The TERRE replacement reserve platform provided financial benefits averaging €23 million per month across participating countries in 2023, rising to €38 million per month in 2024, before the platform was phased out entirely on 1 January 2026 as the European balancing architecture consolidated around PICASSO and MARI. That financial trajectory, and the subsequent withdrawal of a platform delivering measurable value, illustrates the tension at the heart of European balancing market integration: the efficiency gains from cross-border exchange are documented and growing, while the structural fragmentation between national markets persists and resists harmonisation at nearly every layer.
European grid balancing operates through three primary reserve layers, each responding to frequency deviations on a different timescale. FCR, the frequency containment reserve, activates within 30 seconds to arrest deviations and is treated as a symmetrical power product with no separate energy settlement. Automatic frequency restoration reserve responds next, with aFRR units required to ramp fully within five minutes as of December 2024, down from 7.5 minutes previously. Manual frequency restoration reserve, activated by explicit TSO instruction, operates over quarter-hour intervals. Most national systems supplement these with a slower replacement reserve, though TERRE’s retirement consolidates that function back toward the core three-product framework.
FCR: The Most Integrated Layer, With Caveats
FCR is the most harmonised balancing product in Europe, procured through a joint auction involving nine countries that selects the lowest-cost bids across the region while maintaining mandatory minimum volumes within each participating country. The logic is straightforward: frequency is a system-wide property of the synchronous area, so the geographic origin of a rapid primary response is irrelevant to its stabilising effect. Cross-border procurement therefore, extracts genuine cost efficiency without compromising technical performance.
The Belgian FCR market illustrates both the mechanism and its evolution. In 2023 and 2024, Belgium relied heavily on foreign FCR bids, procuring domestically only around the mandatory minimum. From 2025 onward, domestic Belgian FCR provision has grown as local providers, predominantly battery storage, have become more price-competitive. Prequalified FCR capacity in Germany dropped 35% in 2024 to 4.5 GW, reflecting the exit of some conventional assets and a shift in who is providing the reserve rather than a reduction in the volume being procured. Average FCR prices in 2024 hovered slightly above €50 per MW, with Belgium and the Czech Republic recording higher prices and greater volatility, while France maintained traditionally lower prices attributable to its flexible nuclear fleet. The absence of market saturation effects despite growing battery storage prequalification suggests that the FCR market is absorbing new entrants without collapsing prices, which in turn reflects genuine demand growth driven by the variability introduced by renewable generation scaling.
aFRR and the PICASSO Expansion
The automatic frequency restoration layer is where the integration process is most actively evolving, and where the renewable penetration effect on pricing is most visible. PICASSO, the cross-border aFRR energy exchange platform, has been operational since 2022 and has undergone a substantial accession wave. In 2024, Energinet, TenneT NL, Elia, and SEPS joined PICASSO between October and November, and Italy’s Terna reconnected in November 2025 under a revised balancing framework after suspending participation in March 2024 following extreme early price episodes. The platform now covers most of the EU synchronous area, with the broadened participation adding liquidity and deepening the cross-border merit order.
The price dynamics within aFRR capacity markets are increasingly shaped by renewable generation patterns, but unevenly so across countries. In Germany, aFRR capacity prices drop toward zero during nighttime hours when reserve availability is ample, rise for upward capacity in the evening as solar output fades and thermal assets shift their opportunity cost calculus, and rise for downward capacity during daylight hours when solar generation creates surplus that the grid must absorb. These price patterns accurately reflect the underlying opportunity costs of providers adjusting bids based on anticipated wholesale prices and renewable output.
In Spain, the dynamic operates differently. Downward aFRR capacity prices tend to be very low at midday in June because renewables, particularly solar, contribute directly to the available balancing capacity. The Spanish market has integrated renewables into its balancing supply stack more extensively than Germany, where solar participation in balancing capacity remains limited. This contrast reflects a structural difference in market design and prequalification rules rather than simply a difference in installed capacity, and it has meaningful implications for how balancing costs evolve as solar penetration increases across Europe. Countries that fail to incorporate renewable assets into balancing supply will face rising reserve procurement costs as the generation mix shifts, while those that enable renewable balancing participation can offset that upward cost pressure.
A notable trend confirmed through 2024 is the growing importance of downward regulation, with prices for negative aFRR reaching magnitudes comparable to those for positive aFRR, and a similar trend appearing in the mFRR market. This symmetrisation of upward and downward reserve value reflects the increasing frequency with which oversupply conditions, driven by solar and wind output, require the grid to reduce net injection rather than increase it. It is a direct market signal of the balancing challenge that renewable scaling creates.
Divergent National Philosophies in FRR Procurement
While aFRR energy exchange through PICASSO is progressively harmonised, the capacity procurement layer for both aFRR and mFRR remains national, with significant differences in auction design, product duration, settlement mechanism, and bid complexity rules. These differences are not arbitrary; they reflect genuinely different balancing philosophies that have developed alongside different generation mixes and regulatory traditions.
France operates through RTE, its single TSO, using a centralised and proactive balancing model reinforced by strict dual-pricing imbalance settlement that penalises market participants for deviations. This design incentivises early portfolio balancing by commercial participants and leaves RTE relying heavily on mFRR and even replacement reserve for real-time corrections. Germany’s four regional TSOs operate a decentralised, reactive model with intraday gate closures pushed close to real time, allowing continuous commercial trading to resolve most discrepancies and leaving the TSOs to activate only residual imbalances through aFRR. The result is that Germany uses aFRR as its primary active balancing tool with very limited mFRR activation, while France does the inverse.
Neither philosophy is inherently superior; each is calibrated to the market structure and generation mix it operates within. But the divergence creates a material challenge for cross-border integration at the capacity procurement layer. MARI standardises key aspects of the mFRR product, including a 15-minute resolution, harmonised gate closure times, and a pay-as-cleared pricing mechanism, but cross-border mFRR energy exchanges under MARI remain geographically concentrated. Cross-border mFRR exchanges in 2025 were largely confined to the Iberian Peninsula and the Baltics, limiting the cost efficiency gains that broader liquidity would provide.
The Architecture of Cross-Border Efficiency
The layered structure of European balancing integration, with FCR most harmonised, aFRR energy exchange through PICASSO growing rapidly, and mFRR energy exchange through MARI lagging behind, reflects both the technical logic of the reserve hierarchy and the political economy of national market reform. FCR was easiest to integrate because the volumes are relatively small, the product is technically simple, and the geographic indifference argument is unambiguous. aFRR energy exchange is more complex but technically tractable; PICASSO’s pay-as-cleared mechanism and four-second price update cycle create a functional cross-border merit order that can allocate balancing energy to its lowest-cost provider across borders in near real time.
The Netherlands provides a useful illustration of how national market design interacts with this framework. TenneT NL procures aFRR capacity across two time horizons, a daily auction and four-hour blocks, and the volume of free energy bids available in the Netherlands’ aFRR market significantly exceeds contracted capacity. This surplus bid availability reflects a deep pool of flexible assets willing to offer balancing energy beyond their contracted obligations, which benefits system operators by providing additional optionality but also indicates that contracted capacity volumes may understate actual available response capability.
The broader efficiency case for integration rests on reducing the total cost of reserve procurement across the synchronous area by allowing TSOs to draw on the cheapest available bids regardless of origin. The financial performance of TERRE before its retirement, and the documented price convergence effects as PICASSO membership has expanded, establish that cross-border exchange delivers measurable cost reductions. The outstanding question is whether the MARI platform can generate comparable efficiency gains in the mFRR layer as more TSOs connect and the geographic scope of the cross-border merit order widens beyond its current concentration in two regional clusters.
