How should TSOs address the growing problem of imbalance oscillations?

Oscillations are back in the headlines, with the latest update from Entso-e on the Iberian blackouts (https://lnkd.in/eBK8xPsp) providing more data on the frequency and power oscillations.

But another type of oscillation has been causing concerns to TSOs in recent weeks. Although not mentioned in the Iberian context, both Tennet and Elia have published reports, while decisions from Energinet and RTE also reflect the emerging problem.

Imbalances are deviations from scheduled consumption and production from energy markets, increasingly requiring TSOs to activate balancing reserves due to forecast errors and less due to forced outages.

Imbalance oscillations are when system operators activate both upwards and downwards reserves within the same settlement period.

This oscillating activation of explicit flexibility from Balancing Service Providers is not only the result of renewable generation, but increasingly the unintended consequence from the move to a decentralised balancing model.

The role of a Balance Responsible Party used to be exactly that - to ensure that their generation, consumption and market trades are balanced. For each settlement interval, each BRP's imbalance volume is settled at the relevant imbalance price.

Dual imbalance prices incentivised BRPs to maintain their own balance by penalising both positive (excess supply) and negative (excess demand) imbalances.

With Imbalance Settlement Harmonisation, the EU’s target model is for imbalances to be settled at a single price, regardless of whether they result from excess supply or excess demand.

In theory, this incentivises BRPs to take actions that support the balance between electricity supply and demand. TSOs publish imbalance forecasts, and activations of balancing energy close to real time, to signal their requirements.

In practice, more batteries means more BRPs can react more quickly to real-time imbalance than TSOs can instruct BSPs, overshooting the system need.

"Faster BRP responses are therefore unwanted as the aFRR product has its limits and therefor BSP cannot respond faster."

ACER's 2024 assessment found that 20 member states have implemented single imbalance pricing, and 5 have dual pricing, just as the single price has become more of an incentive to respond to power imbalances rather than to support the energy balance.

Tennet's approach in the Netherlands is aligned with the target model when activations are in one direction: there is a single imbalance price to reward supportive imbalances - Regulation State 1 (or -1).

But when both upwards and downwards ancillary services are activated, dual pricing is applied - Regulation State 2.

The balance delta for each minute is published, delayed by 2 minutes, so each BRP must take a view on the system imbalance before the settlement period ends.

"A delay of a few minutes in the processing and presentation of the real-time information from different systems is unavoidable."

Due to asset responses overshooting the system need, the frequency of Regulation state 2 is increasing, as is the magnitude of imbalance oscillations, reaching 2GW in December 2024.

This would not be an issue under a polluter pays principle, the driving rationale behind the move to a single imbalance price. But when assets can not only respond to a balance delta in one direction, but also respond to an overshoot by bringing their own position back into balance within the settlement period, they can avoid the punitive Regulation State 2.

Although not explicitly specified, the increase in imbalance oscillations has coincided with the rise in grid-scale batteries, capable of responding much faster than the assets which traditionally provided flexibility.

The response from Tennet included increasing the delay in publishing the balance delta to 5 minutes, meaning that assets and their optimisers would face a greater risk of incorrectly forecasting the regulation state. Needless to say:

"Delay of the balance delta publication did not have the desired effect."

On 1 July 2025, the delay will return to two minutes. 

Elia maintains the philosophy that assets which cannot participate in ancillary service markets can contribute implicitly to grid balancing by reacting to the Imbalance Price signal, a:

"robust and reliable price signal in real-time, supporting an effective decentralized balancing model and allowing assets which cannot participate explicitly to valorize their flexibility implicitly"

Elia believes that explicit participation should deliver more value to assets like grid scale batteries. And yet we have seen the markets that deliver most value become saturated, and like locusts on search of a new feast, the batteries move on.

Aiming to publish as close to real time as possible, the consultation on the Imbalance Price formula describes the objective of discouraging oscillations, with the price reflecting the average conditions over each 15 minute interval.

This is complicated by the European market coupling platforms, with the potential of mFRR activations from the MARI platform contributing to the price, yet unnecessary for local conditions.

Although a deadband may address small activations affecting the price most of the time, it creates a step change in the price when activations cross the line.

Although most European markets having completed or in the process of moving to 15 minute intervals, particularly for intraday trading, in a dynamic market it may still be challenging to define "average conditions".

However, as we have seen from the additional complexity and pressures on systems causing a delay to 15 minute SDAC, ever decreasing intervals bring their own challenges. And yet, in Australia the NEM introduced 5 minute settlement in 2021.

So what's the solution?

With more grid-scale batteries coming on to the system, increasingly supplemented by smart charging of EVs, and optimised heat pumps, the capacity of assets able to respond more quickly than traditional flexibility is rapidly exceeding the procurement of explicit flexibility.

The deeper, more liquid, wholesale markets are becoming a significant part of the value stack, with passive balancing potentially even more lucrative.

Increasing the risk of passive balancing by delaying the publication of data proved not only to be a retrograde step, but also didn't achieve the desired outcome.

Publishing more data close to real-time, reducing the interval durations, activation deadbands and even returning to a dual imbalance price have all been considered.

Will this solve the growing challenge of imbalance oscillations?