Does my battery look big in this?

Published April 2016

Sizing battery systems can cause project developers a real headache. We share our key takeaways for system sizing success in the UK’s upcoming Enhanced Frequency Response tender.

 “Does my battery look big in this?”

This is the question asked by battery project developers around the world, as they strive to ensure that their storage unit squares up with the services required. And yes, battery sizing is tough. Whilst a battery’s power rating is shaped by the available grid connection, defining the required energy capacity is a much more complex techno-economic optimisation problem.

This is the kind of analysis we can’t resist. In this blog, Nick Baldock and I share our key takeaways, taking the UK’s new Enhanced Frequency Response (EFR) revenue stream as a case study.

1. Being too big prices you out of the market

From crunching the numbers, it’s clear that there are risks associated both with being too small, and being too big. Firstly, being too big can prove an expensive mistake: installing a system with substantial redundancy pushes up capex. At best, this gold-plating means squeezed margins; at worst, it means that you fail to even secure a contract in a competitive tender.

This can be understood at a high-level through simple analysis of the data. National Grid’s EFR guidance states that sufficient capacity for 45 mins worth of storage is required. But analysing grid frequency data reveals that for 95% of the time, a storage unit would only need to provide EFR below the frequency deadband for 114 seconds or less, per event; see the chart below.  This is true also for support required above the frequency deadband.  It raises the question of whether it is really optimal to size your battery for the rare tail end events – which can only be answered through more in-depth temporal analysis.

For the vast majority of events below the frequency deadband, your battery only needs to deliver for up to 2 mins at a time.

2. Being too small risks technical non-performance

But on the flip side, being too small brings the obvious technical risk that the battery unit is not sufficiently available to deliver the services required. Under EFR, National Grid imposes steep penalties for failing to meet National Grid’s 95% performance availability requirement. Fail to hit this by even 1%, and you suffer a payment reduction of 25% in that settlement period. If financiers sniff out this risk, securing project finance could be an uphill battle.

So, here are our three takeaways for approaching this techno-economic optimisation problem:

1. Fully analyse past frequency data: In our chart above, we explore some basic probabilities, but this is just a starting point. Most importantly, there is a need to model in the time-domain, to properly capture the proximity of high and low frequency events. This will help you determine how much you charge the battery between events, since you won’t want to be fully charged if called upon to reduce the grid frequency.

2. Stress test against future frequency pattern changes: Future frequency deviations over the contractual 4-yr period may differ from historic frequency patterns – and your business model should be resilient to this possibility.

3. And most importantly – get to grips with the technical spec of your battery: This blog post has focused on service requirements – but the battery itself is also key! For instance, you need to understand the minimum and maximum operational state of charge. And depending on the technology, you might consider oversizing your battery to provide some lifetime degradation redundancy.

There you have it! Reaching a definitive answer to the question “does my battery look big in this?” requires some very interesting analysis, and we’re happy to help.

Luckily, in personal circumstances outside EFR, the answer is much simpler: always “No”. Just trust us on that one…

Note: Since EFR is a pilot scheme, National Grid is continually updating its guidance material on how the scheme will operate. This blog post is based on material published in March 2016.