Assumptions behind intraday optimisation

There are a host of ways to model how a battery might trade in intraday markets and these influence the revenue outcome. The assumptions we use are below.

• The battery's capacity can be optimised in the Day-Ahead markets. Any spare capacity (both in MW and MWh) can then be traded intraday.
• The volume of ancillary service contracts a battery can win in future is limited in line with today's saturation levels. Ie a 100MW battery cannot win 100MW of Dynamic Containment or Dynamic Regulation.
• Battery operators can accurately forecast intraday prices two hours ahead of delivery, and there is sufficient liquidity in the market to buy or sell the desired volume.
• We require a £5/MWh spread in intraday prices for a buy and sell order.

Why did we decide to optimise like this?

We examined various ways to optimise in intraday markets. Such as:

• Varying the amount of capacity available to trade in intraday markets - like reserving MW or MWh specifically for intraday trading.
• Varying the strike price at which you can trade.
• Using different-sized rolling windows.
• Permitting different levels of ancillary service contracts.
• Varying the foresight of prices - eg 12h, 8h, 6h, 4h (and settled on 2h).

Different combinations of these led to different intraday uplifts. For example, a longer window of price foresight increased revenues. Reserving 50% of the MWh of a battery specifically for intraday trading increases the revenues from intraday trading but reduces the revenues in wholesale trading, and the overall revenue doesn't change much at all. Allowing the battery to participate in more ancillary services means it has less capacity for trading, so the intraday (and day ahead wholesale) revenues are smaller.

To decide on the 'best' parameters for these, we ask questions like:

By which point are the majority of trades placed in the intraday market? How realistic is it to trade a battery volume (and be close to buying or selling at the RPD price) 8 hours out from delivery? Historically, how much volume have batteries placed in dynamic frequency response services on average? Where would battery operators place most volume: risk leaving it to a potentially illiquid continuous intraday market close to real-time, or secure most revenues Day-Ahead?

Ultimately, in setting the parameters of the optimisation, we were informed by data and our expectations of real-world behaviours.