As more renewables come onto the system, grid frequency becomes more volatile. One way to manage this is through frequency response services - which are usually provided by battery energy storage. So, as frequency patterns change, how does this affect those batteries that are helping to stabilize the grid?
Frequency response services
The frequency of the grid represents the balance between generation and demand. When both are perfectly balanced, system frequency is at 50 Hz.
When generation and demand aren’t balanced - which is essentially all of the time - frequency moves away from 50 Hz. Frequency increases when generation outweighs demand, and vice versa.
National Grid ESO aims to keep frequency between 49.8 and 50.2 Hz - these are its “operational limits”. (There is also a statutory requirement to ensure frequency remains between 49.5 and 50.5 Hz.)
- To manage this, National Grid ESO relies on grid services. In this article, we’ll be focusing on frequency response services - which are usually provided by battery energy storage systems.
- If frequency is less than 0.015 Hz away from 50 Hz, no action is required - frequency is within the “deadband”.
- Frequency response services kick in when frequency shifts more than 0.015 Hz away from 50 Hz, in either direction - i.e. when it moves outside of the “deadband”.
Frequency is becoming more volatile
So far in 2023, frequency is spending less time in and around the deadband, and more time further away from 50 Hz. This means frequency has become more volatile - and more frequency response is required.
This is mostly because there is more renewable generation on the grid than ever before.
- More renewables have reduced the inertia of the grid (read about this here), which means that frequency shifts more quickly when there are imbalances in generation and demand.
- It also introduces additional volatility in frequency, as wind gusts and moving cloud cover can cause short-term changes in generation output.
In the past couple of years, there has been an increase in frequency variance - i.e. the average distance of frequency away from 50 Hz. On average, frequency is now further from 50 Hz than ever before - in both directions.
So, what does this mean for battery energy storage systems providing frequency response?
Cycling rates from frequency response are increasing
With this increase in frequency variability, the energy throughput required to provide frequency response services is also increasing. This means that batteries have to cycle more to deliver each service.
On average, the energy throughput requirements for delivering frequency response services have increased by more than 3% in 2023.
Delivery throughput here refers to the number of cycles required to deliver each service, for a one-hour battery contracted at 100% available power.
But... these cycling rates vary massively
Frequency is not always predictable. On any given day, frequency can be more or less volatile, or move significantly in one direction. This can have huge effects on the cycling requirements of frequency response services.
- Dynamic Regulation has the biggest absolute variation in cycling - but Dynamic Moderation has the most significant variation in percentage terms.
- This is caused by the increase in output once frequency moves more than 0.1 Hz away from 50 Hz.
- On those days when frequency strays furthest from 50 Hz, Dynamic Moderation requires almost triple the energy throughput - and therefore cycles - that batteries would normally need to provide in the service.
Daily cycling rates are driven by the variance in frequency across the day. The further away from 50 Hz that frequency gets, the more batteries have to cycle in frequency response services.
In general, cycling rates in high- and low-frequency services are inversely proportional. The highest cycling days in one direction tend to be the lowest cycling days in the other. This can have huge impacts on state of charge management.
Frequency response case study: 1st April 2023
On 1st April 2023, frequency remained below 50 Hz for most of the day. The average frequency across the day was 49.96 Hz - by far the lowest frequency day since before 2020! Despite this, frequency mostly stayed within the operational bands of 0.2 Hz.
This low frequency had big consequences for providers of frequency response.
- There wasn’t much impact on Dynamic Containment (due to its lower energy throughput requirements).
- However, Dynamic Moderation would have required more than one full discharge cycle (from a one-hour battery contracted at 100% available power).
- Firm Frequency Response would have needed nearly two full discharge cycles.
- And Dynamic Regulation? 4.8 discharge cycles, to just one charge cycle.
How did this affect Dynamic Regulation providers?
On the 1st April, the gap between the required energy needs of the high- and low-frequency Dynamic Regulation services meant that any systems contracted to provide them had to make heavy use of the wholesale market - to maintain enough state of charge to fulfill their obligations.
This put a big dent in the revenues that Dynamic Regulation providers were able to earn on that day. In some cases, the costs of managing state of charge (by buying energy on the wholesale market) completely wiped out revenues from frequency response.
Kemsley, a 50 MW battery, contracted an average 56% of its power into low-frequency Dynamic Regulation on the 1st April. To fulfill its obligations, the asset had to repeatedly charge in the wholesale market.
What have we found out?
- Cycling requirements for frequency response services are increasing (on average). With more renewable generation coming onto the system, and lower inertia, this is likely to continue.
- There’s also more volatility in terms of how many cycles batteries need to do. Managing this can be expensive, and can severely disrupt scheduled operations.
- The volatility in cycling demands for Dynamic Regulation means that providing the service on certain days can breach an asset’s cycling constraints - if not properly managed.