate with solar PV at a treatment works. The water firm aims to work out the potential of longer-duration storage in maximising use of solar power.

The deal is for four of RedT’s flow machines, totalling 60kW/300kWh. These will sit alongside 450kW of PV at a ‘pathfinder’ site in Norfolk.

While the main benefit of these kind of installations is to reduce power bills by being able to store and use solar instead of drawing from the grid at peak times, the technology also enables upside revenues from grid services, including frequency response, as well as arbitrage.

The firms will work with aggregator Open Energi to optimise consumption and stack revenues.

Read the full article here.