Issuu

Flexibility Needs Assessment report for Belgium October 8, 2026 Version approved by CREG

Flexibility Needs Assessment report for Belgium

October 8, 2026 Version approved by CREG

This report is drafted by Elia, agreed to be the designated entity for the Belgian Flexibility Needs Assessment. This report is based on system needs analyses conducted by Elia and network needs analyses conducted by Elia and all distribution system operators Fluvius, Sibelga, Ores, Resa and Arewal, supported by Synergrid. This draft report is submitted to the CREG for approval on October 2, 2026 after a consultation between July 24 and September 4, 2026. This version was approved on October 8, 2026.

FNA report for Belgium 2026, version approved by CREG

Introduction to the report This document builds on a non-binding template for the national Flexibility Needs Assessment Reports (FNA reports), which must be adopted in 2026, and biennially thereafter, by the regulatory authority or another authority or entity designated by the Member State in accordance with Article 19e(1) of Regulation (EU) 2019/943 (designated body). Pending the implementation of the Belgian legal framework for the national Flexibility Needs Assessment, Elia was confirmed as the designated entity for this first Belgian exercise and entrusted with the preparation of the report in cooperation with the Belgian DSOs. Elia conducted a public consultation from 24 July 2026 until 4 September 2026. The Directorate-General for Energy provided its opinion to the CREG on 4 September 2026. Following this consultation, the revised report was submitted to the CREG on October 2, 2026. By draft decision of October 8, 2026, the CREG approved the report, pending the adoption of the Belgian legislative framework governing future Flexibility Needs Assessments. The final report, as approved by the CREG and amended where applicable in line with the conditions set out in the CREG decision, was published on October 8, 2026 and transmitted to ACER. The flexibility needs assessments included in the FNA reports must comply with the Flexibility Needs Assessment Methodology (FNA methodology) as approved by ACER and Article 19e of Regulation (EU) 2019/943 (Electricity Regulation). Developed by ACER, this template is intended to support the designated bodies by facilitating the preparation of the FNA reports and enhancing consistency across them. The use of the template was considered optional, and the designated bodies remain free to decide whether and how to use it. The template is not exhaustive of all aspects that may be reported under the FNA methodology, and designated bodies may provide additional information beyond that suggested in the template.

FNA report for Belgium 2026, version approved by CREG

Table of contents Table of acronyms ................................................................................................................................ 8 Executive summary .............................................................................................................................. 9 Scenario and assumptions ................................................................................................................. 10 Key findings ........................................................................................................................................ 12 Interpretation of key findings .............................................................................................................. 14 Policy options addressing uncovered needs ..................................................................................... 16 Barriers to flexibility, regulatory recommendations and actions ........................................................ 22 Areas for further development of the next FNA ................................................................................. 23 1.

Introduction ............................................................................................................................... 24 1.1.

Purpose and scope of the national report ............................................................................. 26

1.2.

Legal and regulatory context ................................................................................................ 30

1.3.

Governance: bodies involved in the FNA report ................................................................... 31

1.4.

Key definitions ...................................................................................................................... 31

Methodological aspects ........................................................................................................... 35 2.1.

Data sources ......................................................................................................................... 35

2.2.

Target years considered ....................................................................................................... 36

2.3.

Modelling approach .............................................................................................................. 36

Scenarios and assumptions description ............................................................................... 45 3.1.

Scenarios used to assess system and network needs ......................................................... 45

3.1.1.

Scenarios for system needs ............................................................................................ 45

3.1.2.

Scenarios for network needs ........................................................................................... 46

3.1.2.1.

Scenarios for DSO data ............................................................................................... 46

3.1.2.2.

Scenarios for TSO data (if applicable) ......................................................................... 47

3.2.

Target for RES integration .................................................................................................... 47

3.3.

Portfolio of flexibility resources available .............................................................................. 48 Results – RES integration needs ............................................................................................ 54

FNA report for Belgium 2026, version approved by CREG

4.1.

Characterisation of residual load time series........................................................................ 54

4.2.

Characterisation of RES generation curtailment .................................................................. 58

4.2.1.

Daily curtailment profile ................................................................................................... 61

4.2.2.

Monthly curtailment profile .............................................................................................. 62

4.2.3.

Correlation of curtailment with renewable generation and demand ................................ 63

4.2.4.

Curtailment duration ........................................................................................................ 64

4.2.5.

Conclusion ....................................................................................................................... 65

4.3.

Uncovered RES integration needs ....................................................................................... 66

4.4.

Flexible resources needed to meet the uncovered needs .................................................... 69

4.4.1.

Quantification with single optimisation ............................................................................ 69

4.4.2.

Quantification of flexibility resources with combined optimisation .................................. 73

4.4.2.1.

Optimal portfolio sizing under full RES integration (zero acceptable curtailment) ...... 73

4.4.2.2.

Portfolio sensitivity to tolerated RES curtailment ......................................................... 76

4.4.2.3.

Trading off flexible resources against renewable capacity .......................................... 81

4.4.2.4.

Conclusion ................................................................................................................... 83

Results – Ramping needs ........................................................................................................ 85 5.1.

Ramping needs ..................................................................................................................... 85

5.2.

Characterisation of ramping capacity ................................................................................... 87

5.3.

Uncovered ramping needs.................................................................................................... 88

5.4.

Flexible resources needed to meet uncovered ramping needs ........................................... 90

Results - Short-term flexibility needs ..................................................................................... 92 6.1.

Short-term flexibility needs ................................................................................................... 92

6.1.1.

Quantification of the flexibility needs ............................................................................... 92

6.1.2.

Duration of forecast errors ............................................................................................... 95

6.1.3.

Sensitivity on the yearly forecast improvements ............................................................. 96

6.2.

Characterisation of short-term flexibility margins ................................................................. 96

6.3.

Uncovered short-term flexibility needs ............................................................................... 100

6.3.1.

Uncovered short-term upward flexibility needs ............................................................. 100

FNA report for Belgium 2026, version approved by CREG

6.3.2.

Uncovered short-term downward flexibility needs......................................................... 103

6.3.3.

Energy from uncovered downward short-term flexibility events .................................... 111

6.3.4.

Sensitivities on uncovered downward short-term flexibility ........................................... 114

6.4.

Flexible resources needed to meet the uncovered needs .................................................. 115

6.4.1.

Reserve capacity needs ................................................................................................ 116

6.4.1.1.

FCR needs ................................................................................................................. 117

6.4.1.2.

FRR needs - methodology and assumptions ............................................................ 117

6.4.1.3.

FRR needs - results ................................................................................................... 119

6.4.1.4.

aFRR needs ............................................................................................................... 120

6.4.1.5.

Relation to flexibility needs and means analysis ....................................................... 121

6.4.2. 7.

Additional flexibility resource needs .............................................................................. 123

Results - Network flexibility needs ....................................................................................... 127 7.1.

TSO network needs ............................................................................................................ 128

7.2.

Network needs on T/DSO connection points...................................................................... 129

7.2.1.

Context .......................................................................................................................... 129

7.2.2.

Results ........................................................................................................................... 130

7.2.3.

Conclusions ................................................................................................................... 134

7.3.

DSO network needs ............................................................................................................ 135

7.3.1.

Overvoltage ................................................................................................................... 136

7.3.2.

Internal DSO Congestion .............................................................................................. 137

7.4.

Unavailability following grid constraints .............................................................................. 137

7.4.1.1.

Available data ............................................................................................................ 138

7.4.1.2.

Result ......................................................................................................................... 139

7.4.1.3.

Future improvements ................................................................................................. 141

Synthesis - Guiding criteria application outcomes ............................................................. 142 8.1.

Finetuning ........................................................................................................................... 143

8.2.

Guiding criteria .................................................................................................................... 145

8.2.1.

Non-fossil fuel flexibility needs for RES integration needs ............................................ 145

FNA report for Belgium 2026, version approved by CREG

8.2.2.

Additional RES response required to cover RES curtailment ....................................... 149

8.2.3.

Sensitivity on nuclear and end-user flexibility ............................................................... 155

8.2.4.

Qualitative assessment of the economically efficient balance ...................................... 157

Barriers for flexibility and contribution of digitalisation .................................................... 158 Barrier 1 : Lack of proper legal framework for market access to new entrants and small actors .... 159 Barrier 2 : Lack of enablers and incentives to provide flexibility ...................................................... 160 Barrier 3 : Restrictive requirements to provide balancing services .................................................. 162 Barrier 4 : Restrictive requirements to provide TSO congestion management ............................... 163 Barrier 5 : Complex, lengthy, and discriminatory administrative requirements................................ 164 Barrier 6 : Lack of regulatory incentives to system operators to consider non-wire alternatives..... 164 9.1.

10.

Digitalisation ....................................................................................................................... 165 Conclusions ............................................................................................................................ 166

10.1.

Key findings ........................................................................................................................ 166

10.2.

Next steps ........................................................................................................................... 167

11.

References ............................................................................................................................... 170

Glossary ............................................................................................................................................. 171 Annex: Analysis of uncovered renewable response needs before any renewable reaction. ... 172

FNA report for Belgium 2026, version approved by CREG

Table of acronyms Acronym

Meaning

Adeqflex’25

Adequacy and Flexibility Study for Belgium over the horizon 2026-36, published in June 2025

AF25

Adeqflex’25 scenarios

aFRR

Automatic Frequency Restoration Reserve

BAT+

Scenarios accounting for battery potential in Belgium

BESS

Battery Energy Storage System

CC (EU-BASE-CC)

Current Commitments scenario, aligned with existing policies

CRM

Capacity Remuneration Mechanism

CT (EU-BASE-CT)

Constrained Transition scenario, with slower electrification

Day-Ahead (market timeframe)

DSO

Distribution System Operator

DSR

Demand Side Response

EDS

Detailed Study

Economic Dispatch

Electric Vehicle

FCR

Frequency Containment Reserve

FNA

Flexibility Needs Assessment

FRR

Frequency Restoration Reserve for restoring system frequency

FRCE

Frequency Restoration Control Error

FSP

Flexibility Service Provider

Heat Pump

Intraday market timeframe

mcyear

Monte Carlo (simulated) year

mFRR

Manual Frequency Restoration Reserve

Megawatt

MWh

Megawatt hour

NECP

National Energy Climate Plan

NRAA

National Resource Adequacy Assessment

Photovoltaic (solar generation)

RES

Renewable Energy Sources

RTE

Round-trip efficiency of storage technologies

ToE

Transfer of Energy (framework for flexibility services)

TYNDP

Ten Year Network Development Plan

TSO

Transmission System Operator

Cross-Border

FNA report for Belgium 2026, version approved by CREG

Executive summary EU Regulation 2019/943 requires that every Member State adopts a report on the estimated flexibility needs for a period of at least the next 5 to 10 years at national level. Elia and DSOs cooperated within Synergrid to develop a plan for the national implementation in Belgium. The implementation framework was developed in cooperation with the CREG, which approved the implementation plan and is responsible for approving the FNA report before its submission to ACER. While awaiting the revision of the Belgian Federal Electricity Law in line with the EU Regulation, Elia was designated as the entity adopting the report in cooperation with Belgian DSOs Fluvius, Sibelga, ORES, RESA and Arewal. This cooperation was supported by Synergrid. The Flexibility Needs Assessment (FNA) report is intended to support Member States in defining indicative national objectives for non-fossil flexibility and justify the need for support mechanisms where necessary. To this end, the assessment integrates the three methodological building blocks as shown in Figure 1: 1. RES integration needs assess RES curtailment in relation to a RES integration target. This target is the maximum level of RES curtailment compatible with the achievement of national RES targets. They assess how well the power system can manage predictable excess during periods of very high renewable generation. 2. Short-term and ramping needs assess the system’s capability to cope with large ramps of demand or generation, unexpected outages and forecasting errors. It assesses as well upward as downward flexibility needs. Particular attention is given on periods of excess generation, and to identify situations where available flexibility may be insufficient to safely manage system operation. 3. Network needs assess the impact of congestion on transmission and distribution networks. In this first edition, the assessment focuses on periods of excess generation. It evaluates the extent to which network constraints may limit the ability of the system to integrate renewable generation and make use of available flexibility resources.

Figure 1 : Overview of the FNA methodology

These needs are combined into a single assessment framework used to identify the additional flexibility needs of the power system after accounting for existing and expected flexibility delivered through market developments and adequacy mechanisms such as the CRM. The assessment is based on the amount of RES curtailment exceeding a predefined RES curtailment target, where such a target exists. As Belgium currently does not define an explicit target regarding acceptable levels of RES curtailment, the assessment instead evaluates the implications of different curtailment outcomes and the associated flexibility requirements. FNA report for Belgium 2026, version approved by CREG

To this end, the assessment integrates the three methodological building blocks into a single, coherent framework. Rather than identifying a single preferred pathway, it evaluates two boundary cases: 1. Additional non-fossil flexibility needs, required to achieve full RES integration which implies zero curtailment; 2. The level of RES curtailment that would occur in the absence of additional flexibility beyond what is already expected from market developments and adequacy mechanisms. In the latter case, additional RES generation capacity must be installed to compensate for the curtailed energy and reach the National Energy Climate Plans (NECP) target while the system must dispose of sufficient response on renewable assets to facilitate this curtailment. It is important to stress that these two boundary cases do not represent policy recommendations. Instead, they provide a transparent framework for assessing the trade-offs between deploying additional flexibility, accepting RES curtailment and installing additional renewable generation capacity. This would allow policy makers to set indicative targets for non-fossil flexibility, and justify the need for support mechanisms where necessary, in line with the objective of the FNA. The level of RES curtailment can be assessed as level of energy (on yearly basis) and in capacity (on instantaneous basis). Based on these results, the assessment can provide insights into the extent to which the existing and expected response of renewable assets is sufficient to manage periods of RES curtailment in the absence of additional flexibility, or whether additional renewable response may be required. These results provide policy makers with quantitative information to assess the trade-offs between additional non-fossil flexibility and acceptable levels of RES curtailment, as well as the implications of such curtailment levels in terms of additional renewable response requirements and additional renewable generation capacity needed to compensate for curtailed energy and achieve renewable energy targets. The purpose of the assessment is not to prescribe a preferred solution, but to quantify the implications of different policy choices. The results can support the definition of indicative national objectives for non-fossil flexibility and contribute to future NECPs. At European level, the outcomes of national FNAs are also used by ACER to support further electricity market integration. The Belgian FNA report was submitted on October 5, 2026 to the CREG for approval after a consultation between July 24 and September 4, 2026. Following the opinion of the Federal Public Service Economy (SPF Energy) the CREG approved the report on October 8, 2026 prior to its publication and transmission to ACER. The next version of the report is awaited in the first semester of 2028. The exact planning of the next version will be determined via a revision of Article 7bis of the Federal Electricity Law, currently under development, specifying the national implementation process of the Adequacy and Flexibility Study framework.

Scenario and assumptions The FNA is built around four scenarios. Two transition pathways are derived from Belgium’s National Resource Adequacy Assessment (NRAA), Adeqflex’25, namely the Current Commitments (CC) and Constrained Transition (CT) scenarios. These are combined with two battery deployment assumptions in order to assess the sensitivity of the identified flexibility needs to future storage deployment. The first battery deployment assumption corresponds to the assumptions ‘without gap filling’ used in Adeqflex’25 and therefore only includes battery capacities existing or contracted through CRM auctions before CRM 2025, and thus up to CRM 2024. These align exactly with the reference scenarios in Adeqflex’25, in line with the requirements set in the FNA methodology, and agreed as such in the implementation plan. This approach allows the assessment to build on existing economic dispatch simulations without requiring additional economic dispatch modelling. •

a ‘Current Commitments’ scenario (CC-AF25) aligned with current European, federal and regional policy ambitions.

FNA report for Belgium 2026, version approved by CREG

• a ‘Constrained Transition’ scenario (CT-AF25) reflecting a slower deployment of electrification, renewable generation, storage and infrastructure. The FNA also wants to identify flexibility needs beyond the flexibility expected to be delivered through market-based investments and existing adequacy mechanisms such as the CRM. To evaluate the impact of battery deployment assumptions on the identified flexibility needs, a second battery deployment assumption is considered, accounting for additional expected large-scale battery parks. The FNA does not currently distinguish between medium-scale and large-scale batteries. For modelling purposes, all non-residential battery assets (> 10 kW) connected to the transmission or distribution network are grouped under the large-scale battery category, while residential batteries are modelled separately. This simplification has no material impact on the identified flexibility needs, which are assessed based on the aggregated storage contribution to the system. While the CC-AF25 and CT-AF25 scenarios assume an installed large-scale battery capacity of 1.5 GW in 2030 and 2035, they do not yet account for the additional 0.7 GW contracted through the capacity remuneration mechanism in 2025, nor for additional capacities that may result from future CRM auctions. The BAT+ scenarios reflect a higher battery deployment assumption, corresponding to the large-scale battery potential identified in Adeqflex’25, amounting to 4.5 GW in 2030 and 5.7 GW in 2035. • a ‘Current Commitments’ scenario (CC-BAT+) 1 • a ‘Constrained Transition’ scenario (CT-BAT+) Given that additional battery capacity has already been contracted beyond the volumes included in the AF25 scenarios and that future CRM auctions remain to be organised, the CC‑AF25 and CT‑AF25 scenarios should be regarded as representing a minimum battery deployment assumption rather than a forecast of future battery deployment. Conversely, the CC‑BAT+ and CT‑BAT+ scenarios reflect the battery potential identified in Adeqflex’25 and provide an indication of how the identified flexibility needs evolve under a more extensive deployment of storage. It is to be noted that this potential is derived from a selected share of projects identified through feasibility and connection studies and are not necessarily economically viable. As such projects can be delayed, or stopped, and there is no certainty this capacity will be fully realised. Where the AF25 scenarios can be seen as a lower bound, the BAT+ scenarios should be regarded as representing a maximum battery deployment assumption rather than a forecast of future battery deployment.

Figure 2: Scenarios, sensitivities and target years of the FNA

The BAT+ scenarios in the FNA reflect a higher battery deployment assumption based on the battery potential identified in Adeqflex'25. This should not be confused with the BAT+ sensitivities used in the Adeqflex'25 Economic Viability Assessment, which reflect more favourable market conditions for battery investments.

FNA report for Belgium 2026, version approved by CREG

In all four scenarios, the assessment already incorporates the following sources of flexibility in the Belgian power system: • Conventional flexibility from larger thermal units, demand side management and storage, primarily large-scale batteries complementing existing capacities such as pumped storage, • End-user flexibility which is assumed to increase considerably driven mainly by small-scale batteries, electric vehicles and heat pumps, • RES flexibility, which assumes that at least all large capacity reacts, further complemented with decentral shares of reactions based on extrapolations of current observations. • Cross-border flexibility is accounted for through economic dispatch simulations which are conducted for Belgium in an interconnected European system (as outlined in Adeqflex’25). Shares of available cross-border flexibility in intra-day and balancing capacity are accounted based on current observations. In addition to the four scenarios, several sensitivities are assessed to evaluate the impact of key assumptions on the identified flexibility needs. These sensitivities include higher levels of renewable response, and even up to full renewable response, additional nuclear lifetime extensions, and lower levels of end-user flexibility. They are used to assess how the identified flexibility needs vary under alternative assumptions regarding system flexibility and generation mix.

Key findings After finetuning of the RES integration needs including short-term, ramping and identified network needs, RES generation curtailment is used as an indicator of additional flexibility needs. The results shown in Table 1 are expressed as maximum need (GW), average annual RES curtailment is identified in amount of curtailed energy (TWh), and as percentage of yearly generated wind and solar generation. The additional flexibility needs are determined following the RES curtailment exceeding the predefined RES curtailment target. As Belgium currently does not define an explicit target regarding acceptable levels of RES curtailment, the assessment first quantifies the expected levels of RES curtailment resulting from the different scenarios. These results provide the basis for evaluating the trade-offs between additional flexibility, renewable response and acceptable levels of RES curtailment presented later in the assessment. In the absence of such a target in Belgium, and assuming a zero RES curtailment target for Belgium, the uncovered capacity is equal to the RES curtailment. Figure 3 shows that total expected RES curtailment remains limited across all assessed scenarios and amounts in 2030 to: • 1.0% of annual wind and solar generation in the CC-AF25 scenario; • 0.5% of annual wind and solar generation in the CC-BAT+ scenario; • 1.3% of annual wind and solar generation in the CT-AF25 scenario; • 0.4% of annual wind and solar generation in the CT-BAT+ scenario. This is expected to increase in 2035 to: • 1.9% of annual wind and solar generation in the CC-AF25 scenario; • 1.1% of annual wind and solar generation in the CC-BAT+ scenario; • 2.5% of annual wind and solar generation in the CT-AF25 scenario; • 1.1% of annual wind and solar generation in the CT-BAT+ scenario. Both scenarios result in relatively low levels of RES curtailment. Compared with the AF25 assumptions, which only account for battery capacities existing or contracted through CRM auctions up to 2024, the BAT+ assumptions, which are based on the battery potential identified in Adeqflex’25, facilitates FNA report for Belgium 2026, version approved by CREG

renewable integration and reduces the associated flexibility needs: the limited annual RES curtailment is further reduced by approximately 40 - 60% in 2030 and by 30 - 50% in 2035. Table 1: Key findings - RES integration (fine-tuned); the ranges reflect the differences between the AF25 and BAT+ scenarios.

Type

Flexibility need

Indicator

Target Year

Scenario

RES integration

[% of RES production]

340 (1.0%)

CC BAT+

2.7

(0.5 %)

4.8

(1.3%)

2.6

(0.4 %)

7.8

(1.9%)

5.9

(1.1 %)

7.9

(2.5%)

5.4

(1.1 %)

AF25

System needs

Uncovered energy [GWh]

4.4

AF25

2030

Max uncovered capacity* [GW]

RES generation curtailment - total

BAT+

CC AF25

CC BAT+

2035 CT AF25

CT BAT+

160 410 130 1020 600 1180 500

*Based on the 99.9th percentile.

Figure 3: Total RES curtailment resulting from RES integration, short-term needs, ramping needs and network needs

The general conclusion is that the amount of RES curtailment following excess generation in Belgium remains limited in annual energy terms across all assessed scenarios, despite the strong growth in renewable capacity towards 2035. It is however to be kept in mind that additional RES curtailment can occur in practice following negative prices driven by excess generation in other countries. The identified curtailment is concentrated in a limited number of periods characterised by excess renewable generation. These results provide the basis for evaluating the trade-offs between additional flexibility, renewable response and acceptable levels of RES curtailment presented later in the assessment.

FNA report for Belgium 2026, version approved by CREG

Interpretation of key findings The individual drivers of system and network flexibility needs - RES integration needs, Ramping needs and Short-Term needs and Network Needs - are summarised in Table 2 and Table 3. The RES curtailment indicator is used in these tables to compare the contributions of the different flexibility needs types in terms of annual curtailed energy and maximum uncovered capacity. Overall, the results show that RES integration needs are the main contributor to the flexibility needs. The values presented for the different flexibility needs cannot be summed, since the final assessment accounts for overlaps and interactions between the different needs through the finetuning process. CC

2030

Type

Flexibility need (downwards)

Max. need* [MW]

RES Integration needs System needs

Network needs

– AF25

BAT+

Max. uncovered capacity* [MW]

Uncovered energy [GWh]

Max. need* [MW]

[% of RES production]

140 – 290

2700 - 4100

– AF25

BAT+

2600 - 4400

(0.4 – 0.8%)

0.1 – 4

Ramping needs

3400

0 - 10

Short-term needs

2800

1000 - 1800

(0 – 0.1%)

T/DSO connection point needs

(~0%)

Max. uncovered capacity* [MW]

(~0%)

3 – 20

Not applicable

TSO network needs

Not applicable

[% of RES production]

100 – 340 (0.3 – 1.1.%)

0.5 – 7

3300

0 - 300

2700

1200 - 2100

(0 – 0.1%)

(~0%)

DSO network needs

Uncovered energy [GWh]

(~0%)

5 – 33

Table 2: Summary of res integration needs, ramping needs, short-term needs and network needs for 2030. Maxima in this table correspond to the 99.9th percentile. The ranges reflect the differences between the AF25 and BAT+ scenarios. The BAT+ scenarios always correspond to the lower bound and the AF25 scenarios to the higher bound. CC

2035

Type

Flexibility need (downwards) RES Integration needs

System needs

Network needs

BAT+

Max. need* [MW]

– AF25

Max. uncovered capacity* [MW]

BAT+

Uncovered energy [GWh] [% of RES production]

Max. need [MW] 1

480 – 850

5800 - 7500 4300

0 - 100

Short-term needs

3400

1000 - 2300

T/DSO connection point needs

550

0.4 – 10 (~0%)

2 – 31 (~0%)

130 (0.3%)

Max. uncovered capacity* [MW]

Uncovered energy [GWh] [% of RES production]

360 – 980

5400 - 7500

(0.9 – 1.6%)

Ramping needs

– AF25

(0.8 – 2.1%)

0.2 – 10

4100

0 - 500

3300

1400 - 2900

(0 – 0.2%)

550

(0.3%)

DSO network needs

Not applicable

TSO network needs

Not applicable

(~0%)

5 – 57 130

Table 3 : Summary of res integration needs, ramping needs, short-term needs and network needs for 2035. Maxima in this table correspond to the 99.9th percentile. The ranges reflect the differences between the AF25 and BAT+ scenarios. The BAT+ scenarios always correspond to the lower bound and the AF25 scenarios to the higher bound.

FNA report for Belgium 2026, version approved by CREG

• RES integration needs (before finetuning) In terms of annual energy volumes, the uncovered RES integration needs represent the largest driver of the identified flexibility needs. They correspond to the expected RES curtailment resulting from forecasted system conditions before accounting for the contribution of ramping, short-term and network needs. As Belgium currently does not define an explicit target regarding acceptable levels of RES curtailment, they are equal to the maximum needs. These values should be interpreted as the RES curtailment resulting from the assessed scenarios rather than the result of a policy target. They provide the basis for evaluating later in the assessment the trade-offs between additional non-fossil flexibility, acceptable levels of RES curtailment and additional needs for renewable response. Nevertheless, RES curtailment remains limited when expressed relative to total renewable generation. Under the AF25 assumptions, which only account for battery capacities existing or contracted through CRM auctions up to 2024, annual RES curtailment amounts in the CC scenarios to approximately 290 GWh (0.8%) in 2030 and 850 GWh (1.6%) in 2035. Under the BAT+ assumptions, reflecting the battery potential identified in Adeqflex’25, these values decrease to approximately 140 GWh (0.4%) in 2030 and 480 GWh (0.9%) in 2035. Similar reductions are observed in the CT scenarios, where annual curtailment decreases from 340 GWh (1.1%) to 100 GWh (0.3%) in 2030 and from 980 GWh (2.1%) to 360 GWh (0.8%) in 2035 when moving from AF25 to BAT+. In terms of frequency of occurrence, domestic curtailment remains concentrated in a relatively limited number of hours. Under the AF25 assumptions, curtailment occurs during approximately 200 hours in 2030 and 400 hours in 2035. Under the BAT+ assumptions, this decreases to approximately 100 hours in 2030 and 250 hours in 2035. This illustrates that additional battery deployment reduces both annual curtailed energy volumes and the number of periods during which curtailment occurs. Nevertheless, rare but large events become more relevant toward 2035. While annual curtailed energy remains limited, instantaneous curtailment levels may reach 2.7–4.4 GW in 2030 and 5.4–7.5 GW in 2035 (99.9th percentile), depending on the battery deployment assumptions. The lower values correspond to the BAT+ scenarios, while the higher values correspond to the AF25 assumptions. Curtailment tends to be concentrated around midday and is observed more frequently between March and September, when excess renewable generation conditions are most likely to occur. Long-duration curtailment events lasting up to 15–19 hours may also occur under exceptional conditions. These results further illustrate that additional battery deployment reduces not only annual curtailed energy volumes and the number of curtailment hours, but also the magnitude of the most severe curtailment events. • Ramping Needs Ramping needs, calculated on forecasted hourly residual load variations, can be substantial in capacity terms, reaching up to 3 – 4 GW in 2030 – 2035. However, uncovered ramping needs remain very limited, contribute only marginally to annual curtailed energy and their magnitude is further reduced in the BAT+ scenarios. This is because the situations giving rise to ramping needs largely overlap with the same periods of excess generation already captured by the RES integration needs. As a result, the same flexibility resources are expected to address both needs. • Short term flexibility The short-term flexibility needs, based on forecast errors and forced outages from intra-day to realtime, confirm that both upward and downward short-term flexibility needs increase over time as the system becomes more weather driven. Total downward short-term flexibility needs increase from 2.4 GW in 2026 to 2.8 GW in 2030 and 3.4 GW in 2035, while total upward flexibility needs increase from 3.4 GW in 2026 to 3.8 GW in 2030 and 4.7 GW in 2035 in the CC scenarios. In the CT scenario the upand downward flexibility needs are around 100 MW lower, mainly due to lower installed wind capacities and lower demand.

FNA report for Belgium 2026, version approved by CREG

While uncovered upward needs are expected to continue being managed through existing cross-border balancing markets and reserve procurements mechanisms, maintaining sufficient downward flexibility may require additional flexible resources. Uncovered downward needs remain during periods of excess renewable generation which amount to: • 1.0 - 1.8 GW in 2030 and 1.0 - 2.3 GW in 2035 for the CC-BAT+ and CC-AF25 scenarios respectively; • 1.2 – 2.1 GW in 2030 and 1.4 - 2.9 GW in 2035 for the CT-BAT+ and CT-AF25 scenarios. in CT-BAT+ and CT-AF25 respectively. The amount of uncovered short-term flexibility needs is significantly influenced by the availability of battery storage. Compared with the AF25 assumptions, which only account for battery capacities existing or contracted through CRM auctions up to 2024, the BAT+ assumptions reduce the identified uncovered downward needs by approximately 30% to 50%, depending on the scenario and target year. This illustrates the contribution that additional battery deployment can make to managing short-term imbalances and periods of excess generation. In annual energy terms, however, these volumes remain very limited. This is explained by the probabilistic nature of forecast errors and forced outages. Short‑term flexibility needs are evaluated at the P99.9 forecast‑error level, corresponding to extreme but infrequent activation frequencies. As a result, although large downward flexibility capacities are required during these situations, the number of hours concerned remains limited and the associated energy volumes remain small. • Network Needs The assessment, in this first edition of the FNA, focusses on quantifying the impact of TSO/DSO coupling point constraints on local RES curtailment and network limitations affecting the availability of flexible resources. Both are found to have a relatively limited impact on the overall results compared to system-level flexibility needs. Concerning the network needs on T/DSO connection points: in 2035, the network needs quantified by DSOs and resulting from TSO/DSO coupling point constraints amount to approximately 0.3% of total RES generation in Belgium in both the CC and CT scenarios. When expressed in terms of RES integration needs, they represent around 3% of the RES integration needs in 2030 and 14% in 2035 in the AF25 scenarios. In the BAT+ scenarios, the corresponding shares amount to around 11% in 2030 and 37% in 2035. The higher percentage for the BAT+ scenarios is explained as the same network needs are compared with a lower curtailment level for system needs, thus the resulting percentage is higher. It is to be noted that these results are to be considered conservative, as they do not yet account for future flexibility activations or additional grid reinforcements. Curtailment is concentrated at a limited number of TSO/DSO coupling points and may reach approximately 0.5 GW. Annual curtailment volumes are distributed almost equally between Wallonia, where they are mainly associated with wind generation, and Flanders, where both wind and solar generation contribute. No network-related curtailment is identified in Brussels. In Wallonia, curtailment mainly occurs during high-wind and low-load situations, while in Flanders it can occur during both windy conditions and periods combining strong solar generation with low demand. Concerning the unavailability of flexible resources: in 2030 and 2035, a quantitative assessment based on connection studies shows that limitations on transmission level connected flexible assets, such as batteries, to cover downward flexibility needs, is expected to be very limited.

Policy options addressing uncovered needs To support policy makers in defining indicative national objectives for non-fossil flexibility, the assessment combines the different flexibility needs into a single, coherent framework. For Belgium, the

FNA report for Belgium 2026, version approved by CREG

RES target in the current National Energy & Climate Plan (NECP) does not, at least explicitly, consider RES curtailment. The assessment therefore evaluates two boundary cases: (A) the additional non-fossil flexibility needs required to achieve full RES integration (B) the level of RES curtailment that would occur in the absence of additional flexibility beyond what is already expected in the system, together with the additional renewable generation capacity required to compensate for the curtailed energy. When considering the second case, the assessment also investigates (C) the additional RES response required to safely manage periods of curtailment. These three perspectives provide policy makers with quantitative information on the trade-offs between developing additional flexibility, accepting RES curtailment and ensuring sufficient renewable response to operate the system safely. The present FNA adopts a technology-neutral approach and therefore quantifies the implications associated with different boundary cases rather than identifying an economically optimal outcome. In the absence of an explicit RES curtailment target in Belgium, future editions of the FNA could be complemented by a quantitative social welfare assessment aimed at identifying the economically efficient balance between additional flexibility deployment, acceptable RES curtailment and the associated additional renewable generation capacity. Such an assessment could help identify the economically justified level of flexibility and curtailment and provide additional input for the definition of indicative flexibility objectives and is to be investigated and confirmed in the implementation plan for the FNA 2028. The present FNA nevertheless provides (D) a qualitative assessment of where the economically efficient balance may lie between additional flexibility deployment, acceptable RES curtailment and the associated additional renewable generation capacity. A. Additional non-fossil fuel flexibility needs to ensure full integration This first boundary case assesses the amount of additional non-fossil flexibility that would be required to achieve near-full RES integration, corresponding to a situation where RES curtailment is limited to 0.1% of annual renewable generation (Figure 4 left). Under this assumption, the assessment identifies the following additional flexibility needs: • 11 to 21 GW in the CC-AF25 scenario for 2030 and 2035 respectively • 11 to 23 GW in the CT-AF25 scenario for 2030 and 2035 respectively • 9 to 18 GW in the CC-BAT+ scenario for 2030 and 2035 respectively • 9 to 17 GW in the CT-BAT+ scenario for 2030 and 2035 respectively Achieving near-full RES integration requires significant additional flexibility volumes, reflecting the fact that the remaining curtailment is concentrated in a limited number of periods characterised by high renewable generation and correspondingly high flexibility requirements. Compared with the AF25 assumptions, which only account for battery capacities existing or contracted through CRM auctions up to 2024, the battery potential identified in Adeqflex’25 reduces these additional flexibility requirements. The costs associated with developing such additional flexibility resources should therefore be assessed against the alternative boundary case presented in Section B, namely accepting RES curtailment and compensating the curtailed energy through additional renewable generation capacity.

FNA report for Belgium 2026, version approved by CREG

Figure 4: Combined fine‑tuned results showing (left) the additional non‑fossil flexibility required to achieve near‑full RES integration and (right) the additional renewable generation capacity required to compensate for RES curtailment when no additional non‑fossil flexibility is developed

B. Additional RES capacity to cover for RES curtailment This second boundary case assesses the amount of additional renewable generation capacity that would be required to compensate for RES curtailment when no additional non-fossil flexibility is developed beyond the flexibility already assumed in the scenarios. Figure 4 (right) shows the additional solar PV or onshore wind capacity that would be required to compensate for the curtailed energy. They represent alternative ways of compensating for the same curtailed energy through either additional solar PV or additional onshore wind capacity. Under this assumption, the additional renewable generation capacity required amounts to: • 0.4 to 1.3 GW for solar PV and 0.2 to 0.5 GW for onshore wind in the CC-AF25 scenario for 2030 and 2035 respectively • 0.5 to 1.7 GW and 0.2 to 0.5 GW for onshore wind in the CT-AF25 scenario for 2030 and 2035 respectively • 0.2 to 0.7 GW and 0.1 to 0.3 GW for onshore wind in the CC-BAT+ scenario for 2030 and 2035 respectively • 0.2 to 0.6 GW and 0.2 to 0.2 GW for onshore wind in the CT-BAT+ scenario for 2030 and 2035 respectively These results indicate that relatively modest additions of solar PV or onshore wind capacity can compensate for the curtailed energy while maintaining the achievement of renewable energy targets. The battery potential identified in Adeqflex’25 further reduces both RES curtailment and the corresponding amount of additional renewable generation capacity required compared with the AF25 assumptions, which only account for battery capacities existing or contracted through CRM auctions up to 2024.

FNA report for Belgium 2026, version approved by CREG

C. Additional RES response required to safely manage periods of curtailment under the second boundary case This section complements the second boundary case by assessing whether the RES curtailment levels identified under that case can be managed safely and which level of additional renewable response may be required. Accepting RES curtailment does not automatically imply that periods of excess generation can be managed safely. During such periods, renewable assets may need to provide additional response by reducing output when required. If insufficient renewable response or other flexibility resources are available, excess generation may remain in the system and could create operational challenges. The assessment of renewable response needs should therefore not be interpreted as implying that renewable response is the only possible solution. Rather, it quantifies the residual downward flexibility needs associated with the second boundary case. These needs could be addressed through additional non-fossil fuel flexibility resources as considered in the first boundary case. In practice, the two boundary cases should be considered as stylised extremes, and intermediate solutions combining additional flexibility with a certain level of RES curtailment may also be envisaged. This distinction is particularly relevant for the AF25 scenarios, which only account for battery capacities existing or contracted through CRM auctions up to 2024. Additional battery deployment, as represented in the BAT+ scenarios, reduces these residual needs significantly before renewable response is considered. While the two options are presented as boundary cases, the requirement to eliminate curtailment entirely is an extreme assumption, as it requires additional flexibility to be developed to address even rarest curtailment events. At the other end of the spectrum, accepting the full level of assessed curtailment would still imply curtailing only around 2.5% of total RES generation in the most extreme scenario. The level of renewable response required depends both on the battery deployment assumptions and on the selected reliability criterion. The assessment therefore evaluates renewable response needs under 1‑hour and 3‑hour reliability criteria. As no reliability criterion for periods of excess generation is defined in either the European FNA methodology or Belgian legislation, both criteria should be interpreted as reference points rather than targets. The 1‑hour criterion is aligned with system security needs and is justified due to lack of measures to manage excess generation (compared with load shedding plans used to manage shortages). The 3‑hour criterion is aligned with the reliability standard applied in Belgium's adequacy framework although no such support mechanisms exist for downward flexibility at the time of writing this report. Consequently, the renewable response needs identified under the 1‑hour criterion should be regarded as cautious, while the 3‑hour criterion results in significantly lower needs. Figure 5 presents the residual renewable response needs under the CC/CT and AF25/BAT+ scenarios. The values shown correspond to the 1‑hour reliability criterion. The assessment already accounts for the renewable response expected in the scenarios, including full participation of onshore wind generation and a decentralised PV response corresponding to 5% of installed PV capacity, based on market behaviour observed in 2025 2. A sensitivity assuming additional decentralised PV participation is also assessed.

The reference assumptions for solar PV response are based on a sample of observed reactions of individual assets during 1% of negative price periods in 2025. The 1% threshold, based on a very limited amount of observations, was retained to account for capacities reacting during the most extreme negative price events.

FNA report for Belgium 2026, version approved by CREG

Figure 5: RES response needs when accepting curtailment under the CC/CT – AF25/BAT+ scenarios after accounting full participation of all decentralised wind generation and 5% of PV fleet – these uncovered needs account for all expected flexible in the scenarios.

Under the 1‑hour reliability criterion, and after accounting for all expected flexibility and full onshore wind response, the remaining renewable response needs amount in 2030 to: • 2.7 – 3.6 GW generating power in the CC/CT-AF25 scenarios • 0.8 – 2.2 GW generating power in the CC/CT-BAT+ scenarios By 2035, these needs increase to: • 3.5 – 5.2 GW generating power in the CC/CT-AF25 scenarios • 1.7 – 3.5 GW generating power in the CC/CT-BAT+ scenarios Under the 3-hour reliability criterion, the identified renewable response needs are substantially lower, illustrating the sensitivity of the results to the selected reliability standard. The remaining renewable response needs amount in 2030 to: • 1.6 – 2.8 GW generating power in the CC/CT-AF25 scenarios • 0.0 – 1.0 GW generating power in the CC/CT-BAT+ scenarios

By 2035, these needs increase to: • 2.7 – 4.5 GW generating power in the CC/CT-AF25 scenarios • 0.6 – 2.1 GW generating power in the CC/CT-BAT+ scenarios

FNA report for Belgium 2026, version approved by CREG

Compared with the AF25 assumptions, which only account for battery capacities existing or contracted through CRM auctions up to 2024, the battery potential identified in Adeqflex’25 substantially reduces the remaining renewable response needs. As the assessment already assumes full participation of onshore wind generation, the remaining renewable response needs are associated with solar generation. While the remaining renewable response needs are expressed in terms of renewable response, they should not be interpreted as requiring renewable response only. These needs could also be reduced through additional non-fossil flexibility resources, including demand-side response and other sources of flexibility. The uncovered flexibility needs or remaining renewable response needs is largely impacted by the assumed battery deployment and reaction of renewable such as solar power on market prices. Any additional reaction of solar in the market on top of the 5% of total installed capacity assumed to be flexible in the scenarios will proportionally reduce the uncovered needs. Note that larger non-residential installations are expected to amount up to 5 to 6 GW in respectively 2030 and 2035. Note that when comparing with installed solar installations, a de-rating factor should be applied accounting that only 60 – 70 % of capacity is effectively generating during production. The capacity to preserve the residential segment will depend on the speed of decision to make new existing units steerable, the development of the market and the capacity to retrofit existing units. Sensitivities on nuclear lifetime extensions and lower end-user flexibility suggest that the identified renewable response needs could be higher under alternative assumptions. In 2025, observed solar PV response volumes reached up to 650 MW, corresponding to around 5% of installed PV capacity. This response level is assumed to grow proportionally with future PV deployment in the scenarios. Together with the backstop mechanisms developed by Elia and DSOs (1 GW), these volumes may already cover part of the identified needs. However, these volumes are expected to overlap to a large extent and are not sufficient to cover all scenarios under the 1-hour criterion. Furthermore, increasing renewable participation in response to market signals, together with recent and upcoming measures such as the continued rollout of smart meters, dynamic contracts and other market reforms, is expected to further enhance renewable participation during periods of excess generation. However, their impact could not be quantified in this first FNA study and is therefore not reflected in the results presented above. The remaining renewable response needs therefore remain highly dependent on the realization of the battery deployment assumptions and on the effectiveness of market-based measures aimed at increasing renewable participation. D. Qualitative assessment of the economically efficient balance This first FNA does not include a quantitative social welfare assessment comparing the costs and benefits of developing additional non-fossil flexibility with those associated with accepting a certain level of RES curtailment and deploying additional renewable generation capacity to compensate for the curtailed energy. It therefore does not identify the economically optimal balance between these options. Nevertheless, the quantitative results provide useful indications of where this balance may lie. In the BAT+ scenarios, which incorporate the battery potential identified in Adeqflex’25, RES curtailment remains limited to approximately 0.4–0.5% of annual wind and solar generation in 2030 and 1.1% in 2035, without additional non-fossil flexibility beyond the resources already included in those scenarios. These results suggest that the economically efficient outcome is likely to be close to a situation in which no significant additional non-fossil flexibility needs to be developed beyond the battery potential identified in Adeqflex'25. However, this qualitative conclusion could be confirmed when a quantitative social welfare assessment in a future FNA (after such assessment is confirmed for the implementation

FNA report for Belgium 2026, version approved by CREG

plan of the FNA 2028), taking into account the costs of additional flexibility, RES curtailment and the additional renewable generation capacity required to compensate for curtailed energy. Within the scope and assumptions of this first FNA, the results do not indicate a need for a dedicated flexibility support mechanism in addition to existing market arrangements and adequacy mechanisms (CRM). Priority should instead be given to ensuring that the battery potential identified in Adeqflex’25 can effectively materialise, notably by preventing grid connection procedures and constraints from becoming barriers to the development of economically viable battery projects. Measures aimed at strengthening renewable participation in response to market signals should also be pursued. Their effectiveness will be important in determining the residual flexibility needs and the level of curtailment that can be safely managed. Demand-side response may provide an additional contribution and should be considered alongside storage and renewable response as a source of nonfossil flexibility.

Barriers to flexibility, regulatory recommendations and actions Unlocking large-scale flexibility requires coordinated efforts across the entire energy ecosystem. Since Elia’s earlier publications, several general key enablers have progressed and continue to evolve. Accelerated and improved outcomes on the general key enablers will depend on stronger collaboration among all stakeholders in the areas summarized in this section. The evaluation of removal of regulatory barriers, proposal of relevant mitigation measures and incentives, and possible improvements to markets and system operation services or products uses the ACER Recommendation on NRA’s reporting on barriers for non-fossil flexibility as a basis. ACER identifies six regulatory and market design barriers and associated which are discussed accordingly in this section: 1.

Lack of proper legal framework for market access to new entrants and small actors

Lack of enablers and incentives to provide flexibility

Restrictive requirements to provide balancing services

Restrictive requirements to provide congestion management

Complex, lengthy, and discriminatory administrative requirements

Lack of regulatory incentives to system operators to consider non-wire alternatives

This list of barriers is not exhaustive, and ACER encourages Member States to identify and assess any additional barriers they consider relevant to the development of non-fossil flexibility. It provides a high level overview of the barriers identified by ACER, and potential enablers, complemented with an additional barrier identified by Elia on infrastructure and interoperability. In summary, the barrier assessment highlights the following key takeaways: •

Make it easy for customers (simplicity, market access and infrastructure): flexibility requires attractive and easy-to-understand offers, clear communication and the continued roll-out of smart meters. Particular attention is needed to unlock the flexibility potential of electric vehicles and heat pumps.

•

Give the right price incentives: price signals and tariff structures should appropriately reward consumers for shifting consumption to the right time and support the growth of flexibility.

•

Deliver infrastructure and interoperability: common standards, interoperability between digital devices and platforms, and efficient implementation of processes for requests and connection of flexible assets are key to enabling flexibility at scale.

FNA report for Belgium 2026, version approved by CREG

•

Enable market access: open and proportionate rules are needed to enable aggregators and flexibility providers to participate. This includes improvements to the Transfer of Energy (ToE) model to facilitate participation in day-ahead and intraday markets, simplification of onboarding and prequalification processes, scalable communication solutions, and facilitating participation across multiple flexibility products.

•

Incentivise non-wire alternatives: on TSO level, the TOTEX methodology is under development In addition, a robust regulatory framework is needed to apply the use of flexibility for managing grid constraints. On DSO level, solving congestion with flexibility faces specific challenges requiring a specific framework including real-time monitoring and forecasting of network needs, assessing availability of cost of flexibility and collaboration between TSO and DSO.

Areas for further development of the next FNA This first FNA represents an important step towards a comprehensive assessment of flexibility needs in Belgium. Several areas for further methodological development have already been identified and could be addressed in future editions of the FNA. This will be investigated and confirmed first in the implementation plan for the FNA 2028, notably: •

TSO Network Needs: develop and implement a comprehensive methodology to quantify flexibility needs associated with internal transmission network congestions and integrate these needs into the overall flexibility assessment.

•

DSO Network Needs: investigate whether and how the quantitative assessment could be extended beyond TSO/DSO coupling point constraints to address overvoltage, internal distribution network congestion and the availability of flexibility resources in distribution grids. Given the technical complexity of these assessments, the feasibility, methodology and implementation approach will first need to be evaluated in the implementation plan for FNA 2028.

•

Availability of flexible resources under network constraints: improve the assessment of the effective availability of flexibility resources during periods of network congestion, develop more robust methodologies beyond the current exploratory approach based on connection studies, and extend the assessment to distribution networks where relevant.

•

Renewable participation and market response: further refine the assumptions regarding renewable response to market signals, taking into account consultation feedback, observed market behaviour and the assumptions that will be developed in future studies (Adeqflex’27). Such assessments will include an update of renewable response assumptions based on observed market behaviour and the impact of recent market reforms.

•

Assessment of flexibility enablers and barriers: evaluate the effectiveness of ongoing measures aimed at unlocking flexibility across the following dimensions: customer empowerment, price incentives; infrastructure and interoperability, market access and incentives for non-wire alternatives.

•

Relation with network investments: Investigate how to carry out a cost-benefit analysis to justify the use of flexible solutions as an alternative to grid expansion. Such methods shall be mindful to avoid overlap with investment plans.

•

Social welfare assessment: asess the relevance of introducing a quantitative social welfare assessment aimed at identifying the economically efficient balance between additional flexibility deployment, acceptable RES curtailment, renewable response and the associated additional renewable generation capacity. Such an assessment could provide additional support for the definition of indicative flexibility objectives and curtailment targets.

FNA report for Belgium 2026, version approved by CREG

1. Introduction In the last decade, the electricity sector’s attention towards flexibility increased gradually. While adequacy ought to ensure sufficient capacity is available in the system to manage the peak demand, flexibility ensures that the system disposes of the right capabilities to cope with variations and uncertainty of the demand and generation assets. Obviously, the rise of renewable generation with the energy transition increased the importance of flexibility. Elia already pioneered this approach in 2016. Ever since, the flexibility needs assessment has been an integral part of Elia’s biennial Adeqflex study as shown in Figure 6. The study integrated adequacy and flexibility every two years in one study as mandated by the Belgian law. Elia’s AdeqFlex studies follow Article 7bis of the Federal Electricity Law. This resulted already since 2019 in a national adequacy and flexibility study, conducted every two years for Belgium, with a 10 year time horizon. This is conducted in concertation with the NRA (CREG) and in collaboration with the Federal Public Services and the Federal Planning Bureau. Traditionally, Elia consults stakeholders on the methodologies, scenarios and assumptions taken.

Figure 6: History of flexibility needs assessments for Belgium

New European legislation introduced in 2024 results in a stand-alone flexibility needs assessment presented in this report. Elia and DSOs cooperated within Synergrid to develop a plan for the national implementation in Belgium. While awaiting the revision of the Belgian Federal Electricity Law in line with the EU Regulation, Elia was agreed to be designated as entity developing the analysis and writing the report in cooperation with Belgian DSOs Fluvius, Sibelga, ORES, RESA and Arewal. This cooperation was supported by Synergrid. Synergrid represents the network operators that distribute gas and electricity to businesses and the population throughout Belgium The FNA report will be submitted to the CREG for approval after a consultation between July 24 and September 4, 2026. The decision of CREG will be preceded by an opinion of the Federal Public Service Economy (SPF Energy). The next version is awaited in the first part of 2028. The exact planning of the next version will be determined via a revision of Article 7bis of the Federal Electricity Law, currently under development, specifying the national implementation process of the Adeqflex framework. In the future as outlined on Figure 7, and subject to the Federal Electricity Law under development, The adequacy study carried out by Elia and approved by the Minister will serve as a primary basis for the scenarios and target years of the FNA, without precluding the use of additional scenarios or target years where appropriate. The adequacy study would be initiated with consultations on scenarios which are used first in the adequacy study in order to assess the adequacy needs and economic viability of generation. Once the adequacy study is approved by the Minister, the process to establish the FNA starts. The FNA builds on the scenarios and economic dispatch simulations used in the adequacy study, while also allowing for the use of additional scenarios or target years where appropriate and in accordance with Article 7 of the European methodology. The selection of scenarios and target years is agreed with the CREG and the DSOs through an implementation plan. The process will be further specified in the revision of Article 7bis of the Federal Electricity Law, currently under development. FNA report for Belgium 2026, version approved by CREG

Figure 7: Evolution of the AdeqFlex framework (Federal Electricity Law currently under elaboration)

FNA report for Belgium 2026, version approved by CREG

Box 1 - What is new compared to Elia’s previous publications on flexibility? Elia already conducted five flexibility studies as part of the Adequacy and Flexibility studies, in line with Article 7 of the Federal Electricity law. This 6th study follows a new framework set by a European legislation, a European methodology (developed by ENTSO-E, approved by ACER) and new Belgian framework involving cooperation with DSOs and approval by the CREG . Under this new framework, the flexibility study becomes a stand-alone flexibility study, split off from the Adequacy study. The Adequacy study, with scenarios developed by Elia, consulted with the market, and approved by the Minister remains while the Flexibility study (FNA) will follow one year later based on aligned scenarios and time horizons, as well as on the results of the adequacy study, while allowing for the use of additional scenarios or time horizons where relevant. Methods and assumptions related to the FNA will be submitted for approval to the CREG. A new process is implemented based on a co-creation between DSOs, Elia and CREG in which an implementation plan was set up, aligning scope of the methods, scenarios and target years. Compared to the scope of Adeqflex, the scope became more holistic, extending aspects of flexibility studied from focus on short-term needs (for intra-day and balancing markets) to all system needs (including day-ahead markets) but also network needs. The RES integration needs assessment quantifies RES curtailment and compares it with a RES integration target, defined as the maximum level of curtailment compatible with the achievement of national RES targets. In the absence of an explicit RES integration target in Belgium, a range of curtailment levels is assessed through sensitivity analyses. This approach allows evaluating the flexibility needs associated with different acceptable levels of curtailment, as well as alternative cost-effective pathways, including additional flexibility resources or, where relevant, additional renewable capacity. Ramping needs capture expected variations in residual load under perfect forecast conditions, while short-term flexibility needs address unexpected variations, such as forecast errors and forced outages. These indicators complement adequacy assessments by focusing on operational flexibility requirements and are analysed in a consistent framework to avoid overlaps, notably by accounting for additional RES curtailment resulting from ramping and short-term needs. Network flexibility needs are assessed separately and reflect local and time-specific requirements to address congestion and voltage issues on transmission and distribution networks. In this first implementation, the analysis focuses on the available data, notably on RES curtailment at DSO level following constraints at T/DSO interface level, as well as the unavailability of flexible assets on transmission level due to grid constraints. Future developments aim to extend the assessment to a broader set of network constraints, where relevant. Overall, the FNA complements the adequacy assessment by focusing on flexibility needs related to RES integration, residual load variations, short-term uncertainties and network constraints, while ensuring consistency with adequacy assumptions and avoiding overlaps.”.

1.1. Purpose and scope of the national report A. The flexibility needs assessment in relation to adequacy and network studies The FNA complements the national adequacy study prepared by Elia and approved by the Minister, by focusing on flexibility needs beyond generation adequacy, and supports network development by assessing the role of flexibility in managing congestion and optimising investment needs. This integrated approach, illustrated in Figure 8, ensures consistency between generation adequacy, system flexibility, and grid development, while allowing for coherent decision making across different planning instruments and governance levels. At the core of this framework lies the scenario definition, which establishes ex ante assumptions regarding installed generation capacity, demand evolution, grid infrastructure, storage, and associated FNA report for Belgium 2026, version approved by CREG

flexibility across Europe. These scenarios are developed (depending on the study) through a structured co-creation process between TSOs and DSOs, with stakeholders, and are subject to approval by the relevant regional and federal authorities. They provide multiple storylines and sensitivities and build on inputs from existing studies and policy frameworks. Based on these scenarios, the adequacy study (NRAA – National Resource Adequacy Assessment) determines the amount of existing and new generation capacity required to maintain system adequacy, while also assessing the economic viability of this capacity and potential needs for support mechanisms.

Figure 8: Framework of forward-looking electricity system studies

After delivery of the adequacy study, the flexibility needs assessment (FNA) identifies the additional flexibility required to ensure secure system operation, with particular attention to managing renewable energy integration and mitigating both system wide and network related RES curtailment or variability risks. For the FNA, the methodologies and assumptions, including those related to flexibility, are subject to approval by the CREG. As foreseen in European methodologies, the FNA builds on aligned scenarios and findings of the adequacy study (notably future capacity), while allowing for the use of additional scenarios or assumptions where justified, in order to ensure overall consistency. As foreseen in the European methodologies, the FNA builds on the scenarios and findings of the adequacy study (future capacity) to ensure consistency of both studies. In parallel, grid development studies assess additional transmission and distribution grid needs, including reinforcement requirements, and will evaluate their costs and benefits using a total expenditure and cost benefit analysis approach. These assessments consider different analytical drivers depending on the grid voltage level—covering vertical, horizontal, and cross border interactions—and ultimately lead to grid development plans approved by the Minister. All these components form a coherent planning framework that balances investments in grid infrastructure, generation capacity, and flexibility solutions, ensuring a cost efficient and resilient transition of the Belgian power system. B.

The scope of the flexibility needs assessment

The FNA goes further than previous AdeqFlex flexibility studies. Earlier publications mainly focused on short-term flexibility needs and system security—i.e., how much flexibility is needed to keep the system balanced in real-time. The FNA keeps this objective but broadens the lens with two additional aspects: renewable integration (how flexibility can help absorb high renewable output and limit structural excess) and grid constraints (how congestion at transmission and distribution level can restrict flexible resources and lead to renewable curtailment). FNA report for Belgium 2026, version approved by CREG

The renewable integration needs refer to the quantity of flexibility required for the Member State to achieve its annual RES integration target. They assess how well the power system can manage predictable structural excess during periods of very high renewable generation. The analysis quantifies the flexibility needs required to integrate renewables efficiently and achieve the defined renewable energy targets. These results help inform public policy. They can support the case for non-fossil flexibility solutions, such as storage and demand response, and help authorities define flexibility objectives in the National Energy and Climate Plans. At European level, the outcomes of national FNAs are also used by ACER to support further electricity market integration.

For system operators, the system security needs means ensuring the system can cope with large ramps, unexpected outages and forecasting errors, especially during periods of excess generation. In this respect, the FNA complements the Adequacy Study by quantifying the flexibility needs required to ensure the secure operation of the system, in particular under high levels of renewable generation. The analysis focuses on assessing the volume and characteristics of flexibility required to maintain real-time system balance.

The FNA assesses grid constraints and the impact of congestion on transmission and distribution networks during periods of excess generation. It quantifies the effects of these constraints on the availability of flexibility and the level of renewable curtailment. Initial quantitative results are provided by Elia and the DSOs, with methodologies to be further expanded in the next FNA in 2028. This objective underpins—and reinforces—the first two. C. The objective of flexibility needs assessment The FNA aims to assess the overall flexibility needs of the electricity system in a technology-neutral manner, in accordance with the European legislation and the FNA methodology established by DSO Entity and ENTSO-E and approved by ACER. Its results are intended to support Member States in defining indicative national objectives for non-fossil flexibility. FNA report for Belgium 2026, version approved by CREG

In the absence of an explicit RES integration target in Belgium, in particular regarding acceptable levels of RES curtailment, the assessment is carried out using a range of curtailment levels through sensitivity analyses. This approach allows estimating the additional flexibility needs associated with different levels of acceptable RES curtailment. In this context, the analysis makes it possible to derive: • the additional non-fossil fuel flexibility required to achieve full RES integration, implying a zero curtailment; • and, conversely, the level of RES curtailment that would be required in the absence of additional flexibility beyond what is already expected in the system. In this case, additional RES generation capacity must be installed to reach the NECP target. The level of RES curtailment can be assessed as level of energy (on yearly basis) and in capacity (on instantaneous basis). Based on these results, the assessment can also provide insights into the extent to which the existing and expected response of renewable assets is sufficient to manage RES curtailment in the absence of additional flexibility, or whether additional RES response may need to be unlocked. The choice between investing in additional flexibility resources or increasing the controllability of renewable generation falls within the scope of public policy. The FNA does not determine such choices and does not assess the relative economic efficiency of these options. It provides a transparent and consistent basis for assessing flexibility needs, which can then be used to inform decision-making. In addition, the FNA identifies potential barriers to the development and use of flexibility, such as regulatory, market or technical constraints, thereby complementing the assessment of flexibility needs and supporting the effective deployment of flexibility solutions. Figure 9 illustrates how system flexibility needs resulting from the three building blocks of the FNA are translated into additional flexibility needs through the analysis of RES curtailment. The additional needs are calculated after accounting all flexibility expected to be delivered through the market and the capacity remuneration mechanism. In the absence of explicit curtailment targets in the Belgian NECP, different levels of curtailment are assessed, reflecting two extremes: the additional non-fossil fuel flexibility needs for full RES integration (no curtailment) and the RES curtailment in the absence of additional non-fossil fuel flexibility.

Figure 9: Expected output of the FNA

This analysis is conducted by iteratively adding additional non-fossil fuel flexibility to the system until RES curtailment is covered. As the Belgian NECP does not specify explicit curtailment thresholds, the analysis evaluates flexibility needs across a range of curtailment levels up to zero curtailment. This FNA report for Belgium 2026, version approved by CREG

approach enables an identification of the relationship between acceptable RES curtailment and the corresponding non-fossil flexibility capacity which can be used for setting national objectives on nonfossil fuel flexibility. Figure 9 further illustrates how RES curtailment in absence of additional non-fossil fuel flexibility is resulting in additional RES response needs. These additional needs are calculated after accounting expected RES controllability in the market based on current trends. To operationalise this approach, the assessment applies a reliability criterion for covering RES curtailment. These curtailment levels can further be analysed in relation to the level of additional flexibility deployed, by assessing the maximum renewable capacity that must become flexible to manage this RES curtailment in view of operational security. By comparing these requirements with the existing and expected levels of RES response in the scenarios, the analysis provides a quantified view of the additional RES response to be unlocked on renewable assets.

1.2. Legal and regulatory context The European legislative framework introduced in 2024 establishes a dedicated, stand-alone flexibility needs assessment. Article 19(e) of Regulation (EU) 2024/1747 requires Member States to conduct this assessment based on a standardized methodology that: • requires every two years a report on the estimated needs for flexibility for a period of at least the next 5 to 10 years at national level; • shall be used by Member States to define indicative national objectives for non-fossil fuel flexibility and can be used to justify non-fossil flexibility support schemes; • is subject to cooperation between TSOs (for system needs, network needs) DSOs (for network needs) and NRA (for approval). In line with Article 19(e), ENTSO-E and the EU DSO Entity developed the common methodology and data formats, approved by ACER on 25 July 2025. Building on this framework, Elia and Synergrid defined the national implementation approach for Belgium, formalized in an implementation plan approved by CREG on 11 December 2025. While awaiting the revision of the Belgian Federal Electricity Law in line with the EU Regulation, Elia has been designated as Designated Entity responsible for carrying out the analysis and preparing this report, in cooperation with Belgian DSOs Fluvius, Sibelga, ORES, RESA and Arewal. This cooperation is supported by Synergrid. The Belgian Federal Electricity Law is currently still under elaboration and will specify the process of the implementation of the FNA in Belgium.

Box 2 - Relevant documents and references • •

• •

European Legal framework: Regulation (EU) 2019/943 : https://eurlex.europa.eu/eli/reg/2019/943/oj/eng European Regulatory framework: FNA method approved by ACER : https://www.acer.europa.eu/news/acer-approves-eu-wide-methodology-assess-national-electricityflexibility-needs Belgian Implementation plan approved by CREG : https://www.creg.be/nl/publicaties/beslissingb3152 Assumptions and methods for the Belgian FNA approved by CREG : https://www.creg.be/nl/openbare-raadplegingen/prd3184

FNA report for Belgium 2026, version approved by CREG

1.3. Governance: bodies involved in the FNA report This report is drafted by Elia, designated as the Designated Entity. This report is based on a system needs analyses conducted by Elia and a network needs analyses conducted by Elia and all distribution system operators. Section 7.2 (Network needs on T/DSO connection points) and Section 7.3 (DSO network needs) was drafted with DSOs based on the analyses conducted by DSOs while Section 9 on barriers has been developed in concertation with DSOs. As Designated Entity, Elia consolidated the TSO and DSO calculations and prepared this FNA report in concertation with the DSOs. This cooperation is supported by Synergrid. Elia, DSOs and Synergrid prepared an Implementation Plan detailing the high-level scope and methodology, focusing on the implementation decisions taken in the context of the FNA Methodology and on the expected inputs, outputs and timeline of the study. This Implementation Plan was approved by CREG on December 11, 2025 and presented to stakeholders in a dedicated workshop on February 13, 2026. In a later step, a detailed national methodology for the analyses to be conducted by Elia was developed, detailing the calculation and analysis steps for this report. This methodology was consulted by CREG from March 20, 2026 to March 30, before it was approved by CREG. This methodology is described in Annexes A, B and C of this FNA report. Through a structured cooperation in a Synergrid working group, Elia and DSOs agreed on the methodologies to calculate the DSO network needs and DSOs delivered the analyses. The methodology is described in Annex D of this FNA report. The results are elaborated in Section 7. DSOs aligned where needed with the regional regulators. This FNA report was presented to stakeholders in a dedicated workshop organized by Elia, Synergrid and DSOs on July 17, 2026. It was agreed between Elia, the DSOs, and CREG to launch a public consultation on 24 July 2026, closing on 4 September 2026. The report will then be updated and submitted to CREG for approval. An opinion from FPS Economy – Directorate-General for Energy will also be provided to the CREG on the same date.

1.4. Key definitions This section defines the key concepts used in the FNA report. Definitions follow the Electricity Regulation and the FNA Methodology. Flexibility “Flexibility” means the ability of an electricity system to adjust to the variability of generation and consumption patterns and to grid availability, across relevant market timeframes. This legal definition was introduced by the Electricity Market Reform (Regulation (EU) 2024/1747). As shown Figure 10, power systems and markets need flexibility to cope with three types of uncertainty (also known as ‘flexibility drivers’): • the variability and uncertainty of demand, as it is not possible to know ex-ante what the exact electricity demand will be in real-time, since it depends on external variables such as consumer preferences and weather conditions; • the variability and uncertainty of renewable and distributed generation, as renewable generation such as wind and solar power, as well as other highly distributed generation sources such as combined heat and power or run-of-river hydroelectricity, are characterised by uncertainty, since they are subject to variable and uncertain weather conditions; and • unexpected generation unit or transmission asset outages as forced outages are an inherent characteristic of generation and transmission systems and are, by definition, unpredictable. These events result in a sudden loss (or excess) of power. FNA report for Belgium 2026, version approved by CREG

Figure 10: Drivers and sources of flexibility

In order to keep the system in balance, which is a fundamental prerequisite for system security, these expected and unexpected variations in demand and generation must be covered at all times through the use of flexibility sources, also referred to as the ‘flexibility sources or means’ of the system. Flexibility means are delivered by technologies which are controllable: they can alter their generation or demand upon request in a relatively short timeframe. Relevant categories of such technologies include: • generation units, since most conventional thermal units can modify their output within a certain timeframe; • demand side assets, since some of these can provide flexibility by modifying their demand following a reaction to explicit signals, or (implicit) price signals; • electricity storage, since this technology is generally very flexible and is characterised by an ‘energy’ reservoir; and ; • interconnectors, which can import (or export) flexibility from/to other regions by means of crossborder forward, intra-day/day-ahead, or balancing markets. Key terms are given for the individual building blocks identified in this section. These definitions are introduced by the FNA methodology.

Downward flexibility needs

needs whose solution requires either decreasing injection into the network or increasing demand from the network;

Upward flexibility needs

needs whose solution requires either increasing injection into the network or decreasing demand from the network

Uncovered flexibility needs

additional flexibility needs identified by the FNA methodology, not expected to be met by new or existing resources based on ERAA or NRAA inputs

FNA report for Belgium 2026, version approved by CREG

A. RES integration needs

RES integration needs

Quantity of flexibility required for the Member State to achieve its annual RES integration target

Residual load

Total demand minus non-dispatchable generation (as individual units or as aggregated units), such as from wind or solar as well as generation, subject to must run conditions per market time unit

RES curtailment

Limiting the generation or transmission of renewable power for system operational reasons or grid-capacity reasons

RES integration target

RES target

Uncovered RES integration needs

The maximum RES curtailment amount or share compatible with the RES target for the Member State. It is expressed either as a maximum RES curtailment in absolute terms, or as a maximum RES curtailment in relative terms to the total RES generation or to the total electricity demand, or symmetrically as the minimum effectively integrated (i.e., non-curtailed) RES generation in absolute terms or in relative terms to the total RES generation or to the total electricity demand

National target for RES in the electricity sector of the most recent NECP or from another relevant official national document consistent with the most recent NECP. It is either expressed as a RES volume in absolute terms or as a share of RES production in relative terms to the total electricity demand

Amount of RES curtailment exceeding the maximum level of RES curtailment compatible with the RES integration target

B. Short-term and ramping needs

Ramping flexibility

Short-term needs

Needs associated with variations of the residual load assuming perfect forecast conditions

Needs associated with unexpected variations of the residual load such as forced outage of assets during the intra-day or balancing timeframe

Very fast flexibility

Capacity able to react to aggregated D-1 or shorter horizons forecast error variations (5 to 15 minutes before real-time)

Fast flexibility

Capacity able to react to aggregated D-1 or shorter error variations (15 to 60 minutes before real-time)

Slow flexibility

Capacity able to react to aggregated D-1 or shorter horizons forecast error (60 minutes to 24 hours before real-time)

FNA report for Belgium 2026, version approved by CREG

C. Grid constraints

Network flexibility needs

Voltage issue

Congestion issue

Flexibility needed to adjust for grid availability, by means of preventing or solving congestion or voltage issues, across relevant timeframe

Situation when voltage levels are not within the operational limits

Means a situation in which the electric current flow through a physical asset exceeds operational limits

FNA report for Belgium 2026, version approved by CREG

2. Methodological aspects This section describes the methodological framework used to perform the flexibility needs assessment, in line with the FNA Methodology. It begins by specifying the data sources used, including the selected report as the primary input dataset. The target years considered for simulations are indicated next. The modelling approach details the analytical methods employed for the flexibility needs calculation and finally any additional data or analyses adopted. Together, these elements constitute the methodological framework to evaluate the national flexibility needs. The methodologies are described in detail in four Annexes. Three annexes on system needs were implemented by Elia in this report and were submitted to CREG for approval on March 10, 2026. These annexes were subject to consultation and assessment by CREG prior to their approval. • Annex A: RES integration needs • Annex B: Short-term needs and ramping needs • Annex C: Unavailability of flexible resources, finetuning and guiding criteria A fourth annex has been drafted by DSOs in concertation with Elia in the run up to this report. • Annex D: DSO network needs

2.1. Data sources A. System needs In line with the FNA methodology and as agreed in the implementation plan, Elia will conduct the system needs as a post-processing of the Economic Dispatch (ED) results from Adeqflex’25 (NRAA). The scenarios ‘Current Commitments’ (EU-BASE-CC) and ‘Constrained Transition’ (EU-BASE-CT) are used as the basis. The scenario assumptions are further elaborated in Section 3. The system needs are assessed under two configurations reflecting different assumptions on the availability of flexibility resources. • A first configuration, agreed in the implementation plan, was based on a “no gap-filling” approach, where flexibility resources are limited to economically viable units and those already contracted through CRM auctions prior to 2025, as considered in Adeqflex’25. All existing thermal capacities are assumed to remain available, in line with the same source. In this lower bound, this configuration does not reflect more recent developments in contracted capacity or expected additional flexibility resources emerging from adequacy mechanisms and market-based investments. • A second configuration reflects an extended availability of flexibility resources, where both the “Current Commitments” (EU‑BASE‑CC) and “Constrained Transition” (EU‑BASE‑CT) scenarios are adapted to account for the large-scale battery potential identified in Adeqflex’25 for the target years 2030 and 2035, as further described in Section 3. This can be considered a higher bound on large-scale batteries. Together, these configurations frame the range of flexibility resources considered in the assessment. In line with the objective of the FNA, the analysis aims to identify flexibility needs that are not expected to be covered by existing or anticipated market-based outcomes, including adequacy mechanisms such as the CRM. Therefore, four scenarios are assessed for both 2030 and 2035. Additionally, given other uncertainties and their impact on the flexibility needs (e.g. during moments of incompressibility), some of the analyses are complemented with additional sensitivities: • More existing nuclear capacity (3 or 4 GW instead of 2 GW) given the ongoing discussions on potential extensions of the existing nuclear fleet in Belgium; FNA report for Belgium 2026, version approved by CREG

• Less ‘end user flexibility’ on residential assets (EV, heat pumps and small-scale batteries) compared to the baseline assumptions as some segments are shown to be developing more slowly than expected. This sensitivity was also defined in Adeqflex’25 and used as sensitivity analysis for adequacy results. B. Network needs Network needs are in this first version of the FNA based on four types of needs: o

TSO network needs for which all analysis out of scope for FNA 2026. An implementation plan for the FNA 2028 is presented by Elia in Section 7.2.

Network needs on T/DSO connection point for which a quantitative analysis is conducted by DSOs on RES curtailment on distribution level following constraints on the TSO transformer at the T/DS connection point. This analysis is conducted under Article 11 of the FNA methodology. Data and analyses provided by DSOs based on their DNDPs, i.e. latest distribution network development plans (investment plans) approved where available. An overview of the main assumptions on renewable capacity and flexibility is given in Section 3 and Annex D.

DSO network needs for which a qualitative analysis is conducted by DSOs on voltage and internal congestions. Quantitative analysis out of scope for the FNA 2026.

Unavailability of flexible resources for which a quantitative analysis is conducted by Elia on unavailability of TS-connected batteries during high-RES conditions. This is based on information available in flexible connection agreements made available in connection studies.

2.2. Target years considered Economic dispatch simulations are performed for the target years 2030 and 2035, where reference year 2030 includes September ’29 – August ’30 and 2035 includes September ’34 – August ‘35. As agreed in the implementation plan of this study, these are considered best aligned with objectives and requirements of the FNA: • At least two years in the first version of the FNA • At least one EU policy year (2030) • Aligned with ERAA target years

2.3. Modelling approach The FNA report assesses the ability of the Belgian power system to manage increasing shares of renewable energy and growing volumes of excess energy by applying the Flexibility Needs Assessment (FNA) framework to the results of the economic dispatch simulations used for the Adeqflex’25 study. The methodology itself is fully described in the four method appendices and follows the framework of the FNA method approved by ACER. The three annexes on system needs methodologies have already been approved by CREG ahead of this report. This section therefore focuses on briefly explaining how that methodology is applied in practice and how the results are interpreted to characterize system flexibility needs. As shown in Figure 11, the study methodology is structured around the three complementary building blocks that together assess the system’s ability to manage increasing shares of renewable energy and growing volumes of excess energy. These building blocks focus on renewable integration, system security and grid constraints, and jointly provide a comprehensive view of the flexibility challenges faced by the Belgian power system under future scenarios.

FNA report for Belgium 2026, version approved by CREG

Figure 11: Overview of FNA methodology and outputs

The renewable integration building block assesses the ability of the system to manage structurally predictable periods of excess renewable generation. In a first step, RES integration needs are assessed based on indicators derived from the economic dispatch simulations performed in Adeqflex’25. In particular, residual load (i.e. demand net of nondispatchable renewable generation) and RES curtailment are computed from the simulation outputs to capture, respectively, the net demand to be covered by dispatchable resources and the volume of renewable generation that cannot be integrated under the simulated system conditions. In the absence of an explicit RES integration (curtailment) target in Belgium, the assessment is conducted across a range of curtailment levels. This allows analyzing how different acceptable levels of RES curtailment translate into different flexibility needs. In a second step, an optimization model is used where the residual load and curtailment indicators are used to determine how much additional flexibility of each type would be needed in the system to reach a certain level of renewable integration in the system. This can be set off against accepting an amount of renewable curtailment, and the additional renewable capacity to be developed to compensate for this lost renewable energy. In second instance, an optimization model is used where the residual load and curtailment indicators are used to determine how much additional flexibility of each type would be needed in the system to reach a certain level of renewable integration in the system. This can be set off against accepting an amount of renewable curtailment, or the additional renewable capacity to be developed to compensate for this lost renewable energy. The system security building block evaluates the system’s robustness against variability and uncertainty. It focuses on the ability of the system to manage periods with excess energy, large residual load ramps as well as unexpected deviations caused by forecast errors and forced outages. In first instance, the short-term flexibility needs are assessed through a statistical analysis of historical forecast errors and forced outages, extrapolated towards 2030 and 2035 based on installed capacity. In second instance, the short-term means are assessed by calculating the residual margins on generation, storage and demand assets after the economic dispatch simulations used in Adeqflex’25. This allows to identify the uncovered short-term flexibility needs. The grid constraints building block accounts for the fact that flexibility is not always fully available due to network limitations. It assesses the system’s ability to manage congestion and identifies situations in which flexible resources are unavailable because of grid constraints, prequalification requirements or temporary limits. These limitations may result in additional renewable curtailment or reduced effectiveness of flexibility, even when sufficient theoretical flexibility capacity exists. The RES curtailments in the distribution grid are assessed through a comparison of injection profiles in 2030 and 2035 from the distribution grid in the transmission grid, against limits on the TSO transformer on the coupling point between Elia and the DSOs. The unavailability of flexible resources is assessed based on an aggregation of information captured in issued connection studies. Note that these methods will be further developed towards the next FNA.

FNA report for Belgium 2026, version approved by CREG

The FNA reports are intended to support Member States in defining indicative national objectives for non-fossil flexibility. To this end, the assessment integrates the three building blocks into a single, coherent framework, leading to the identification of overall flexibility needs. In the absence of an explicit RES integration (curtailment) target in Belgium, the assessment is conducted across a range of curtailment levels, reflecting different possible levels of renewable integration. This approach captures the trade-off between two extreme situations: on the one hand, full integration of renewable energy (i.e. zero curtailment), requiring additional non-fossil flexibility; and on the other hand, a situation with no additional non-fossil flexibility, leading to a corresponding level of RES curtailment. On this basis, the framework allows the quantification of: 1. Non-fossil flexibility needs, corresponding to the amount of additional flexibility required to achieve full renewable integration, i.e. to avoid curtailment; 2. Renewable flexibility needs, derived from the situation without additional non-fossil flexibility. In this case, the resulting level of curtailment defines the required contribution from renewable generation. By comparing this requirement with the flexibility already available or expected from renewable generation, the additional renewable flexibility needs can be determined. These two extreme situations define the boundaries of the assessment and illustrate the trade-off between deploying additional non-fossil flexibility and accepting higher levels of curtailment. On this basis, the framework also allows estimating the additional renewable capacity required to compensate for curtailed energy. Figure 11 illustrates already how flexibility needs are derived from system flexibility needs identified across the different building blocks. In the absence of an explicit curtailment target, the analysis is structured around two extreme situations: on the one hand, full renewable integration (i.e. zero curtailment), which requires additional non-fossil flexibility; and on the other hand, a situation without additional non-fossil flexibility, leading to a corresponding level of RES curtailment. On this basis, the analysis allows to quantify the non-fossil flexibility needs required to achieve full integration, as well as the contribution required from renewable generation in the absence of additional flexibility, and the associated need for additional renewable capacity. As the Belgian NECP does not specify explicit curtailment thresholds, the analysis also evaluates intermediate curtailment levels between these two boundary cases. This makes it possible to identify the relationship between acceptable RES curtailment levels and the corresponding non-fossil flexibility needs, thereby providing input for the setting of indicative national objectives for non-fossil flexibility In the case where no additional non-fossil flexibility is deployed, the resulting level of RES curtailment provides a reference for assessing the contribution required from renewable generation. By taking into account the renewable flexibility already available or expected in the scenarios, the additional renewable flexibility needs can be determined. The same curtailment level also allows estimating the additional renewable capacity that would be required to compensate for curtailed energy. Weather scenarios considered and climate years that have been used for the simulations (as per FNA Methodology, Article 6(5)). As agreed in the implementation plan, Elia conducts the FNA method based on 25 ‘Monte Carlo’ years clustering in line with the short-term flexibility assessments and economic assessment in Adeqflex’25. The 25 Monte Carlo years are derived from a clustering of 199 climate years. Conducting a full analysis on 199 climate years is deemed unnecessary and inefficient. The selected 25-year clustering strikes a balance between analytical precision and resource efficiency, ensuring robust and reliable results. Additional insights on clustering can be found in Appendix G.2 of Adeqflex’25. The justification of the number of Monte Carlo years is elaborated in the Annex A. • The analysis relies on the approach from past AdeqFlex studies by using the forward-looking climate database developed by Météo-France, which is fully aligned with the ERAA methodology. FNA report for Belgium 2026, version approved by CREG

• A reduced set of Monte Carlo years is used for the economic simulations and economic viability assessment (EVA), consistent with the approach applied in Adeqflex’25 A clustering is performed in order to reduce the amount of output to be analysed while keeping the same weights of the analysed parameters and indicators. The resulting hourly time series, covering 25 Monte Carlo years, form the basis for flexibility analyses with a focus on excess energy, ramping requirements and forecast-driven deviations. By relying on the same simulation outputs, the flexibility assessment remains fully aligned with the adequacy study and ensures that conclusions on system needs are coherent across different dimensions of system operation. Specification of use of post-processing or Economic Dispatch modelling (as per FNA Methodology, Article 7(3)). The assessment relies as agreed in the implementation plan on pre-existing models in Adeqflex’25 with post-processing adjustments on two scenarios (CC and CT). In addition, dedicated economic dispatch simulations were performed on two complementary scenarios reflecting the large-scale battery potential identified in Adeqflex’25 (one for CC and one for CT), extending the baseline assumptions on storage deployment. In the baseline scenarios, agreed in the implementation plan, storage capacities are limited to existing installations and capacities contracted prior to 2025. As such, they do not fully reflect additional contracted volumes and expected future developments resulting from market dynamics and adequacy mechanisms such as the latest CRM 2025 auctions and all future CRM auctions. The complementary scenarios are designed to capture this identified battery potential and to assess its impact on system flexibility. This allows the analysis to assess a scenario that maximizes the potential flexibility that could be developed by the market. This extension of the scenario space does not require methodological changes to the economic dispatch modelling framework. Approach to compute additional flexibility resources to meet the RES integration target, if applicable (as per FNA Methodology, Article 8). The methodology is elaborated in detail in Annex A.3. Additional flexibility resources are quantified based on the uncovered RES integration needs, i.e. the RES curtailment exceeding the selected RES integration target. A technology-neutral optimisation determines the minimum flexibility portfolio required to reduce curtailment to the target level, while accounting for interactions between flexibility durations and the temporal patterns of curtailment. Flexibility is represented through generic storage-like resources characterised by their power capacity (MW), energy capacity (MWh), energy-to-power ratio (duration) and round-trip efficiency. The assessed duration categories are 1–8h (short duration), 20–50h (medium duration) and 100–200h (long duration). No explicit response time is assumed. The optimisation is performed on an hourly basis. Consideration of cross-border flexibility, (as per Electricity Regulation, Article 19e(1)). When assessing the RES integration needs, cross-border flexibility is accounted for throughout the assessment of the system needs. In a first instance, the economic dispatch simulations are conducted for Belgium in an interconnected European system. It thus accounts for available generation, demand and storage in other regions, as well as network constraints represented trough a flow-based model. Local grid constraints are not accounted. These assumptions are elaborated in Adeqflex’25. When assessing the contributions of cross-border flexibility for the assessment of ramping needs. It is assumed that sufficient liquidity is generally available in the European system to cover perfectly predicted ramps. Available cross-border flexibility is regarded as fully available except during periods of market tightness (incompressibility and scarcity). Scarcity is identified using the demand shedding (LOLE), while incompressibility is determined based on the RES curtailment indicator described in Section 4.2. During such periods, the potential for constrained cross-border flexibility is considered, reflecting reduced market liquidity.

FNA report for Belgium 2026, version approved by CREG

Short-term needs account for cross‑border flexibility through the additional import and export potential available on top of the scheduled exchanges resulting from economic dispatch simulations. This potential is defined as the difference between these scheduled exchanges and the theoretical maximum import and export capacities. A first step establishes this theoretical flexibility margin, using maximum exchange levels of 7,500 MW for imports and 8,000 MW for exports, derived from market modelling results. In a second step, liquidity constraints are assessed. An analysis on historical observations of available energy bids originating from cross-border resources in intraday and balancing platforms are found to be sufficiently high, meaning that liquidity does not materially limit cross‑border flexibility in the modelling. In contrast, available transmission capacity available for intraday and balancing is found to be the limiting factor for cross‑border flexibility. Capacity constraints are differentiated by flexibility type and market conditions, using observed values under normal and extreme price situations (as defined in the method found in Annex B.2). In practice, this translates into approximately 2,500 MW of import capacity and 900 MW of export capacity under normal price conditions being available for fast cross‑border flexibility following network constraints. However, during periods of extreme prices (1% lowest for export and 99% highest for import), the available volumes are significantly reduced. During such moments, available transmission capacity for balancing platforms can reach below 50 MW (upward flexibility) or 300 MW (downward flexibility) and can remain at zero for intra-day markets. In the third step, contractual reserve sharing arrangements are taken into account. For fast flexibility, minimum exchange volumes of 250 MW (import) and 350 MW (export) are accounted in the calculations. However, note that these capacities are in practice not guaranteed. Approach for the estimation of the 0.1 and 99.9 percentile of the probability distribution of the residual load forecast errors (as per FNA Methodology, Article 10(1)). The short-term flexibility needs are assessed following the expected forecast errors of demand and generation between day-ahead and real-time. These are dimensioned on the largest forecast errors which might occur for each target year 2030 and 2035. The methodology is described in Detail in Annex B.1. The calculation starts from an aggregation of the forecast errors of the demand, offshore and onshore wind generation, solar generation and must run generation in a residual load forecast error. The largest errors are then represented by the 0.1 and 99.9 percentiles of the probability distribution of this residual load forecast errors. The distribution follows an extrapolation of a time series of historical Residual load forecast errors accounting the projected evolution of the load and installed capacity of variable renewable generation, and forecast improvements. A. Time horizon considered The flexibility needs assessment is based on a categorization of three types of flexibility, which are derived from the timeframes during which new information is received by market players. This information may relate to forecast updates or information concerning the unexpected unavailability of a generation or transmission asset. The assessment distinguishes between slow, fast and very fast flexibility as shown in Figure 12 and as outlined below: • Slow flexibility up to 5 hours ahead represents the ability to deal with expected deviations in demand and generation following the intraday forecast update. It concerns information received between the day-ahead market (up to 36 hours before real-time) and the intraday forecast received several hours before real-time, depending on the forecast service. Additionally, this flexibility deals with power plant or transmission asset outages which are announced several hours before real-time (or still not resolved after several hours). This flexibility can be provided

FNA report for Belgium 2026, version approved by CREG

with most of the installed capacity, as there are several hours to change the output of a generation, storage or demand unit and even start or stop a power plant. • Fast flexibility up to 15 minutes ahead represents the ability to deal with unexpected power deviations in real-time, or deviations for which information is received between the last intraday forecast and real-time. It concerns information received between several hours up to a few minutes before real-time, depending on the forecast service. Additionally, this flexibility type needs to deal with forced outages up to several hours until the providers of slow flexibility can take over generation units which are already dispatched and able to modify their output program within a few minutes, or through units which have start or stop time of a few minutes, as well as storage units (pumped-hydro and batteries) and types of demand side management which are considered very flexible • Very fast flexibility up to 5 minute variations, represents the ability to deal with real-time variations in the forecast error and in particular the forecast errors of the last intraday forecast before real-time. It can be expressed as the capacity required for up to 5 minutes, or even per minute (MW/min). This type of flexibility does not cover forced outages which are assumed to be covered by FCR, and relieved by fast and slow flexibility. Very fast flexibility is to be covered by assets which can follow forecast error variations on a minute-by-minute basis and therefore only those units which are already dispatched, as well as some battery storage and demand side management units which are considered very flexible.

Figure 12 : Short-term flexibility needs framework

While very fast reserves can be seen as a subset of the fast reserves, slow flexibility—covering intraday forecast updates acts upstream by reducing forecast errors before real-time. Fast reserves then address the remaining residual forecast errors closer to real-time. As slow and fast flexibility needs are not entirely cumulative (as intra-day forecast updates generally reduce the final forecast error), the total short-term flexibility needs is calculated based on forecast errors between day-ahead and real-time. B. Data sources The assessment is based on historical forecasts of the total load and renewable generation and are based on the results of forecasting tools which are published on a real-time basis on Elia’s website. Although the flexibility needs of the system are driven by the predictions and operational decisions of market players, this forecast data is assumed to be representative of the tools which are used by market players.

FNA report for Belgium 2026, version approved by CREG

In addition, synthetic errors are constructed of the forced outages on generation and transmission asset impacting fast and slow flexibility needs. The probability distribution is based on a time series generated with a ‘Monte Carlo’ simulation, taking into account the generation fleet and relevant HVDC interconnectors for the year for which the simulation is conducted in accordance with the maximum generation or transmission capacity and outage probability of relevant generating units and interconnectors for the relevant target years. The number of forced outages per year are elaborated in Adeqflex’25 (assumption workbook tab 4.3) and shown in Table 4. Further information on the modelling of the forced outages is provided in Annex B.1.4. Table 4: Number of forced outages per year based on Adeqflex’25 assumptions

Category Nuclear CCGT OCGT TJ CHP, waste, biomass Pumped Storage DC links

Number of FO per year 1.2 9.4 3.3 3.4 3.0 7.0 1.8

C. Projection to target years Note that all historic time series values are expressed as a percentage of the monitored capacity (the demand is expressed in terms of the average demand, the renewables and must run generation in terms of installed capacity). This to extrapolate the time series towards projected values in 2030 and 2035. This extrapolation is conducted by means of the installed capacity and demand scenarios for the selected scenarios and target years. The total load, real-time wind and solar generation, as well as other distributed generation forecasts from the selected historical years, are adjusted to reflect expected forecast accuracy improvements for the target years. An average cumulative improvement factor of 1% per year, applied from 2025 onwards, is incorporated, same as the assumptions used in previous studies published by Elia. The forecast improvement value is justified by a scenario of incremental improvements of ‘technology mature’ forecasting tools without large new breakthroughs. To understand the impact of this assumption, a sensitivity analysis without improvement factor, as well with a higher improvement factor, on the resulting short-term flexibility needs will be provided. In addition to this forward-looking adjustment, a backward correction is applied to the historical dataset to account for the observed improvements in forecast accuracy between 2021, 2022, and 2023, relative to 2024 and 2025. The accounted improvement is based on the yearly average observed mean absolute error expressed as a percentage of installed capacity. This ensures that the historical time series are normalized to the current forecast performance level before projecting further improvements towards the future target years. Consideration of network needs (as per FNA Methodology, Article 11 and Article 12). Network needs are in this first version of the FNA based on three types of needs: • TSO network needs for which all analysis out of scope for FNA 2026. An implementation plan for the FNA 2028 is presented in Section 7.2. • Network needs on T/DSO connection point for which a quantitative analysis is conducted on RES curtailment on distribution level following constraints on the TSO transformer at the T/DS connection point. This analysis is conducted under Article 11 of the FNA methodology. Data and analyses provided by DSOs based on their DNDPs, i.e. latest distribution network development FNA report for Belgium 2026, version approved by CREG

plans (investment plans) approved where available. An overview of the main assumptions on renewable capacity and flexibility is given in Section 3 and Annex D. • DSO network needs for which a qualitative analysis is conducted on voltage and internal congestions. Quantitative analysis out of scope for the FNA 2026) Additionally, an analysis is conducted on the unavailability of flexible resources on TSO-level following grid constraints which is based on information available in flexible connection agreements made available in connection studies. The quantitative analysis on network needs on T/DSO connection point focuses on quantifying the flexibility needs for preventing congestion at the TSO/DSO coupling point. The computation elaborated in Annex D considers: • a spatial resolution per DSO area (Fluvius, ORES, RESA, Sibelga & Arewal), • TSO/DSO Coupling points, where the lower voltage of the transformer is coupled to a 5 kV to 63 kV DSO network, • an hourly temporal granularity (7 representative days with each 24 values). The representative days reflect different scenarios in terms of RES generation (low RES, average RES, high wind, high solar) and demand (high, low, average). These representative days are defined with the same RES/load thresholds for all DSOs so all network needs (represented by RES curtailment) can be aligned with Elia’s economic dispatch simulation. The RES/load thresholds are chosen to obtain good results for extreme scenarios on one hand (power), and allow for a representative calculation of the average annual needs (energy) on the other hand. To allow for a representative combination of coupling point results, where installed solar and wind can be very different, separate representative days were introduced for high-solar and high-wind. Each DSO was responsible for using representative scenarios for local generation and load scenarios, in line with their DNDP. The RES curtailment at distribution level is reported as 24 hourly values for each representative day and for each TSO/DSO coupling point. The maximum power and total energy needs are respectively summed per representative day. The total yearly energy needs are obtained via the weighed sum of each representative day’s energy needs with its occurrence in an average Monte Carlo year. The TSO network needs are not implemented in this first version of the FNA. Section 7.1 presents an implementation plan for this voluntary task to study the RES curtailment following congestions in the transmission grid, referred to as TSO network needs. The implementation plan is for the next version of this report foreseen in 2028. As agreed in the implementation plan, the timing of this first version did not allow to develop the specific methods in line with the FNA method. Note that a specific assessment of TSO network constraints is conducted through the assessment of unavailability of flexible resources following grid constraints. In this assessment, the unavailability of transmission system connected battery assets during high renewable conditions is assessed. This is intended to allow a potential finetuning of the results if it would be shown that these assets are regularly unavailable when needed the most. This assessment is conducted based on an analysis of all flexible connection agreements issued by Elia. Finetuning approach (as per FNA Methodology, Article 14). While the TSO Network Needs were not quantified in this first FNA and cannot be taken into account in the finetuning, the Network needs on T/DSO connection point are found to be an order of magnitude smaller than the RES integration needs when calculated solely on system level and remain very limited in terms of annual renewable generation. The impact of the finetuning is presented in Section 8. The unavailability of flexible resources following grid constraints, calculated for transmission system level connected battery assets during high renewable conditions in Section 5.3, was also found to be small and far below the finetuning threshold. These are therefore not further considered in the finetuning FNA report for Belgium 2026, version approved by CREG

of system needs. No assessment is made on unavailabilities of flexible resources following distribution grid constraints. Finetuning has been conducted within the following system needs: • Ramping needs are accounted in the RES integration needs. To avoid double counting of needs, the maximum of RES curtailment and downward ramping needs in each hour is considered. This way, when a downward ramping need occurs at the same time as a RES integration needs, it is accounted that any flexibility addressing such RES integration need will automatically also cover the downward ramping need or vice versa. • Short-term needs are accounted in the RES integration. Here, the downward short-term needs are added to obtain a fine-tuned RES Integration need because RES curtailment to cover excess forecast errors are considered to come on top of RES curtailment following the well-predicted excess generation of the RES integration needs.

FNA report for Belgium 2026, version approved by CREG

3. Scenarios and assumptions description This Section outlines the scenarios, targets, and key assumptions underpinning the flexibility needs assessment, building upon the methodological framework described in Section 2. While Section 2 focuses on how the assessment is conducted—covering data sources, modelling approaches, and analytical techniques—this section specifies what inputs are applied to those methods. It details the scenarios used for system and network needs, the RES integration targets, and the critical assumptions that shape the analysis. Together, these elements ensure transparency in the baseline conditions and the factors influencing the assessment outcomes.

3.1. Scenarios used to assess system and network needs 3.1.1. Scenarios for system needs As agreed in the implementation plan, the analysis is built on the scenario framework from the Belgian NRAA, Elia’s Adequacy & Flexibility Study 2025 (further referred as Adeqflex’25). This approach, shown in Figure 13, ensures consistency between the adequacy and flexibility assessments through the use of common scenarios and underlying assumptions, as required by the FNA methodology. Two main scenarios are considered: •

The ‘Current Commitments’ scenario reflects existing policies and commitments, aligned with the National Energy and Climate Plan, and represents the central transition pathway.

•

The ‘Constrained Transition’ scenario assumes a slower uptake of electrification and flexibility technologies (notably electric vehicles, heat pumps and end-user flexibility) and also considers delays in the deployment of wind onshore generation, storage and grids across Europe, reflecting more challenging macro-economic conditions, supply chain difficulties, slower policy implementation and reduced public acceptance of onshore wind and infrastructure.

Figure 13: Overview of Scenario Framework

It was agreed in the implementation plan that the system needs are assessed for each scenario without gap filling, i.e., the flexibility means are quantified based solely on economically viable units and the units that are already known to be built under the previous auctions (prior to 2025) of the Capacity Remuneration Mechanism as defined in Adeqflex’25. All existing thermal capacities are assumed to remain in the market as also foreseen in Adeqflex’25. It means that part of the needs identified in this FNA study might be solved by the additional capacity (if any) contracted in future CRM auctions for adequacy purpose to fill the gap.

FNA report for Belgium 2026, version approved by CREG

However, the Adeqflex’25 CC and CT scenarios only account 1.5 GW of large-scale battery capacity, this assumption can be seen at the time of drafting the report as an absolute minimum threshold. Indeed, latest CRM auctions already contracted an additional 0.7 GW of large-scale batteries, while many connection studies for large-scale batteries have been delivered by Elia. For this reason, it was decided to complement these two scenarios, as from now referred to as ‘CC-AF25’ and ‘CT-AF25’ with two scenarios accounting the battery potential presented in Adeqflex’25 (Section 3.3.3.2): 4.5 GW in 2030 and 5.7 GW in 2035 of large-scale batteries in total. These scenarios are referred to as from now as ‘CC-BAT+’ and ‘CT-BAT+’. It is important to note that the BAT+ scenarios in the FNA reflect a higher battery deployment assumption based on the battery potential identified in Adeqflex'25. This should not be confused with the BAT+ sensitivities used in the Adeqflex'25 Economic Viability Assessment, which reflect more favourable market conditions for battery investments. The CC-AF25 and CT-AF25 scenarios represent a lower-bound estimate of large-scale battery deployment, as they only include existing and contracted capacities prior to 2025. Conversely, the CCBAT+ and CT-BAT+ scenarios reflect the battery potential identified in Adeqflex’25 and therefore represent an upper-bound scenario for storage deployment. At this stage, no assessments show how an additional 3 GW would come online economically by 2030 without support mechanisms. Therefore, this results in 4 scenarios assessed for both 2030 and 2035. For modelling purposes, all non-residential battery assets (> 10 kW) connected to the transmission or distribution network are grouped under the large-scale battery category, while residential batteries are modelled separately. This simplification has no material impact on the identified flexibility needs, which are assessed based on the aggregated storage contribution to the system. Additionally given other uncertainties and their impact on the flexibility needs (e.g. during moments of incompressibility), some of the analyses are complemented with additional sensitivities for both 2030 and 2035 as will be outlined in section 8 and will only assess the impact on the main indicators: • More existing nuclear capacity (3 or 4 GW instead of 2 GW) given the ongoing discussions on potential extensions of the existing nuclear fleet in Belgium; • Less ‘end user flexibility’ on residential assets (EV, heat pumps and small-scale batteries) compared to the baseline assumptions as some segments are At this stage • developing slower than expected. Specifically, the pace of unlocking flexibility reacting to market prices is reduced by a factor 2. This sensitivity was also defined in Adeqflex’25 and used as sensitivity analysis for adequacy results (Section 7.4.2 in Adeqflex’25).

3.1.2. Scenarios for network needs 3.1.2.1. Scenarios for DSO data To match the assumptions on available flexibility in the scenarios ‘Current Commitments’ (EU-BASECC) and ‘Constrained Transition’ (EU-BASE-CT) used by Elia to calculate the system needs, the DSOs will use the central scenarios from their DNDP as a basis. These scenarios correspond to the installed capacity of Solar PV and onshore wind and decentralised generation as shown in Table 5. Hydro Run of River and Waste are not included as these are negligible compared to the other resources. Table 5: Comparison of scenarios of installed capacities of solar and onshore wind power Belgium Elia (CC) 2030 Solar PV (MW) 2035 Solar PV (MW) 2030 Onshore Wind (MW) 2035 Onshore Wind (MW)

Elia (CT) 16 525 20 895

6 017 7 579

4 198 4 532

Flanders Fluvius 10000 14000 2800 2965

FNA report for Belgium 2026, version approved by CREG

ORES 3600 4680 2304 3312

Wallonia RESA 1120 1677 484 938

Arewal 194 250 107 127

Brussels Sibelga 500 635 -

Currently, DNDPs only consider demand response (upward flexibility) to manage the yearly peak load. As DNDPs do not quantify downward flexibility available during RES abundance, no flexibility was assumed in the calculation of the network needs on T/DSO connection point. Annex D on the DSO network needs methodology further elaborates the scenarios used by each DSO with a Comparison of scenarios of installed capacities of solar and onshore wind power and the DNDP assumptions between the different scenarios for 2030. For 2035, ORES, RESA and Arewal align the respective growth rates with the Elia scenario. The 2035 values are obtained by applying the 20302035 growth rate of the ELEC scenario3 to the DNDP 2030 assumptions. DSO do not plan to have quantified projections on BESS (residential or medium voltage). Fluvius assumes that batteries would either decrease or not impact the synchronous peak. Ores has many batteries in the queue for grid connection and future BESS projections will depend on ongoing and future discussions such as on flexibility and grid access. 3.1.2.2. Scenarios for TSO data (if applicable) As agreed in the implementation plan, an assessment of the TSO network needs is not conducted in this first version of the FNA. An implementation plan is presented in Section 7.2.

3.2. Target for RES integration This section starts giving an overview of installed renewable capacities of 2025 compared to the different scenarios considered (Figure 14). Installed renewable energy capacity is assumed to increase significantly between 2025 and 2035 across all major technologies. Solar photovoltaic capacity grows strongly throughout the period, while both onshore and offshore wind capacity expand substantially, with particularly pronounced growth after 2030 for offshore wind. These developments result in a near doubling of total installed renewable capacity over the study horizon.

Figure 14: Overview of renewable capacities considered in the model for Belgium

The approach for determining the assumptions on Targets for RES integration is approved by CREG through the methods (Annex A). In Belgium, the national RES ambitions are stated in the National Energy & Climate Plan (NECP) in terms of annual renewable energy volumes and global RES-E share. However, the NECP does not specify an explicit target regarding the acceptable level of RES curtailment, nor does it distinguish between installed capacity and effectively integrated renewable generation. Consequently, the

https://www.fluxys.com/-/media/project/fluxys/public/corporate/fluxyscom/documents/public-consultation_hydrogeninfrastructure/updates/scenario_data_excel_04112025.xlsx

FNA report for Belgium 2026, version approved by CREG

assessment is conducted across a range of RES curtailment level. This approach makes it possible to analyse the relationship between renewable integration and flexibility needs, while ensuring consistency with the renewable energy ambitions set out in the NECP. The analysis is structured around two boundary cases: • a zero-curtailment case, representing full integration of renewable generation and requiring additional non-fossil fuel flexibility resources; • a case without additional non-fossil flexibility, resulting in a corresponding level of RES curtailment. In addition, intermediate curtailment levels between these two boundary cases are assessed. These sensitivities help illustrate how different acceptable levels of curtailment affect the resulting flexibility needs and the additional renewable capacity that may be required to compensate for curtailed energy.

3.3. Portfolio of flexibility resources available A. Conventional flexibility Figure 15 shows the considered conventional flexibility in the simulations for the different scenarios and time horizons. It is also referred to as centralised or large-scale capacity. Only downward flexibility reacting to high prices is considered in this figure. The figures for upward flexibility will typically show higher demand-side management values. All capacities are calculated under the assumption of full and immediate response when activated. The values are nominal for storage meaning they do assume full availability (no outages nor maintenance). In practice storage technologies are constrained by their reservoir size. For DSM, some technologies—such as heat pumps—will only provide downward flexibility when there is an underlying demand, meaning availability is higher in winter than in spring or summer.

Figure 15: Overview of conventional flexible capacities considered in the model

The expected evolution of system flexibility is marked by a significant increase in centralised through batteries. By 2035, a substantial portion of conventional flexibility is projected to be provided by newly developed assets, primarily large-scale batteries complementing existing capacities such as pumpedstorage.

FNA report for Belgium 2026, version approved by CREG

Within the CC-AF25 and CT-AF25 scenarios, 1.5 GW of BESS is assumed for both 2030 and 2035, in line with Adeqflex’25. This capacity reflects the battery volumes included in the Adeqflex’25 reference scenarios, which were based on existing installations and capacities contracted prior to 2025. Any remaining adequacy gap was not filled with additional batteries but represented as non-dispatchable dummy capacity, i.e. not contributing to system flexibility. As mentioned earlier in this section, these scenarios are complemented by the CC-BAT+ and CT-BAT+ scenarios, which reflect the battery potential identified in Adeqflex’25. Together, these scenarios provide a range of plausible battery deployment pathways and allow the assessment of flexibility needs under different levels of expected large-scale storage development including on high and medium voltage levels. End-user flexibility The expected evolution of conventional flexibility is also complemented by a significant increase in decentralised resources. This capacity expands considerably, driven mainly by small-scale batteries, electric vehicles and heat pumps. This flexibility is progressively unlocked over time and is assumed to contribute through both market-based mechanisms and local, grid-driven activation. Figure 16 quantifies this evolution in terms of both average available capacity (GW) and number of participating devices. Also here, in practice, some technologies—such as heat pumps—will only provide downward flexibility when there is an underlying demand, meaning availability is higher in winter than in spring or summer. Note that downward flexibility following cooling is not accounted in the scenarios (this effect is under investigation and foreseen to be accounted in upcoming studies). Since this flexibility could play a role during periods of excess renewable generation, excluding it may lead to an underestimation of available downward flexibility. Nevertheless, the currently limited flexibility potential of heat pumps and air conditioning, combined with uncertainties regarding the alignment of cooling demand with renewable surpluses, suggests that further analysis should be conducted in future studies.. Most connected vehicles under dynamic contracts are expected to charge following market prices. It is also important to note that the available flexibility might also depend on the time of the day and whether it is a weekday or a weekend, as the effective availability of electric vehicles depends on the number of vehicles connected to the grid given that for EVs the effectively connected amount of cars also matter. The figures on end-user flexibility do not include medium scale batteries connected at the DSO level (considered as large-scale storage, cf. previous section). As mentioned earlier, for modelling purposes, all non-residential battery assets (> 10 kW) connected to the transmission or distribution network are grouped under the large-scale battery category. In parallel, decentralised flexibility expands considerably, driven mainly by small-scale batteries, electric vehicles and heat pumps. This flexibility is progressively unlocked over time and is assumed to contribute through both market-based mechanisms and local, grid-driven activation. The analysis quantifies this evolution in terms of both average available capacity (GW) and number of participating devices.

FNA report for Belgium 2026, version approved by CREG

Figure 16: Overview of decentralised flexible capacities considered in the model

For residential devices, the assumptions on which shares will react in the market on prices, and which shares will react on tariffs and self-consumption incentives as applied in Adeqflex’25 are illustrated in the figure below. The scenarios incorporate a continued increase in the level of activation of electric vehicles (EVs) and heat pumps (HPs). Regarding home batteries, a progressive shift is assumed from predominantly local optimisation (e.g. PV self-consumption) towards increased participation in marketbased mechanisms, notably through responsiveness to price signals.

Figure 17: Evolution of flexibility residential devices by type (in percentage of assets)

B. Renewable flexibility Figure 18 shows the renewable capacity that is assumed to react on negative prices in the scenarios. The values are based on an assessment of self-curtailment during low price periods in 2025, defined as periods combining abundant wind or solar conditions and negative day-ahead prices, i.e. situations in which a downward reaction from wind installations can reasonably be expected. This algorithm to determine the volume of market response is applied on available metering data. It measures how large and medium-sized installations react during periods of negative day-ahead prices. Two indicators are considered: • •

the total installed capacity of units that respond during more than 10% of negative-price periods; the total installed capacity of units that respond during more than 1% of negative-price periods.

The first indicator represents the capacity exhibiting a structural response to negative prices. The second indicator additionally captures capacities that react only during the most extreme negative-price events. In line with the objective of assessing the capability of renewable generation to contribute to periods of excess generation, the latter is retained as the reference assumption in the scenarios, as it also allows flexibility activated only during the most extreme negative-price events to be taken into account. However, these assumptions should still be regarded as conservative. They do not account for the increasing renewable participation already observed in the market since 2025, nor for the impact of FNA report for Belgium 2026, version approved by CREG

recent and ongoing measures aimed at strengthening renewable response to market signals, including the continued rollout of smart meters, dynamic contracts and other market reforms. The impact of these developments could not be quantified in this first FNA study and is therefore not reflected in the scenarios. Their effect will be further assessed in future editions of the FNA. The impact of higher reaction on negative prices is discussed in the sensitivities in Section 8. For offshore wind power, it is assumed that all capacity engages in downward flexibility on all market time frames. At least, in short-term markets, bidding obligations for parks above 25 MW ensure their participation in the balancing market. For onshore wind power, a quantitative assessment is made on available data covering 80% of onshore wind installations (out of the 3600 MW installed in 2025). Based on this analysed perimeter, and after extrapolations to the entire fleet, 2100 MW is found to exhibit curtailment behaviour during negative day-ahead price periods. This corresponds to approximately 60% of the installed onshore wind fleet already reacting to market signals when considering capacities reacting during more than 1% of negative-price periods. This assumption is retained in the simulations and should still be regarded as conservative.

Figure 18 : assumptions on market reactions of renewable flexibility

For solar power, the same analysis is conducted but shows significantly lower curtailed volumes and much less systematic curtailment behaviour. The solar analysis covers approximately 20% of the Belgian solar installations larger than 0.01 MW (around 3600 MW total installed end of 2025), for which individual metering data is available. This includes units larger than 25 MW but residential solar installations are not included in this analysis, as no individual metering data is currently available for the residential segment. As a result, no conclusions can be drawn at this stage regarding the curtailment behaviour of residential PV. It needs to be stressed that this only concerns stand-alone PV reaction and that all flexibility of other end-user applications (heat pumps, home batteries and electric vehicles) through dynamic prices, capacity tariffs or other incentives including selfconsumption is already accounted separately in the scenarios (cf. previous section) Within this analysed perimeter, solar curtailment is observed to exhibit 650 MW of curtailment behaviour is observed during very sunny periods combined with negative day-ahead prices. This corresponds to approximately 5% of the installed fleet already reacting to market signals when considering capacities reacting during more than 1% of negative-price periods. This assumption is retained in the scenarios and should still be regarded as conservative. Note that this value includes installations above 25 MW and assumes no substantial reaction for installations lower than 10 kW. Based on these observations, the lower bound is accounted in the simulations. Note that: •

The reaction of decentral solar and wind in short-term markets is not accounted as no data is currently available on reaction of solar and wind power assets in intra-day and balancing market prices.

•

Note that this assessment is based on observations from 2025. Renewable participation in response to market signals is expected to further increase in the coming years, reflecting both

FNA report for Belgium 2026, version approved by CREG

the increasing awareness of negative prices and the impact of recent and ongoing measures aimed at strengthening renewable response. As a consequence, the renewable response assumptions retained in this study should be regarded as conservative. Their use may lead to an underestimation of the renewable flexibility available to the system and, consequently, to higher identified residual flexibility and renewable response needs. Following the feedback received in the consultation on the FNA report, Elia will further refine and consult the methods in the next consultation on the scenarios for the Adequacy study’27. The study assumes that only solar and wind units larger than 25 MW react on short-term markets. These units should at least be able to react in balancing markets through regulatory requirements to bid in balancing markets. It is however expected that units smaller than 25 MW can be expected to participate also more and more in short-term markets, directly and through aggregation. On the other hand, it should be taken into account that large(st) part of the capacity is expected to have already reacted in day-ahead and may therefore not be double counted. No analyses and results are currently available to put forward specific numbers and this is subject to further investigation towards next rapports. Note that fixing the thresholds to 5% (solar) and 60% (wind) towards 2030 and 2035 in the scenarios assumes increasing reaction volumes in proportion to additional capacity installed but does not account for the increasing renewable participation already observed in the market since 2025, nor for the impact of recent and ongoing measures aimed at strengthening renewable response to market signals. As a result, the renewable flexibility available to the system may be underestimated and the remaining flexibility and renewable response needs may consequently be overestimated 4. As the impact of such measures is difficult to determine ex ante, sensitivities are conducted in Section 8 to determine the impact of additional reactions of renewable energy actors and show how higher reaction rates of renewables on market prices will reduce the uncovered flexibility needs. Note that the assumptions used in the scenarios will be updated in the scenarios for the Adequacy study’27 based on latest information available. Conclusions - Four main assumptions on flexibility Firstly, the study uses the available estimates on flexibility volumes on all technologies and voltage levels accounted for in the Adeqflex’25 scenarios used for the European system simulations. These are described in detail in Adeqflex’25 report and simulate the operations of all flexibility assets in the EU system subject to interconnection constraints. This also includes batteries and end-user flexibility (up to lower voltages). End-user flexibility assumes to react through market participation via prices and local optimisation via tariffs. It is assumed that over 55% in 2030 and 70% in 2035 of the EV fleet would react to local or price signals. Such an assumption can be seen as optimistic given the current 30% share. A sensitivity with lower shares of end-user flexibility will be accounted for in section 8 when assessing the impact on main indicators. Additional capacity (which could provide additional flexibility) to be developed for adequacy reasons (to guarantee the security of supply level – also called the adequacy gap) is not explicitly taken into account, beyond the capacities already contracted prior to 2025. In contrast to the CC-AF25 and CT-AF25 scenarios, CC-BAT+ and CT-BAT+ scenarios account for the potential increase of large-scale battery capacity assumed in Adeqflex’25.

Recent evolutions and measures that are not accounted include that by 1st of April 2026 net metering ended for almost all residential solar installations in Flanders after the installation of a digital meter (4.7 GW); expected re-powering of older installations getting exposed to prices by losing green certificates and getting replaced by more modern invertors with improved capabilities towards flexibility, the recent decision by the Flemish government to stop operational support for existing assets as of 400 kW during negative price hours, awareness around the deeply negative day-ahead and imbalance prices in April and May 2026 (including an imbalance price event up to -15.000€/MWh).

FNA report for Belgium 2026, version approved by CREG

Secondly, the scenarios assume 2 GW of nuclear capacity in Belgium for 2030 and 2035. It is important that the impact of an extension of additional nuclear generation units towards 2030 and 2035 is not accounted. However, such scenario should not be excluded as it lies within the current ambitions of the federal government. A sensitivity on additional nuclear capacity (+1 GW and +2 GW) is conducted in section 8 analysing the impact on the main indicators. Thirdly, the analyses are conducted with and without accounting additional RES flexibility. The ‘without’ assessment allows to identify the impact on renewable targets while the ‘with’ assessment allows to identify the operational challenge. Indeed, at least larger units than 25 MW but also a share of the decentral capacity is assumed to be available for negative price reactions or TSO activations (regulatory obligation) while current contributions of decentralised RES flexibility beneath 25 MW are not explicitly accounted for in the calculations. This is justified as participation remains limited, particularly for PV. Fourthly, analyses are conducted accounting for cross-border flexibility over all time horizons. Simulations are conducted in an interconnected EU market and additional cross-border flexibility in intra-day and balancing are taken into account, subject to transmission and liquidity constraints (with reserve sharing contracts as minimum thresholds) and all RES assets.

FNA report for Belgium 2026, version approved by CREG

4. Results – RES integration needs The RES integration needs assessment builds on the economic dispatch simulations conducted for the CC and CT scenarios for 2030 and 2035. The analysis first characterises the residual load (Section 4.1) and the resulting RES curtailment indicator (Section 4.2). The RES curtailment indicator is then compared with the RES integration target to identify any uncovered RES integration needs (Section 4.3). Finally, Section 4.4 assesses the flexibility resources that could contribute to addressing these targets.

Figure 19: Calculation steps on RES integration needs

It is important to clarify that the concept of “RES generation curtailment” (“curtailment”) used in this analysis refers specifically to the reduction of domestic renewable generation resulting from a domestic excess of generation. In the absence of such an excess in Belgium, no curtailment is assumed. This implies that RES curtailment in Belgium following negative prices induced by a cross-border excess is not accounted in the calculations. Therefore, in the case of Belgium, the amount of negative price periods exceeds the domestic excess RES periods and hence calculated curtailment periods. In this context, curtailment is assessed on a harmonised basis across bidding zones and reflects situations in which domestic renewable generation is reduced due to a domestic excess of generation. This approach does not explicitly account for technology-specific bidding behaviour, support schemes, contractual arrangements or other factors that may influence curtailment decisions in practice. In addition, given the highly interconnected nature of the Belgian electricity system, domestic prices may become negative as a result of excess generation in neighbouring countries even when no domestic excess generation exists. Given the current limitations in available data and the objective of ensuring a harmonised assessment across countries, this simplified representation provides a consistent basis for assessing RES integration needs in this first FNA. Future iterations of the FNA could further refine this approach by incorporating more granular representations of technology-specific behaviour, market responses and cross-border interactions, subject to improved data availability and methodological developments.

4.1. Characterisation of residual load time series The “residual load” indicator represents the load (demand) minus RES generation and thermal ‘must run’ generation (which includes domestic nuclear generation, certain gas-fired units, biomass and waste units). More details on the calculation on the residual load can be found in Annex A.1.1. It can be considered as a good indicator for ‘gross’ downward flexibility needs as negative residual load represents the part of the generation to be covered with local flexibility, or export. Note that this section does not differentiate between the ‘AF25’ and ‘BAT’ scenarios because the residual load is not impacted by additional batteries. Figure 20 illustrates the average hourly residual load profile for the scenarios and target years, compared to the actual (2024). It represents the average over all Monte Carlo years. The figure provides a clear visual basis to derive three indicators of the residual load, each capturing a different aspect of the daily residual load dynamics. First (1), the average daily difference reflects the typical intra-day variation in residual load. This can be observed in the figure through the deviation of the average residual load profile (solid line) from the horizontal dashed line representing the daily mean. The pronounced dip around midday and the FNA report for Belgium 2026, version approved by CREG

increase towards the evening illustrate the structural daily cycle, mainly driven by solar generation and demand patterns. This indicator provides a measure of the baseline, recurring flexibility requirement that the system must accommodate on a daily basis. For 2030 this results in around 2.5 GW for both scenarios. Where this increases towards 3 GW in 2035. Second (2), the P99 daily spread captures the upper bound of typical daily variability. In the figure, this is represented by the vertical extent of the shaded band (P01–P99 range across the Monte Carlo years). This band shows the range within which most daily outcomes fall, excluding extreme situations. The thickness of this band therefore reflects how much the daily residual load can vary under different conditions. This variability equals 14 GW in 2030 and increases towards 18 GW in 2035. Indicating an increasing daily variability towards 2035. The highest variability occurs during midday. Third (3), the negative peak represents the most extreme downward flexibility requirement. This can be identified in the figure as the lowest point of the lower envelope of the shaded area, typically occurring around midday. These negative values correspond to periods where renewable generation exceeds demand. This indicator highlights the maximum required downward flexibility, directly linked to the integration of renewable energy sources and the need to absorb excess generation. In 2030 the P01 goes down to 6 GW, where this becomes even more negative towards 9 GW in 2035. This reflects more frequent and deeper periods of excess generation relative to demand, increasing the need for downward flexibility.

FNA

Figure 20: Hourly residual load profile over all Monte Carlo years. Indicating the hourly average, daily average and the percentile 1 and 99 over all Monte Carlo years as a range

Figure 21 presents the residual load breakdown in terms of system drivers: all must run generation types for an average daily pattern in the CC 2030 scenario. The shown day is an average over all days of summer (left) and winter (right) and all Monte Carlo years. The breakdown clearly highlights the contribution of solar generation to the midday drop in residual load in summer, which is less visible during winter. In winter, more wind power is observed which depends on wind conditions but does not show any recurrent daily behaviour.

FNA report for Belgium 2026, version approved by CREG

Figure 21: Hourly residual load breakdown for the 2030 current commitment scenario

Figure 22 (top) presents the average residual load profile for every day of the week, calculated as average across all Monte Carlo years, calculated as the sum of the residual load per day and then averaged per day over all days and all Monte Carlo years. During weekdays, the residual load profile remains relatively similar over the different days, while a general reduction of the residual load below the weekly average during weekends is observed because of lower demand. As in previous figures, the range of the residual load widens toward 2035, indicating increasing day-to-day variability in the residual load profile over time.

300

P01 - P

300

over all Monte Carlo years

200

wee ly averages

150

2035

100

2030

50 0

Mon

Tue

Wed

Thu

Fri

Sat

over all Monte Carlo years

200 150

wee ly averages

100

2035

2030

Sun

Mon

-50

-100

7000

Tue

Wed

Thu

Fri

Sat

Sun

7000

6000

Full range over all MonteCarlo years

5000

2035

4000

2030 yearly averages

3000 2000 1000

Residual laod [GWh]

P01 - P

250

Residual laod [GWh]

250

Full range over all MonteCarlo years

5000

2035

4000

2030

3000

yearly averages

2000 1000

0 Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Figure 22: Daily (top) and monthly (bottom) residual load profile over all Monte Carlo years. Indicating the weekly, monthly and yearly average and the minimum and maximum over all Monte Carlo years as a range

Figure 22 (bottom) presents the average residual load profile for every month of the year. The profile in this figure is obtained by first aggregating the residual load over each month and subsequently averaging the results across all Monte Carlo years. For the monthly indicator the minimum and maximum values are shown instead of the 1st and 99th percentiles. The results indicate a clear seasonal pattern in residual load. Average residual load levels are generally lower during the spring and summer months (with lowest values in May), driven by increased solar generation in combination with still FNA report for Belgium 2026, version approved by CREG

relatively high wind generation and lower electricity demand during the summer period, while higher residual load levels are observed during the winter months, mainly due to higher electricity demand. Note that both the CC and CT scenarios exhibit very similar monthly patterns, which is expected given the shared weather conditions. However, due to the lower overall load in the CT scenario, its entire curve is slightly shifted downward compared to the CC case. The fact that lower residual load levels are observed in spring instead of summer can be explained by Figure 23, which shows the monthly RES generation and total load (left) and the monthly ratio between those two (right), averaged over Monte Carlo years in 2035 for the CC scenario. The lowest residual load values occur in May and June, as the total load is lower during those months (e.g. due to bank holidays) while RES generation (in particular solar PV production) peaks during this period. Combining those two effects, the RES generation to total load ratio reaches its peak in May (and not in summer) and its minimum in December.

Figure 23: Monthly average RES generation and total load (left) and average RES generation to Total Load ratio (right) in 2035 (CC scenario)

In conclusion, the residual load profiles evolve with increasing renewable generation indicating higher needs for flexibility. The increasing occurrence of negative residual load periods, together with the widening spread between day peaks and day valleys and between annual minimum and maximum residual load levels, reflects the increasing variability associated with higher renewable penetration. Toward 2035, minimum residual load levels decrease further, reflecting the combined effect of additional wind and solar generation together with periods of lower demand. As a result, the lowest residual load values are typically observed during spring and summer, with a minimum in May, as well as during weekends and public holidays, when demand is structurally lower.

FNA report for Belgium 2026, version approved by CREG

4.2. Characterisation of RES generation curtailment The RES curtailment indicator takes into account availability of flexibility and cross-border imports and exports during periods of generation excess (negative residual load) between neighbouring countries, ensuring a correct attribution of curtailment to each country in the economic dispatch simulations. These hourly time series are calculated through post-processing. The resulting indicator should be interpreted as a stylised representation of renewable generation that cannot be integrated under the assumptions of the assessment, rather than as a prediction of future realised curtailment. More details on the calculation of the RES curtailment indicator can be found in Annex A.1.2. Figure 24 (left) illustrates the yearly values of the RES curtailment indicator per target year. For each target year the average total yearly curtailment across the Monte Carlo years is depicted along with the minimum and maximum total yearly curtailment over the Monte Carlo years. The figure displays an increasing trend of curtailment over the years for both scenarios. For the CC-AF25 scenario, the average yearly curtailment in 2030 is 290 GWh, which accounts for only 0.8% of all generated wind and solar. In the CC-BAT+ scenario this even decreases towards only 140 GWh or 0.4% of all generated wind and solar. Towards 2035 this increases to 850 GWh accounting for 1.6% of total generated wind and solar or 480 GWh in the higher battery scenario.

Figure 24: Annual values of the RES curtailment indicator and descriptive statistics for each target year (left) and total hours with a non-zero RES curtailment indicator due to domestic excess (right), showing the average and range across all Monte Carlo years.

For the CT-AF25 scenario, curtailment is higher due to the combined effect of an assumed lower load and lower end-user flexibility (following lower electrification), which outweighs the impact of the lower installed onshore wind capacities assumed in that scenario. The reduction in load leads to a lower starting point for calculating residual load. When renewables are subtracted from an already lower load level, the residual load is more likely to fall below zero even with slightly less wind generation. As a result, curtailment is higher despite the lower onshore wind capacity. In the CT-BAT+ scenario, the resulting curtailment is lower as additional batteries absorb part of the excess generation. This effect is particularly pronounced because excess generation is more concentrated in short, solar-driven periods, which batteries are especially effective at managing. FNA report for Belgium 2026, version approved by CREG

Overall, the resulting curtailment remains relatively limited compared to total load and total wind and solar generation. In the CC scenarios, the yearly volume corresponds to 0.4–0.8% of total wind and solar generation in 2030, depending on the battery assumptions, increasing to 0.9–1.6% in 2035. For the CT scenarios, the corresponding ranges are 0.3–1.1% in 2030 and 0.8–2.1% in 2035.

Relation between curtailment due to domestic excess (referred here as “curtailment”) and negative prices. Negative or near-zero electricity prices can be explained by three main mechanisms: 1. Domestic excess generation that sets the market price and may result in curtailment 2. Domestic excess generation that is fully absorbed by flexibility, with flexibility resources setting the market price 3. In an interconnected system, imports at negative prices resulting from excess generation in neighbouring countries while Belgium is not experiencing domestic excess generation. This part consists of the main driver of negative prices nowadays given hours with domestic excess in Belgium are very limited but negative prices are not. First, negative prices can arise from domestic excess generation that cannot be absorbed within the system. This relates to the curtailment analysed in this section. Since renewable generators typically have very low marginal costs and may incur opportunity costs or lose financial support when curtailed, they are willing to accept negative prices. In this case, negative prices are a direct signal of structural oversupply. Second, negative prices can also occur when domestic excess generation is fully managed by flexibility, without actual curtailment taking place. In this situation, negative prices can be set by flexibility resources when they are the marginal option needed to absorb surplus generation, as these resources may require zero or negative prices to compensate for efficiency losses, opportunity costs, or operational constraints, even though the system itself remains fully balanced. Third, and the most predominant driver, negative or near-zero prices can occur when Belgium imports electricity from neighbouring countries experiencing excess generation, even if Belgium itself is not structurally long. In an integrated and coupled electricity market, prices are influenced by regional conditions. When neighbouring systems face significant renewable surplus, producers may be willing to pay to produce which can lead to an export of electricity. If interconnector capacity is available and Belgian demand can absorb these volumes, imports displace more expensive domestic generation. As a result, Belgian prices can fall to zero or negative levels despite the absence of domestic excess. In this case, the Belgian price reflects regional oversupply rather than national system conditions alone. In practice, such events can also lead to effective reduction in renewable generation in Belgium despite no domestic excess. It will depend on the prices at which generators in Belgium and abroad are willing to reduce production. Together, these three mechanisms show that negative or near-zero prices should not be interpreted solely as evidence of domestic excess generation (referred in this study as ‘curtailment’). They may also reflect efficient flexibility activation or cross-border trade within an integrated European electricity system, which is particularly relevant when analysing price signals in economic dispatch models. As outlined in the introduction of this Section, curtailment is assessed on a harmonised basis across countries and is intended to identify periods of domestic excess generation that cannot be absorbed through available flexibility or exports.

FNA report for Belgium 2026, version approved by CREG

Figure 24 (right) illustrates the total number of hours with curtailment due to domestic excess and the range across the Monte Carlo years. Overall, the figure indicates that curtailment occurs during a relatively limited number of hours, averaging 90 - 230 hours in 2030 and around 220 – 470 hours in 2035, depending on the scenario. However, these values vary substantially across the Monte Carlo years, reflecting the strong influence of weather conditions (e.g. sunny weekends). Figure 25 presents the hourly distribution of all curtailment events. The y-axis is capped at 0.6% in order to zoom in on the non‑zero curtailment hours. The resulting distribution is strongly right‑skewed, indicating that most curtailment events are small in magnitude, while rare extreme events can lead to very large curtailment volumes.

Figure 25: Hourly curtailment distribution with percentiles per target year

FNA report for Belgium 2026, version approved by CREG

The hourly curtailment distribution for 2030 under the CC-AF25 scenario shows that the majority of curtailment hours remain below 3 GW. In the CC-BAT+ scenario this threshold decreases to 1.8GW. The figure also includes extreme percentiles for each scenario, showing the probability that curtailment exceeds a given level. The 99.99th percentile for 2030 in the current commitment scenario indicates that curtailment exceeds 6.4 GW in only 0.01% of hours (or 5.1 GW in the higher-battery scenario). By 2035, these extreme values increase to around 10 GW. It can be concluded that while curtailment only occurs during a limited number of hours, rare events can lead to very high curtailment volumes. While most events remain relatively small, the increasing frequency and magnitude of extreme peaks toward 2035 highlight the growing importance of managing these tail risks.

4.2.1. Daily curtailment profile Figure 26 presents the daily curtailment profiles for all scenarios and shows a consistent midday peak across scenarios. However, curtailment is significantly reduced in the BAT+ scenarios, indicating that additional battery capacity is able to absorb a substantial share of the excess generation. This recurring increase is driven by the strong positive correlation between curtailment and solar PV generation, which was already observed when looking at the residual load in Figure 21. Correlation plots further illustrating the relationship between curtailment and system conditions are presented in Section 4.2.3.

Figure 26: Hourly curtailment distribution per target year showing the average and 95th and 99th percentile

To better assess the seasonality of the curtailment, Figure 27 shows curtailment profiles averaged over the season providing a better ‘nuanced’ profile than the average day of the entire target year. Across all different scenarios, it can be observed that during spring months curtailment is significantly larger. Also, summer months indicate a high curtailment size on average. Winter and Autumn show the lowest peaks. Note that the year 2030 is not shown because the results show similar conclusions.

FNA report for Belgium 2026, version approved by CREG

Figure 27: Average hourly curtailment distribution per season and scenario for 2035

In conclusion, daily curtailment consistently peaks around midday across all scenarios, reflecting the dominant role of solar PV generation in combination with broader renewable generation patterns and demand conditions in driving curtailment patterns. Seasonal analysis further shows that curtailment risk is highest in spring and summer, while winter and autumn exhibit significantly lower levels, highlighting the combined influence of solar generation and seasonal demand dynamics.

4.2.2. Monthly curtailment profile Figure 28 presents the monthly curtailment for 2035 across the different scenarios. The figure illustrates both the average monthly curtailment and the spread across Monte Carlo years, further highlighting clear seasonal patterns and an increasing level of uncertainty over time. Curtailment is primarily concentrated in the spring and early summer months, with April, May, and June showing the highest average values. During these months, both the median and upper range of curtailment increase, reflecting the combined effect of rising solar generation and relatively lower demand. In contrast, winter months such as January, February, November, and December exhibit very limited curtailment. In general, significant average curtailment (above 50 GWh) between March and September is observed. While the seasonal pattern remains broadly similar across scenarios, the magnitude of curtailment is significantly reduced in the BAT+ scenarios, indicating that additional battery capacity is able to absorb a substantial share of the excess generation throughout the year, particularly during the spring and summer months when curtailment levels are highest. The period with significant average curtailment values is shortened from April to July. Note that the year 2030 is not shown because the results show similar conclusions, but with lower averages compared to 2035.

FNA report for Belgium 2026, version approved by CREG

Figure 28: Monthly curtailment per scenario for 2035

4.2.3. Correlation of curtailment with renewable generation and demand Figure 29 shows scatterplots of curtailment against RES and load for the current commitment scenario in 2030. Each plot includes a trendline indicating the direction and strength of the correlation. A positive slope of the trendline reflects a positive correlation, which is observed for onshore wind, offshore wind, and solar generation. This is intuitive, as higher renewable generation further reduces the residual load, increasing the likelihood that it falls below zero and results in curtailment. The trend is the most prominent for solar, confirming the correlation between solar and curtailment as seen before.

Figure 29: Correlation plots between curtailment and RES and load for 2030 CC FNA report for Belgium 2026, version approved by CREG

In contrast, curtailment exhibits a very weak negative correlation with total load (Table 6). This is also consistent with expectations: lower load levels reduce the initial load from which renewables are subtracted, increasing the probability of negative residual load and hence curtailment. Curtailment does not only occur during the lowest loads but to a wide range of load levels. Conversely, the scatterplots show that curtailment does not occur during very high load hours. In such situations, a substantial volume of non-dispatchable generation would be required before the residual load could drop below zero, making curtailment highly unlikely.

Table 6: Correlation table between curtailment and RES and load across target years and scenarios

4.2.4. Curtailment duration In addition to analysing the magnitude and timing of curtailment, it is also relevant to investigate their duration. Particularly in light of energy-constrained technologies. Longer duration events may be more difficult to manage than short duration ones. For this, a curtailment event is defined as a continuous period during which curtailment is strictly non-zero, starting when curtailment first becomes positive and ending when it returns to zero. Figure 30 presents the distribution of curtailment event durations by target year and scenario. Overall, the figure shows that a number of curtailment events can exceed the typical energy content of batteries (estimated 4 hours at maximum capacity). Most events last less than 10 hours, with a substantial share consisting of single-hour events. In addition, a clear plateau can be observed for event durations between approximately 2 and 10 hours, indicating that multi-hour curtailment episodes are not uncommon but remain relatively limited in duration. When the battery potential identified in Adeqflex’25 is considered (BAT+ scenario), there is a shift towards shorter curtailment events. However, longer-duration events remain present, indicating that additional battery capacity does not fully eliminate all prolonged periods of excess generation. For 2030 under the CC-AF25 scenario, the median (P50) curtailment event duration is around 2 hours, while extreme events remain rare, with the 99th percentile reaching approximately 15 hours. In the CCBAT+ scenario, the median duration decreases to around 1 hour and the upper tail is significantly reduced, with the 99th percentile decreasing to around 6 hours. Under the CT-AF25 scenario, the median duration increases slightly to around 3 hours, with a somewhat shorter upper tail (P99 around 11 hours). In the CT-BAT+ scenario, the median duration decreases to around 1 hour and the upper tail remains the same. The similar upper-tail durations observed in the AF25 and BAT+ scenarios can be explained by the reduction in the total number of curtailment events when additional batteries are included. Because the 99th percentile is calculated over a smaller set of events, it becomes more strongly influenced by the remaining extreme events, resulting in a limited reduction of the upper tail. In addition, batteries tend to reduce the magnitude of curtailment peaks rather than the duration of prolonged events. When the available battery capacity is insufficient to absorb an entire curtailment event, it primarily lowers the peak volume while leaving part of the event duration unchanged. As a result, the impact on extreme event durations remains limited. For 2035 under the CC-AF25 scenario, the median (P50) curtailment event duration increases to around 3 hours, while the upper tail extends further, with the 99th percentile reaching approximately 19 hours. In the CC-BAT+ scenario, the median duration decreases to around 2 hours, while the upper tail remains broadly similar, with the 99th percentile reaching around 18 hours. FNA report for Belgium 2026, version approved by CREG

Under the CT-AF25 scenario, the median duration remains around 3 hours, while the 99th percentile reaches approximately 15 hours. In the CT-BAT+ scenario, the median duration decreases to around 1 hour, while the upper tail reaches around 16 hours. The reason for this similar duration in the 99th percentile is the same as mentioned above.

Figure 30: Hourly distribution of curtailment event durations per target year

4.2.5. Conclusion In terms of annual volumes, curtailment remains limited relative to total renewable generation, reaching on average around 290 GWh (0.8% of wind and solar generation) in 2030 and increasing to approximately 850 GWh (1.6%) in 2035 under the CC-AF25 scenario, which reflects the battery capacity currently installed and contracted through CRM auctions up to 2024. When the battery potential

FNA report for Belgium 2026, version approved by CREG

identified in Adeqflex’25 is considered (CC-BAT+ scenario), these volumes are significantly reduced, reaching around 140 GWh in 2030 and 480 GWh in 2035. In terms of frequency and capacity, curtailment occurs during a relatively small share of hours—about 200 hours in 2030 and 400 hours in 2035 under the CC-AF25 scenario. In the BAT+ scenario these values decrease to approximately 100 and 250 hours respectively. Most curtailment hours are small in magnitude, remaining below 3 GW. Rare but significant events nevertheless remain present. Under the CC-AF25 scenario, rare but significant events become increasingly relevant toward 2035. Hourly curtailment can exceed 6 GW in 2030 and 10 GW in 2035. In the BAT+ scenario, the magnitude of these extreme events is reduced, with the 99.99th percentile decreasing from around 6.4 GW to 5.1 GW in 2030 and from around 10 GW to around 9.2 GW in 2035. These events are primarily concentrated around midday and during spring and summer months, reflecting the combined influence of solar generation, broader renewable generation patterns and seasonal demand dynamics.

4.3. Uncovered RES integration needs Uncovered RES integration needs are defined as the volume of RES curtailment exceeding the maximum level of curtailment compatible with the RES integration target. For Belgium, the current NECP does not define a quantitative target for acceptable RES curtailment. To assess the impact of allowing different degrees of curtailment, sensitivities are performed for different levels of acceptable RES curtailment. These sensitivities illustrate how the identified RES integration needs decrease when part of the assessed curtailment is considered compatible with the RES integration target. RES integration can be achieved through additional non-fossil flexibility resources, through a higher level of acceptable RES curtailment, or through a combination of both. In the absence of a quantified national curtailment target, this and the next section illustrate two boundary cases. The first corresponds to full RES integration, where zero curtailment is considered acceptable. Under this assumption, uncovered RES integration needs are equal to the assessed RES curtailment and can therefore be interpreted as the RES integration needs that would need to be addressed through additional non-fossil flexibility resources. These results will be illustrated in Section 4.4. The second boundary case assumes that no additional non-fossil fuel flexibility resources beyond those already considered in the scenario are available. In this situation, any remaining RES integration needs must be addressed through the response of renewable assets to system conditions and market signals. The assumptions regarding renewable assets expected to respond to market signals are described in Section 3.3 and include all offshore wind capacity, approximately 60% of onshore wind capacity and around 5% of solar PV installations. Given the uncertainty of future market response, additional sensitivity analyses are performed in Section 8, including higher response levels for onshore wind (up to 100%) and solar PV. Additional RES response needs represent the remaining RES integration needs after accounting for both the expected response of renewable assets to market signals (as described in Section 3.3) and a given level of additional non-fossil flexibility resources. In this section, this level is assumed to be zero.

FNA report for Belgium 2026, version approved by CREG

Figure 31 presents the duration curve of the additional RES response needs, illustrating both the volume of renewable generation that would need to respond and the number of hours during which this response would be required in order to achieve the RES integration target. For a reliability criterion of one hour per year, additional RES response needs reach 3.4 GW in 2030 and 4.6 GW in 2035 under the CC-AF25 scenario, which reflects the battery capacity currently installed and contracted through CRM auctions up to 2024. When the battery potential identified in Adeqflex’25 is considered (BAT+ scenario), these values decrease to 1.7 GW in 2030 and 2.9 GW in 2035. Under the CT-AF25 scenario, additional RES response needs increase further to 4.1 GW in 2030 and 5.7 GW in 2035. Considering the battery potential identified in Adeqflex’25 reduces these values to 2.6 GW and 4.0 GW respectively. When relaxing the reliability criterion from one hour to three hours per year, the required response volumes decrease accordingly. Under the CC-AF25 scenario, additional RES response needs decrease from 3.4 GW to 2.3 GW in 2030, while under the CT-AF25 scenario they decrease from 4.1 GW to 3.2 GW. Similar reductions are observed in the BAT+ scenarios. Although the battery potential identified in Adeqflex’25 substantially reduces these needs, significant response volumes remain across all scenarios, highlighting the structural nature of periods of excess generation.

Figure 31: Additional RES response needs after the expected response of renewable capacity

In line with the RES curtailment indicator presented in the previous section, periods of elevated additional RES response needs coincide with conditions characterised by low system load and high renewable generation, particularly solar PV generation. As a result, these needs are primarily concentrated during spring and summer weekends, when lower demand amplifies periods of excess generation. Figure 32 shows the hourly P90 value of the additional RES response needs across all Monte Carlo years. FNA report for Belgium 2026, version approved by CREG

This temporal clustering is clearly visible in the heat map for the CC-AF25 scenario, which highlights recurring daytime patterns of elevated additional RES response needs. An additional RES response is required around 1.2% to 1.9% of the year in 2030 and 2035 respectively, and the highest needs typically occur around midday between April and August in the CC-AF25 scenario. On the other hand, the battery potential identified in Adeqflex’25 substantially reduces these needs, reducing by 95% the number of hours during which an additional RES response is required in CC-BAT+ scenarios compared to CCAF25 scenarios, while reducing the peak values by 65 to 35% in 2030 and 2035 respectively. Remaining hours with non-zero additional RES response mostly take place in May and June in the CC-BAT+ scenarios. Current commitments

Perfect foresight needs JAN

FEB

MAR

APR

MAY

JUN

JUL

AUG

SEP

OCT

NOV

DEC

2030 CC-AF25

24h

2035 CC-AF25

24h

Current commitments

Perfect foresight needs JAN

FEB

MAR

APR

MAY

JUN

JUL

AUG

SEP

OCT

NOV

DEC

24h

2030 CC-BAT+

24h

2035 CC-BAT+

Figure 32: Temporal representation of Additional RES response needs [MW] in the CC-AF25 and CC-BAT+ scenarios (hourly P90th value over all Monte Carlo years).

Section 8 provides additional sensitivities with higher shares of renewable assets responding to market signals, including decentralised wind and solar generation, in order to assess the remaining additional RES response needs under alternative assumptions regarding renewable response capability.

FNA report for Belgium 2026, version approved by CREG

4.4. Flexible resources needed to meet the uncovered needs The previous section introduced two boundary cases for achieving the RES integration target. The first assumes full RES integration (i.e. zero acceptable curtailment), in which the resulting RES integration needs must be addressed through additional non-fossil flexibility resources. The second assumes that no additional non-fossil flexibility resources beyond those already included in the scenario are available, in which case the remaining RES integration needs must be addressed either through an additional response of renewable assets to system conditions and market signals or through the installation of additional generation capacity of renewable assets. This section focuses on the first boundary case and quantifies the additional flexibility resources required to achieve the RES integration target. Section 4.4.1 evaluates the flexibility requirements associated with each flexibility option individually, considering durations ranging from 1 hour to 200 hours. This allows the marginal contribution of each flexibility option to be assessed independently. Section 4.4.2 then evaluates flexibility requirements while allowing multiple flexibility options to be deployed simultaneously. Section 4.4.2.1 assesses the flexibility requirements under full RES integration (zero acceptable curtailment). This allows to calculate the minimum required additional capacity of non-fossil fuel flexibility options to cover all uncovered RES integration needs. Section 4.4.2.2 explores sensitivities around the boundary case investigated in Section 4.4.2.1 by investigating the impact of different acceptable curtailment levels on the resulting flexibility requirements. In the same section, a sensitivity analysis is performed on the round-trip efficiency of flexibility resources. Finally, section 4.4.2.3 assesses, for different levels of acceptable RES curtailment, the amount of additional renewable capacity that would be required to compensate for the curtailed renewable generation and maintain the RES integration target. This way, the results allow policy makers to consider two alternative approaches when updating their national energy and climate plans: developing additional non-fossil flexibility resources to achieve full RES integration (i.e. no curtailment) or increasing renewable deployment while accepting some level of RES curtailment in order to maintain the RES target. In practice, different combinations of these approaches may achieve a given RES integration target. It should be noted, however, that this assessment does not incorporate economic considerations, and therefore does not aim to identify or recommend an economically optimal solution. Note that all figures shown in this section show non-fossil flexibility that needs to be developed on top of already existing flexibility. All numbers are thus additional flexibility not the total needs including existing flexibility.

4.4.1. Quantification with single optimisation This section presents the results of the flexibility quantification performed in alignment with the FNA methodology. In practice, this assessment is carried out by adding a fixed capacity of a single theoretical flexibility type / option—characterised by its duration—to the system and re-optimising RES curtailment independently for each Monte Carlo year. For each capacity increment, the remaining curtailment is calculated, and the average marginal curtailment reduction is derived as the reduction in curtailment achieved per additional GWh of flexibility. By repeating this process in steps of 1 GWh over a wide capacity range and for each flexibility option separately, the marginal curtailment reduction curves are obtained. For interpretability, the marginal reduction curve is shown alongside the curve of average remaining RES curtailment. Together, these curves indicate both the effectiveness of additional flexibility and how FNA report for Belgium 2026, version approved by CREG

much curtailment remains after adding a given amount of capacity. Figure 33 and Figure 34 display these curves for 2030 and 2035, respectively. In 2030 (Figure 33), short-duration flexibility options (1h, 2h, 4h and 8h) reduce curtailment rapidly for relatively small added capacities. Compared to medium- and long-duration options, short-duration flexibility achieves significantly larger curtailment reductions per unit of installed energy capacity. For example, flexibility options with durations above 50 hours show a much flatter reduction curve, indicating lower effectiveness in mitigating the predominantly short-duration curtailment events observed in the simulations. Despite higher curtailment volumes in the CT-AF25 scenario, the required capacity remains close to the CC-AF25 scenario (94 GWh in both cases). This reflects the fact that the sizing is driven by a limited number of extreme curtailment events rather than by average annual curtailment volumes. When the battery potential identified in Adeqflex’25 is considered (BAT+ scenario), the required capacity decreases to 78 GWh for the CC-BAT+ scenario and 67 GWh for the CT-BAT+ scenario. It should be noted that these capacities are very large. This is because the required capacity is determined during the optimisation by forcing curtailment to be reduced to zero across all simulated conditions. In practice, this means eliminating curtailment in each of the 25 Monte Carlo years, which represents a very stringent constraint. As a result, the flexibility portfolio must be sized to cover all extreme events across the full range of scenarios, leading to significantly higher capacity requirements. Such values should therefore be interpreted as upper-bound estimates associated with the objective of achieving full RES integration under all simulated conditions for a specific flexibility option. To assess more realistic levels of RES integration, sensitivities with different acceptable curtailment levels are introduced in Section 4.4.2.2.

FNA report for Belgium 2026, version approved by CREG

Average remaining RES curtailment [GWh]

Average RES curtailment marginal reduction [GWh]

400

94 GWh

200h

300

100h 50h 20h

200

Current commitments 94 GWh

200h 100h

50h 20h

8h 4h

100

8h 4h

2h 1h

Added capacity

0 0

100

[GWh]

Average remaining RES curtailment [GWh]

Added capacity

0 0

100

[GWh]

Constrained Transition

Average RES curtailment marginal reduction [GWh] 94 GWh

400

94 GWh

60 50

300

200h 100h 50h

200

20h 8h

200h

100h 50h

20h 8h

100

0 0

100

Added capacity

100

[GWh]

Average remaining RES curtailment [GWh]

[GWh]

Average RES curtailment marginal reduction [GWh]

400

78 GWh

200h

300

100h 50h 20h

200

Current commitments

78 GWh

200h 100h

50h 20h

100

Added capacity

0 0

Added capacity

100

[GWh]

Average remaining RES curtailment [GWh]

Added capacity

0 0

100

[GWh]

Constrained Transition

Average RES curtailment marginal reduction [GWh] 67 GWh

400

67 GWh

60 50

300

200h 100h 50h

200

20h 8h

200h

100h 50h

20h 8h

100

0 0

100

Added capacity

[GWh]

100

Added capacity [GWh]

Figure 33: Average remaining curtailment and marginal reduction curves for 2030

In 2035 (Figure 34), overall curtailment levels are higher, while the same trends observed in 2030 persist. Shorter-duration flexibility options achieve a steeper reduction of curtailment per unit of capacity, particularly at low capacity levels. This behaviour is consistent with the power‑to‑energy characteristics of the flexibility options. Shorter‑duration flexibility has a higher power‑to‑energy ratio, allowing it to absorb sharp curtailment peaks more effectively than longer‑duration options for the same installed energy capacity. Compared to 2030, the capacity required to achieve full RES integration increases significantly across all scenarios. Under the CC-AF25 and CT-AF25 scenarios, which reflect the battery capacity currently installed and contracted through CRM auctions up to 2024, the required capacities reach 310 GWh and 302 GWh respectively. When the battery potential identified in Adeqflex’25 is considered (BAT+ scenario), these values decrease to 273 GWh for the CC scenario and 233 GWh for the CT scenario.

FNA report for Belgium 2026, version approved by CREG

The relatively limited difference between the CC-AF25 and CT-AF25 scenarios, despite materially different annual curtailment volumes, indicates that sizing is primarily driven by a limited number of extreme curtailment events rather than by average annual curtailment volumes. The relative increase between 2030 and 2035 remains broadly similar across scenarios, with the required capacity increasing by a factor of 3.25 in the AF25 scenarios and by 3.5 in the BAT+ scenarios. As in 2030, these capacities should be interpreted as upper-bound estimates. Achieving full RES integration requires eliminating curtailment under all simulated conditions and across all Monte Carlo years, resulting in flexibility portfolios sized to accommodate rare but particularly severe curtailment events. Sensitivities with different levels of acceptable curtailment are therefore introduced in Section 4.4.2.2. Average remaining RES curtailment [GWh]

Average RES curtailment marginal reduction [GWh] 310 GWh

1.000

Current commitments 310 GWh

90 200h

800

200h

100h 50h

600

20h 8h

400

100h

50h 20h

8h 4h

200

Added capacity

120

160

200

240

280

[GWh]

320

Average remaining RES curtailment [GWh]

2h 1h

Added capacity

120

160

200

240

280

320

Constrained Transition

Average RES curtailment marginal reduction [GWh] 302 GWh

1.000

302 GWh

90 75

800 200h 100h

600

50h 20h

400

200h

100h 50h

20h 8h

200

0 0

120

160

200

240

280

320

Added capacity

120

160

200

240

280

320

[GWh]

Average remaining RES curtailment [GWh] 273 GWh

[GWh]

800

Current commitments 273 GWh

90 200h

200h

100h 50h

600

20h 8h

400

100h

50h 20h

8h 4h

200

Added capacity

0 40

120

160

Added capacity

Average RES curtailment marginal reduction [GWh]

1.000

[GWh]

200

240

280

320

[GWh]

Average remaining RES curtailment [GWh]

Added capacity

0 0

120

160

200

240

280

320

[GWh]

Constrained Transition

Average RES curtailment marginal reduction [GWh] 233 GWh

1.000

233 GWh

90 75

800 200h 100h

600

50h

20h

400

200h

100h 50h

20h 8h

200

0 0

120

160

200

240

280

320

Added capacity

[GWh]

120

160

200

240

280

320

Added capacity [GWh]

Figure 34: Average remaining curtailment and marginal reduction curves for 2035

However, an important limitation of this quantification approach is that additional flexibility options are evaluated independently. Each curve represents the effect of adding only one type of flexibility, without considering interactions or complementarities with other options. As a result, the approach does not allow for the identification of an optimal mix of flexibility resources with different durations. The results should therefore be interpreted as illustrating the relative effectiveness of each flexibility option when assessed in isolation. FNA report for Belgium 2026, version approved by CREG

Within this framework, short-duration flexibility options (1h, 2h, 4h and 8h) appear to be the most effective at reducing RES curtailment per unit of installed energy capacity. However, these results do not allow a clear distinction between the different short-duration options nor do they account for potential complementarities between flexibility resources of different durations. Overall, the figures show that short duration flexibility is the most effective option for reducing RES curtailment within this simplified assessment framework, particularly at low to moderate capacity levels, while longer-duration flexibility becomes increasingly inefficient when assessed in isolation on the basis of the amount of GWh added.

4.4.2. Quantification of flexibility resources with combined optimisation To quantify flexibility resources while accounting for interactions between multiple flexibility options, an additional joint optimisation is performed in which all flexibility sources are assessed simultaneously. Within this optimisation method the flexibility needs are assessed by explicitly accounting for the timing and duration of RES curtailment events. Curtailment is not treated as a uniform volume but as a sequence of events that can range from short intraday peaks to prolonged surplus periods lasting multiple hours or days. Aligning flexibility sizing with these temporal characteristics ensures that different types of flexibility are deployed where they are most effective. For each flexibility duration category, an optimisation determines the maximum power and energy capacity that can effectively reduce curtailment, subject to operational constraints. These flexibility options are introduced sequentially, starting with long-duration flexibility and moving to shorter durations, each time optimising against the remaining curtailment. The results obtained at the end of this step are referred to as the “f a rf ”. This portfolio includes all flexibility options that can effectively reduce curtailment while being compatible with the temporal structure of the curtailment patterns. However, because the optimisation is performed sequentially by duration category, interactions between flexibility types are not fully captured. In particular, shorter-duration flexibility may also contribute during longer curtailment events if it is not otherwise occupied. For example, two 4-hour flexibility assets could jointly cover a single 8-hour curtailment event, potentially substituting for one dedicated 8-hour flexibility option. Therefore, in a final step, all flexibility options are jointly optimised to minimise total installed energy capacity [GWh] while still meeting the curtailment reduction target. This results in an efficient and coherent flexibility portfolio that reflects both the magnitude and persistence of curtailment events across all simulated system conditions. The results of this stage will be referred to as “optimal portfolio”. In general, this approach has three important implications: •

When aiming for full RES integration, flexibility needs are largely driven by rare extreme curtailment events.

•

Allowing a small amount of RES curtailment can significantly reduce the required flexibility volumes.

•

The resulting portfolios are expected to be dominated by short-duration flexibility resources, potentially complemented by medium-duration flexibility.

The remainder of this section explores these conclusions in detail. The analysis starts by sizing the portfolio under the assumption of full RES integration (i.e. zero acceptable curtailment), which serves as the base case. Subsequently, sensitivities are performed by assessing different levels of (1) acceptable RES curtailment, (2) different round-trip-efficiencies for the different flexibility type options. 4.4.2.1. Optimal portfolio sizing under full RES integration (zero acceptable curtailment) As mentioned in the methodology, the optimal portfolio sizing is composed of two main stages:

FNA report for Belgium 2026, version approved by CREG

1. ‘Feasible portfolio’: This portfolio includes all flexibility options that can effectively reduce curtailment while being compatible with the temporal structure of the curtailment patterns as explained above. Note that this option gives a range of options to choose from rather than minimal installed capacities. 2. ‘Optimised portfolio’: Considering all flexibility options together, determine the minimum combination of capacities required to eliminate curtailment, while adhering to the capacities identified in the feasible portfolio. Resulting capacities required to achieve full RES integration (zero acceptable curtailment) for all scenarios are shown in Figure 35, with a breakdown per flexibility option. In 2030, the optimal portfolio requires 93 GWh in both the CC-AF25 and CT-AF25 scenarios. When the battery potential identified in Adeqflex’25 is considered (BAT+ scenario), the required capacity decreases to 77 GWh and 63 GWh for the CC-BAT+ and CT-BAT+ scenarios respectively. In 2035, the required capacities increase substantially across all scenarios, reaching 318 GWh and 306 GWh for the CC-AF25 and CT-AF25 scenarios respectively, and 278 GWh and 231 GWh for the CCBAT+ and CT-BAT+ scenarios. Relative to 2030, the required capacity increases by a factor of approximately 3.5 across all scenarios, reflecting the increasing magnitude and clustering of excess generation events. The corresponding installed power per flexibility option is shown in Figure 36. In 2030, the optimal portfolios require between 15 GW and 17 GW of installed power across the different scenarios, with limited variation between AF25 and BAT+ assumptions. In 2035, the required installed power increases substantially, reaching between 27 GW and 30 GW depending on the scenario. Despite large differences in installed energy capacities, the resulting optimal portfolios exhibit relatively similar power requirements. It should be noted that the resulting flexibility capacities are very large. This is because eliminating 100% of curtailment implies removing curtailment across all Monte Carlo years, which effectively means sizing the flexibility portfolio to cover extreme events that may occur only once every 25 years. These extreme cases typically arise from the clustering of several large curtailment events in close succession, combined with limited opportunities to discharge or shift energy due to insufficient positive residual load before, between, or after these events. As a result, the flexibility portfolio must be dimensioned to withstand rare but particularly unfavourable system conditions, leading to conservative and potentially oversized capacity requirements. Despite the oversizing of the portfolio, a few observations can be made on the results: • Short- and medium-duration flexibility dominates the portfolios. Across all scenarios and years, most installed capacity is in the 1-hour to 20-hour range. • Long-duration flexibility plays a minor role in optimal portfolios. In the optimal results, long-term flexibility is only selected in the 2035 constrained transition scenario, representing a small share of total capacity (33 GWh out of 306 GWh), indicating that most curtailment can be addressed without extensive long-duration solutions. • Very long-duration flexibility is never selected. Although 200-hour flexibility was included as an option, it is not chosen in either feasible or optimal portfolios, suggesting that curtailment events do not persist long enough to justify such long-duration assets. • Portfolio sizes are smaller in higher battery uptake scenarios. In these scenarios, the battery potential identified in Adeqflex’25 already absorbs part of the excess renewable generation by shifting it towards periods with positive residual load. This reduces both annual curtailment volumes and the severity of individual curtailment events. This is confirmed by the fact that medium- and long-term duration flexibility options experience the largest reductions in installed capacity, while short-duration flexibility requirements remain relatively stable

FNA report for Belgium 2026, version approved by CREG

f r 100

a M

Current commitments

36 Feasible

15 33

2030

185

Feasible

13 25

36 Optimal

15 33

Installed

Optimal

capacity [GWh]

100 150 200 250 300 350 400 450

13 25 27

5 10 Feasible

106

133

425

Feasible

5 10 Optimal

146

Installed

100 150 200 250 300 350 400 450

115

107

106

318

GWh

119

Optimal

278

Constrained Transition

15 37

2030

178

Feasible

12 27

36 Optimal

Installed

100 150 200 250 300 350 400 450

Installed

50 100 150 200 250 300 350 400 450

capacity [GWh]

423

150

Feasible

78 Optimal

117

12 27 15 65

Optimal

capacity [GWh]

2035

15 37

Feasible

364

Feasible

capacity [GWh]

2035

231

285

306

Optimal

20h

50h

100h

Figure 35: Feasible and optimal portfolio installed capacities per flexibility option in GWh for a full RES integration (i.e. zero acceptable curtailment) in all Monte Carlo years for all scenarios f rf

ra a M

. . r a ar ar

)

Current commitments

Feasible

Optimal

Current commitments

1 19

Feasible

Optimal

1 19

2030

Installed

power [GW]

Installed

Feasible

1 31

Feasible

Optimal

power [GW]

2035

Constrained Transition

Feasible

Optimal

Constrained Transition

1 19

Feasible

2 1 17

Optimal

1 15

2030

Installed

power [GW]

Installed

Feasible

2 1 30

Feasible

2 0 28

Optimal

50h

100h

power [GW]

2035

20h

Figure 36: Feasible and optimal portfolio installed power per flexibility option in GW for a full RES integration (i.e. zero acceptable curtailment) in all Monte Carlo years for all scenarios

FNA report for Belgium 2026, version approved by CREG

4.4.2.2. Portfolio sensitivity to tolerated RES curtailment As no national target for acceptable RES curtailment is currently defined, Sections 4.4.1 and 4.4.2.1 first assess the boundary case corresponding to full RES integration (i.e. zero acceptable curtailment). However, such an assumption results in flexibility portfolios sized to cover a limited number of extreme curtailment events. The present section therefore explores intermediate levels of acceptable RES curtailment by incrementally increasing the amount of RES production considered compatible with the RES integration target (e.g. from 0% in steps of 0.1%). This allows assessing how flexibility requirements evolve when part of the curtailment is considered acceptable, thereby illustrating the trade-off between RES integration and additional flexibility requirements. The analysis is performed both for the AF25 scenarios, which reflect the battery capacity currently installed and contracted through CRM auctions up to 2024, and for the BAT+ scenarios, which additionally include the battery potential identified in Adeqflex’25. The number of sensitivities depends on the total curtailment level observed in each scenario and target year. For example, in the 2030 CC-AF25 scenario, total RES curtailment represents approximately 0.8% of total RES production. Consequently, eight additional sensitivities are performed, each increasing the acceptable curtailment level by 0.1 percentage point, as illustrated in Table 7. Similarly, in the 2030 CC-BAT+ scenario, which includes the battery potential identified in Adeqflex’25, total RES curtailment represents approximately 0.4% of total RES production, so that 4 additional sensitivities are performed. The same methodology is applied to the target year and to the constrained transition scenarios, these need a higher number of sensitivities due to the higher amount of curtailment. Base case

Sensitivities to perform

Curtailment allowed [% of total RES production]

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Curtailment allowed [GWh]

102

136

170

204

238

272

Curtailment to be mitigated with additional flexibility in CC-AF25 [GWh]

287

253

219

185

151

117

Curtailment to be mitigated with additional flexibility in CC-BAT+ [GWh]

147

113

Table 7 : Sensitivities performed for year 2030 (CC-AF25 and CC-BAT+ Scenarios)

Figure 37 illustrates the total installed capacity of optimal portfolios per target year and scenario, across the different RES curtailment sensitivities. The very stringent nature of the base case, corresponding to full RES integration (zero acceptable curtailment), is highlighted by the strong reduction in required flexibility capacity that can be achieved by allowing only a limited amount of RES curtailment. In the CC-AF25 and CT-AF25 scenarios, accepting 0.1% RES curtailment reduces the required capacity by approximately 75–80%, from 93 GWh to 21–25 GWh in 2030 and from 306–318 GWh to 72–80 GWh in 2035. In the BAT+ scenarios, which include the battery potential identified in Adeqflex’25, the reduction is even more pronounced, reaching 82.5–87.5%. The required capacity decreases from 77 GWh to 14 GWh in the CC-BAT+ scenario and from 63 GWh to 8 GWh in the CT-BAT+ scenario in 2030, while decreasing from 278 GWh to 49 GWh and from 231 GWh to 46 GWh respectively in 2035. FNA report for Belgium 2026, version approved by CREG

The strong reduction in required installed capacity is driven by the fact that the base-case portfolio is sized to fully cover the most extreme curtailment event. This worst-case event dominates the sizing, while all other curtailment events are smaller in terms of energy content. Allowing even a limited amount of RES curtailment—on average 0.1% of total RES production— substantially reduces the energy volume of this extreme event. This effect is even more pronounced in the BAT+ scenarios an upfront reduction of the most extreme curtailment event’s severity. As a result, the flexibility portfolio no longer needs to be dimensioned for this highly exceptional case, leading to a significantly less conservative and more realistic capacity requirements. The required capacity decreases further as higher levels of RES curtailment are considered acceptable. For example, allowing 0.5% RES curtailment reduces the required capacity to around 4 GWh and 7 GWh in 2030 for the CC-AF25 and CT-AF25 scenarios respectively, and to around 22 GWh and 25 GWh in 2035. In the BAT+ scenarios, where the battery potential identified in Adeqflex’25 is considered, allowing 0.5% of RES production to be curtailed virtually eliminates the need for additional flexibility in 2030. In 2035, the remaining flexibility requirement decreases to around 6 GWh, representing an additional reduction of approximately 85–90% compared to the 0.1% curtailment case. Nonetheless, the marginal reduction in flexibility requirements rapidly diminishes as higher levels of RES curtailment are considered acceptable, indicating that the flexibility requirements under full RES integration are primarily driven by a limited number of extreme curtailment events. Installed capacities [GWh]

RES curtailment allowed [% of total RES production]

0,5%

RES curtailment allowed [% of total RES production]

0% 0,1%

2030

Current commitments

0% 0,1%

21 13 9 6 4 3 1

23 16 12 9 7 6 4

0,5%

1,0%

Installed capacity

0 0% 0,1%

0,5%

2035 1,0%

1,5%

0% 0,1%

47 35 27 22 18 15 13 10 8 7 5 4 3 2 1 0

1,0%

1,5%

2,0%

RES curtailment allowed [% of total RES production]

0,1%

20h

Current commitments

Installed capacity

100

150

200

250

300

0,1%

Installed capacity

150

200

250

300

[GWh]

0,2%

0,3%

100

231

0,1%

0,3%

0,2%

278

0,4%

0,3%

[GWh]

Constrained Transition

0,1%

0,2%

100h

[% of total RES production]

0,3%

50h

[GWh]

306

RES curtailment allowed

0,2%

2035

2030

50 100 150 200 250 300 350 400 450 500

50 38 30 25 22 19 16 14 12 11 9 8 7 6 5 4 3 2 2 1 0

0,5%

Installed capacity

318

3 2 1 0

1,0%

[GWh]

50 100 150 200 250 300 350 400 450 500

Constrained Transition

0,4%

0,5%

0,6%

0,7%

0,8%

0,9%

20h

50h

Figure 37: Resulting installed capacity per flexibility option per target year and scenario FNA report for Belgium 2026, version approved by CREG

Another important aspect of the additional flexibility resources sizing is the consequently installed power. Figure 38 shows the installed power per flexibility option, together with the maximum used power, i.e. the maximum total net power used by the flexibility portfolio across all timesteps and Monte Carlo years to shift excess renewable generation. It can first be noted that allowing 0.1% RES curtailment leads to a much smaller reduction in required power capacity than in required energy capacity. While the required energy capacity decreases by approximately 75–80% in the AF25 scenarios and by 82.5–87.5% in the BAT+ scenarios, the corresponding reduction in installed power is limited to around 35–50%. In the CC-AF25 and CT-AF25 scenarios, installed power decreases from 16 GW to 9 GW and 10 GW respectively in 2030, and from 30 GW to 20 GW and 21 GW respectively in 2035. In the BAT+ scenarios, which include the battery potential identified in Adeqflex’25, installed power decreases from 17 GW to 8 GW in the CC scenario and from 15 GW to 7 GW in the CT scenario in 2030. In 2035, the corresponding reductions are from 29 GW to 18 GW and from 27 GW to 17 GW respectively. This difference stems from the composition of the optimal portfolios. In the base case, installed power capacity is largely dominated by short-duration flexibility, accounting for around 85% of total installed power capacity in 2030 and 75% in 2035. In the BAT+ scenarios, these shares increase further to approximately 95% and 90% respectively, reflecting the fact that the battery potential identified in Adeqflex’25 already mitigates a large share of the medium- and long-duration curtailment events. Since the flexibility options that are removed when allowing 0.1% of RES curtailment are mainly medium- and long-duration options, their contribution to total installed power capacity is relatively limited. Consequently, the overall reduction in required additional power capacity is mechanically constrained and remains significantly smaller than the reduction observed for energy capacity.

FNA report for Belgium 2026, version approved by CREG

Installed power [GW] Current commitments

RES curtailment allowed

[% of total RES production]

[% of total RES production] 0.0% 0.1%

0.0% 0.1% 7

6 5

0.5%

10 8

2030

Constrained Transition

RES curtailment allowed

0.5%

3 2

1.0%

Installed power

0.0% 0.1%

Installed power

0.5%

14 13 12 11 11 10

1.0%

7 6

1.5%

[GW]

32 30

1.0%

11 10

2035

0.5%

0.0% 0.1%

1 0

[GW]

1.5%

4 3

2.0%

2 1 0

20h

Current commitments

RES curtailment allowed

0.1%

[GW]

0.7%

0.8%

[GW]

8 5

0.6%

0.5%

0.7%

0.4%

0.6%

0.3%

0.5%

0.2%

0.4%

0.1%

0.3%

Installed power

0.0%

0.2%

0.1%

0.9%

0.3%

Installed power

0.0%

2035

0.2%

0.3%

0.0%

0.2%

Constrained Transition

[% of total RES production]

0.0% 0.1%

100h

RES curtailment allowed

[% of total RES production]

2030

50h

2 1

0.8%

0.9%

20h

50h

Figure 38: Optimal additional installed power per flexibility option per target year and scenario

The amount of additional installed power of flexibility might seem surprisingly high but can be explained by two effects: •

For higher levels of acceptable RES curtailment, the optimisation increasingly becomes an energy-capacity problem rather than a power-capacity problem. Once installed power capacity exceeds the maximum curtailment power or the maximum residual load value, adding further power capacity provides little additional benefit. At that stage, additional RES curtailment can primarily be reduced by increasing the energy capacity available to shift excess generation over longer periods.

•

For low levels of acceptable RES curtailment, the remaining curtailment volume is relatively small and consists mainly of short-duration events. Under these conditions, short-duration flexibility options, particularly the 1-hour duration option, are favoured by the optimisation as

FNA report for Belgium 2026, version approved by CREG

they minimise the required installed energy capacity while providing relatively high power capacity per unit of energy capacity. As a result, relatively large installed power capacities remain necessary even when the required installed energy capacities decrease significantly. Optimal portfolio sensitivity to round-trip-efficiency The assumed round-trip efficiency (RTE) of each flexibility option has a direct impact on the sizing of flexibility needs. This effect is present both when defining the feasible portfolio—where the maximum installable capacity per flexibility option is determined—and when deriving the optimal portfolio, where all flexibility options are jointly optimised. To assess the influence of RTE assumptions, a sensitivity analysis was performed in which all flexibility options were assigned the same round-trip efficiency of 85%, corresponding to the value assumed for the shortest-duration flexibility option. In the reference case, short-, medium- and long-duration flexibility options are assumed to have round-trip efficiencies of 85%, 75% and 60% respectively. The analysis was conducted for the 2035 constrained transition (CT) scenario, as this is the only scenario in which a meaningful share of long-duration flexibility is selected in the reference case and therefore the only case where the effect of the RTE assumptions can be clearly observed. The results in Figure 39 show that the assumed RTE mainly affects the role of long-duration flexibility options in portfolios targeting full RES integration. When all flexibility options are assigned a round-trip efficiency of 85%, the required 100-hour flexibility capacity decreases from 150 GWh to 74 GWh in the feasible portfolio and from 33 GWh to 12 GWh in the optimal portfolio under full RES integration (0% acceptable curtailment). While this may appear counter-intuitive given the higher assumed efficiency of long-duration options, it can be explained by the objective of the optimisation, which focuses on eliminating curtailment. In this context, a lower round-trip efficiency can actually be beneficial, as energy losses effectively dissipate surplus energy and reduce the need for additional discharge opportunities. The reduction in long-duration flexibility is mainly compensated by an increased deployment of medium-duration flexibility, which remains attractive even under the higher efficiency assumption, similar to the reference case. However, the influence of the RTE assumptions rapidly diminishes once a limited amount of RES curtailment is considered acceptable. For the 0.1% curtailment sensitivity, the resulting optimal portfolio remains almost unchanged compared to the reference case, both in terms of installed energy capacity and installed power capacity.

Figure 39: Installed capacity and power per flexibility option for RTE sensitivity in 2035 (CT scenario)

FNA report for Belgium 2026, version approved by CREG

The sensitivity was not performed for the BAT+ scenarios, as the battery potential identified in Adeqflex’25 already substantially reduces the need for medium- and long-duration flexibility and, in some cases, virtually eliminates the need for additional flexibility capacity. As a result, the impact of alternative RTE assumptions is expected to be even more limited than in the AF25 scenarios 4.4.2.3. Trading off flexible resources against renewable capacity As discussed in the previous section, achieving the RES integration target does not necessarily require all curtailed energy to be integrated through additional flexibility. An alternative approach consists in increasing renewable deployment such that, despite some level of curtailment, the effective renewable generation still achieves the RES integration target. In order to avoid negative implications for system security, which are addressed in Section 6, any additional renewable capacity considered in this analysis is assumed to be able to reduce its output during periods of excess generation. The additional renewable capacity required to achieve a given level of acceptable RES curtailment is calculated by combining the wind and solar capacity factors with the simulated curtailment time series. During periods of curtailment, the effective capacity factor for this additional capacity is assumed to be zero, as any additional generation produced during these periods would also be curtailed and therefore would not contribute to the RES integration target. Accordingly, the analysis assumes that any additional renewable capacity deployed to compensate for curtailed energy should be flexible and therefore will curtail during moments when curtailment is already observed. The resulting impact on average capacity factors is summarised in Table 8. Across scenarios and target years, both wind and solar experience a reduction in effective capacity factor of roughly 1 to 3 percentage points due to curtailment for AF25 scenarios. This decreases to 0.5 to 2.5 percentage points in the BAT+ scenarios, where fewer curtailment hours limit the impact on capacity factors. These values should be interpreted as a minimum possible reduction, as they only account for whether curtailment already occurs and do not assess whether additional capacity could be added without creating new curtailment moments.

Table 8: Change in (percentage point) capacity factor for solar and onshore wind for additional capacities in the system

FNA report for Belgium 2026, version approved by CREG

Figure 40 illustrates the additional capacity of either solar or wind that would need to be installed to compensate for allowed curtailment levels, expressed as a share of total renewable generation. For 2030 under the current commitment scenario (CC-AF25), allowing 0.8% of total RES generation to be curtailed requires approximately 0.4 GW of additional solar capacity or around 0.1 GW of additional wind capacity. The effect becomes considerably stronger in 2035 due to higher renewable penetration and increased curtailment volumes. Under current commitments (CC-AF25), allowing 1.6% curtailment requires approximately 1.1 GW of additional solar capacity or around 0.4 GW of wind capacity. The same relationship between acceptable curtailment and required additional renewable capacity is obtained for the BAT+ scenarios, which include the battery potential identified in Adeqflex’25. However, the range of curtailment levels considered is smaller due to the lower total curtailment observed in these scenarios. The vertical line indicates the investment threshold for the BAT+ scenario. Comparing scenarios, the constrained transition case consistently shows a lower impact on onshore wind capacity factors and lower additional onshore wind capacity requirements. This is explained by the lower installed onshore wind capacity in the constrained transition scenario, allowing additional wind to be integrated with relatively less curtailment. Overall, Figure 40 and Table 8 indicate that RES integration target can be maintained by installing a relatively limited amount of additional onshore renewable capacity. For example, in 2030 under the CC scenario, allowing 0.8% of RES generation to be curtailed requires approximately 0.4 GW of additional solar capacity or around 0.1 GW of additional onshore wind capacity. In 2035, allowing 1.6% curtailment requires approximately 1.1 GW of additional solar capacity or around 0.4 GW of additional onshore wind capacity. These values remain limited when compared with the substantially larger additional flexibility capacities required under the boundary case corresponding to full RES integration (zero acceptable curtailment). The results therefore illustrate that increasing renewable deployment can constitute an alternative approach to maintaining the RES integration target when some level of RES curtailment is considered acceptable.

Figure 40: Necessary additional capacity of Solar or Onshore Wind to install to cover the allowed curtailment FNA report for Belgium 2026, version approved by CREG

4.4.2.4. Conclusion Overall, the joint optimisation framework provides a quantification of flexibility needs by explicitly accounting for the timing, duration, and clustering of RES curtailment events, as well as interactions between different flexibility options. Figure 41 combines the main conclusions from this section. The results show that flexibility sizing under full RES integration is largely driven by a limited number of extreme curtailment events. Allowing even a very limited amount of RES curtailment—as little as 0.1% of total RES production—reduces required energy capacity by 75–80% in the AF25 scenarios and by up to 87.5% in the BAT+ scenarios, which explicitly incorporate the battery potential identified in Adeqflex’25, while still maintaining very high integration levels. Further increases in acceptable curtailment yield progressively smaller reductions in flexibility requirements, confirming that flexibility needs under full RES integration are dominated by the most extreme events. Across all scenarios, the resulting portfolios (to avoid curtailment) are primarily composed of short-duration flexibility options, complemented by medium‑duration flexibility in some cases, while long‑duration options play only a minor role and very long‑duration flexibility is never selected. This reflect the fact that most curtailment events are concentrated over relatively short periods and that flexibility requirements are strongly influenced by periods of high renewable generation, particularly around hours of elevated solar PV production. Together, these results highlight that modest curtailment allowances can drastically reduce additional flexibility requirements and lead to more realistic, efficient portfolios without materially undermining renewable integration objectives. It should also be noted that, if this flexibility were deployed solely to achieve renewable targets, its utilisation would remain very limited, with activation occurring only during a small number of hours per year. The BAT+ scenarios, show substantially lower flexibility requirements than the AF25 scenarios. In the BAT+ scenarios, allowing limited levels of RES curtailment virtually eliminates the need for additional flexibility in 2030. In 2035, the presence of the potential battery capacity halves the level of allowed curtailment necessary to remove the need for additional flexibility sources compared to the AF25 scenarios. This result is particularly relevant in the context of the FNA, whose objective is to identify potential flexibility needs beyond those already addressed through adequacy-related investments. While the AF25 scenarios, which only include battery capacities installed and contracted through CRM auctions up to 2024, still exhibit a limited residual flexibility need, the BAT+ scenarios indicate that these residual flexibility needs become very limited when the battery potential identified in Adeqflex’25 is considered. Finally, the analysis shows that maintaining the RES integration target does not necessarily require all curtailed energy to be integrated through additional flexibility resources. Increasing renewable deployment can also compensate for curtailed generation, with the required additional renewable capacity remaining relatively limited compared with the flexibility capacities required under the boundary case corresponding to full RES integration. The results therefore illustrate the trade-off between additional non-fossil flexibility resources, acceptable RES curtailment and additional renewable deployment when pursuing a given RES integration target.

FNA report for Belgium 2026, version approved by CREG

Current commitments

Additional flexibility requirements needed

Additional renewable capacity needed

RES curtailment allowed [% of total RES production] 0,0 0,0

0.1%

2030

0,2

0.5%

0,1

1.0% 1.5%

Installed power [GW]

0,0

2035

Installed power

1,5

2,0

Installed power

[GW]

0,7 0,2 1,1

1.5%

2,0

0,1

1.0%

1,5

0,4

0.5%

1,0

0,1 0,0

0.1%

0,5

0,4

20h

Additional flexibility requirements needed

Current commitments

Additional renewable capacity needed

RES curtailment allowed [% of total RES production] 0,04

0.1%

2030

0,02

0,08

0.2%

0,03

0,13

0.3%

0,05

Installed power [GW]

0,0

0,5

1,0

[GW]

0,1

0.1% 0,0

0.3%

2035

0.6%

0,2 0,1 0,4 0,1 0,6

0.9% 0,2

Figure 41: Additional flexibility requirements and additional renewable capacities needed in 2030 and 2035 to maintain the RES integration target across acceptable RES curtailment levels (CC scenarios)

FNA report for Belgium 2026, version approved by CREG

5. Results – Ramping needs The indicator of the ramping needs represents the amount of flexibility needed to cover the expected variation of the hourly residual load, taking into consideration the technical constraints of available flexible generation units. As the economic dispatch simulations already account for the hourly ramps of the residual load, including cross-border network constraints and abilities of units in other bidding zones to ramp on hourly basis, it is expected that this analysis confirms the ability of the system to capture these hourly variations, unless during moments with scarcity or incompressibility: • during normal conditions, the system is able to access sufficient energy locally or cross-border to cover hourly ramps, represented by the economic dispatch simulations. • during scarcity or incompressibility, insufficient local flexibility may be available to manage the corresponding upward or downward ramps, while the availability of cross-border flexibility is constrained. The analysis begins by determining the required ramping flexibility needs. Next, the available means are quantified and compared against these needs, enabling the identification of any uncovered ramping needs. The methodology is described in detail in Annex B.4.

5.1. Ramping needs The assessment of ramping flexibility needs follows the steps outlined in Figure 42. The starting point is the calculation of the residual load. The residual load is computed as defined in section 4.1 by subtracting the generation from energy sources (offshore wind, run-of-river, onshore wind, solar, and other renewables), as well as the must-run output from thermal units, from the total load. In line with Article 9(1) of the FNA method, the ramping flexibility needs are defined as the hourly changes in residual load between two consecutive timestamps 𝑅𝐿𝑡+1 − 𝑅𝐿𝑡 .

Figure 42: Illustration of the ramping needs, ramping means and uncovered ramping needs

Figure 43 illustrates the probability distribution of residual load variation for both the CC and CT scenarios in 2030 and 2035. The residual load variation is not impacted by the choice between the AF25 and BAT+ scenarios, as the difference in battery capacity assumed between the AF25 and BAT+ scenarios does not affect the underlying residual load calculation. Consequently, the results presented FNA report for Belgium 2026, version approved by CREG

in this section apply equally to both AF25 and BAT+ scenarios, and only a single distribution is shown for each CC and CT scenario. While the overall pattern remains relatively consistent between 2030 and 2035, the magnitude of the variations becomes more pronounced in 2035. This trend can be attributed to changes in the main components of the residual load (demand, and the different types of renewable generation). The overall electricity demand is expected to increase, resulting in larger hourly variation. In addition, the installed capacity of renewable energy sources is projected to increase, thereby amplifying the impact of variability in generation from these sources. Regarding the differences between the CC and CT scenarios, the CT scenario assumes lower renewable energy capacity and lower total demand. As a result, it exhibits reduced residual load variation compared to the CC scenario. Figure 43 also highlights that large residual load variations are linked to solar generation. The main drivers are investigated further in the following analysis.

Figure 43: Distribution of the hourly residual load variation for CC and CT in 2030 and 2035

To understand the drivers of the residual load variation, the contribution of the main drivers of the residual load solar, wind and load is shown in Figure 44 for the simulated year 2030. The resulting figure for 2035 shows similar patterns and is therefore not shown separately. The other category represents the flexibility associated with heat pumps, industry, and electrolysers. The figure ranks the hourly residual load variations from small to large and looks at the contributions of the individual technologies FNA report for Belgium 2026, version approved by CREG

in these variations (using their individual variations). For the low residual load variations, the main drivers are wind power variation and load variation. But for the large residual load variations, the main driver is pre-dominantly solar power variations. For 2030, the solar variation contributes up to 87 % to positive residual load variations and up to 79 % to negative residual load variations. This dominant contribution of solar power to the highest residual load variations can be attributed to the steep fluctuations in solar generation, particularly during morning and evening periods, combined with a large installed solar capacity.

Figure 44: Drivers of the residual load variations for Current commitments in 2030

5.2. Characterisation of ramping capacity As shown in Figure 45, the residual load variations are compared with the ramping residual margins. The latter concerns available flexibility and encompasses generation, storage, end-user flexibility, demand response, and interconnections. These are assessed based on their respective dispatches or schedules, taking into account the technical constraints specific to each technology, as outlined in the short-term flexibility framework. Importantly, the calculation of flexible resources excludes any capacity set aside for balancing purposes. In line with Article 9(5) of the FNA method, the calculations account for the possibility to ramp up capacities prior to their dispatch or offtake to cover the needs. Available cross-border flexibility is regarded as fully available except during periods of market tightness (incompressibility and scarcity). Scarcity is identified using the demand shedding (LOLE), while incompressibility is determined based on the RES curtailment indicator described in Section 4.2. During such periods, the potential for constrained cross-border flexibility is considered, reflecting reduced market liquidity. As shown in Figure 45 (up), available flexibility drops sharply during the sunniest hours and months. This is largely because high solar output forces most thermal units to already operate at reduced output. Additionally, low electricity prices during these sunny periods incentivize battery charging, which further reduces remaining downward flexibility available. As previously discussed, the primary driver of largest residual load variations is the fluctuation in solar power but solar production also significantly reduces the availability of remaining downward flexibility for ramping.

FNA report for Belgium 2026, version approved by CREG

Figure 45: Hourly average available downward (down) and upward (up) flexible resources excluding cross-border flexibility for CC-AF25 (left) and CC-BAT+ (right) in 2030

Although to a lesser extent, the influence of demand can also be observed in Figure 45 (left), as available flexibility increases during morning and evening peaks, when more thermal units are usually running. Note that a similar trend is obtained for the other scenarios considered. In contrast, the impact of solar leads to an increase of the average upward flexible resources, as seen in Figure 45 (down). Indeed, the high solar production leads to the charging of the batteries and to the shutdown or modulation of thermal units. At these moments, batteries can provide upward ramping flexibility by stopping their charging and/or starting their discharging. Thermal units can equally ramp up or start up again, under the assumption they anticipate the need of the market. Consequently, the high solar production may even increase available upward flexibility. The profiles of average upward and downward flexible resources are broadly similar across the AF25 and BAT+ scenarios. Nevertheless, the BAT+ scenarios explicitly incorporate the battery potential identified in Adeqflex’25. Additional battery capacity increases both upward and downward ramping capabilities and therefore increases the available ramping residual margins as can be seen in the average residual magins in Figure 45.

5.3. Uncovered ramping needs Once the ramping flexibility needs and the available flexible resources are calculated, uncovered ramping needs can be calculated on an hourly basis by identifying the situation where the available flexible resources are not sufficient to cover the ramping flexibility needs. As the uncovered needs are expected to occur during periods of scarcity or incompressibility, the assessment focuses on these conditions. During normal conditions, economic dispatch simulations do not demonstrate flexibility shortages through demand shedding (LOLE) or RES curtailment and no ramping needs are expected to occur. Figure 46 shows the percentage of time during which there are no uncovered ramping needs. Ramping requirements are almost always covered (above 99.7% of time) when looking at all hours of year across FNA report for Belgium 2026, version approved by CREG

all scenarios for both 2030 and 2035. In the BAT+ scenarios, almost all occurrences of uncovered ramping needs disappear. This contrasts with the AF25 scenarios. The additional battery potential included in BAT+ increases both upward and downward ramping residual margins, thereby eliminating the limited number of uncovered ramping situations observed in the AF25 scenarios. Figure 47 shows the average number of hours per year during which uncovered ramping needs occur, for both downward and upward directions. Upward uncovered ramping needs arise less than one hour per year in the AF25 scenarios. The vast majority of uncovered ramping needs occur on downward flexible resources and only on the AF25 scenarios (9 - 23 hours depending on the scenario and target year). These situations result from insufficient remaining flexibility during incompressibility situations: •

Downward storage flexibility is unavailable, as storage assets are already charging;

•

Limited flexibility from thermal units, which are already operating at minimum output (or not dispatched and hence not being able to provide ramping flexibility);

•

No cross-border flexibility available (regional events).

In the BAT+ scenarios, which explicitly incorporate the battery potential identified in Adeqflex’25, uncovered ramping needs are reduced to negligible levels (<1 hour per year) and disappear entirely in some scenarios.

Figure 46: Proportion of events without uncovered ramping needs

Figure 47 : Average number of hours per year with uncovered ramping needs

Figure 48 presents, for each scenario and target year, the maximum ramping need derived from residual load variation, the 99.9th percentile of uncovered downward ramping needs, and the total volume of FNA report for Belgium 2026, version approved by CREG

uncovered energy for downward ramping needs. The BAT+ scenarios substantially reduce the occurrence of uncovered ramping needs and the associated volume of uncovered energy, which falls below 1 GWh in all cases.

Figure 48: Maximum ramping needs, maximum uncovered capacity and uncovered energy for each scenario and target year for downward

5.4. Flexible resources needed to meet uncovered ramping needs The uncovered ramping needs assessment indicates that flexibility shortages may occur during lack of available flexibility resources, particularly on the downward side, and very rarely on the upward side. In the AF25 scenarios, which only include battery capacities installed and contracted through CRM auctions up to and including 2024, a limited number of uncovered ramping situations remain. This is consistent with the fact that the adequacy gap identified for 2035 is not covered in these scenarios. As a consequence, the flexibility resources represented in the AF25 scenarios are not sufficient to fully address all ramping needs under scarcity and incompressibility conditions. By contrast, in the BAT+ scenarios, which explicitly incorporate the battery potential identified in Adeqflex’25, uncovered ramping needs are reduced to negligible levels. This suggests that the battery potential identified in Adeqflex’25 is sufficient to bridge the remaining flexibility gap observed in the AF25 scenarios and to cover the identified ramping needs. The results therefore suggest that the limited upward ramping shortages observed in the AF25 scenarios are primarily linked to the remaining adequacy gap rather than to a distinct ramping issue. Once the battery potential identified in Adeqflex’25 is taken into account, these shortages disappear, indicating that the flexibility resources required to address adequacy needs are also sufficient to cover the corresponding upward ramping needs. As a result, the analysis does not point to a significant FNA report for Belgium 2026, version approved by CREG

residual upward flexibility need beyond the flexibility development required to address the adequacy gap. When looking at Hitzenflaute events, i.e. situations characterised by a sharp upward ramp in residual load resulting from declining renewable generation combined with increasing demand, the economic dispatch simulation does not indicate demand shedding or ramping needs. This reflects the perfectforesight assumption applied in the dispatch model, whereby flexible resources can anticipate such situations and position themselves accordingly, for instance through day-ahead scheduling decisions. Similarly to upward flexibility, downward ramping shortages occur primarily during incompressibility situations. In the AF25 scenarios, a limited number of uncovered downward ramping situations remain. By contrast, in the BAT+ scenarios, these shortages are reduced to negligible levels. This suggests that the battery potential identified in Adeqflex’25 is sufficient to bridge the remaining flexibility gap observed in the AF25 scenarios. More generally, the results indicate that the measures and flexibility resources required to address incompressibility situations are also sufficient to address the corresponding hourly ramping needs. More generally, the results indicate that the flexibility resources required to address adequacy needs and incompressibility situations are also sufficient to address the associated hourly ramping needs. As a result, the analysis does not point to a significant residual ramping-related flexibility need beyond the flexibility resources already required to ensure adequacy and manage incompressibility.

FNA report for Belgium 2026, version approved by CREG

6. Results - Short-term flexibility needs This section on short-term flexibility needs discusses the needs of the system associated with unexpected variations such as forecast errors and forced outage of assets occurring during the intraday or balancing timeframe. This analysis complements the ‘perfect-foresight’ assumptions applied in the economic dispatch simulations used in Section 4 and therefore covers flexibility needs from the dayahead timeframe to real-time including reserve capacity close to real-time, which are not reflected in the RES integration assessment. • As short-term flexibility needs aim to address all forecast errors and forced outages in the system, part of the short-term needs will be covered by the frequency restoration reserve (‘FRR’) capacity pursuant to Article 157 of Commission Regulation (EU) 2017/1485. • Downward ramping needs and short-term flexibility needs - leading to RES curtailment - are calculated first and then added to the RES curtailment time series used to calculate the RES integration needs, in order to avoid overlaps and double counting between these indicators. Annex B specifies the methodology for the short-term needs assessment in detail. Figure 49 illustrates how the section starts with a description of the short-term flexibility needs (section 6.1) based on a calculation of largest forecast errors, variations and outages at worst case percentiles. In a second step, the available short-term flexibility means (Section 6.2) are identified based on a calculation of residual margins on economic dispatch results (per asset, per hour, per Monte Carlo year) to react in the intraday and balancing time frames.

Figure 49 : Calculation steps on short-term flexibility needs

This allows in a third step to identify the uncovered short-term flexibility needs (Section 6.3) comparing the residual margins with the flexibility needs in terms of hours at risk and volumes at risk. In the last step, flexibility resource needs (Section 6.4) are identified as the capability of the different sources of flexibility to meet uncovered needs quantified: • in first instance, the expected up- and downward reserve capacity needs are assessed that are required by Elia to cover flexibility needs related to system imbalances in line with European legislation and regulatory requirements. • In a second instance, the remaining flexibility needs are assessed which require to be dealt with through additional measures.

6.1. Short-term flexibility needs 6.1.1. Quantification of the flexibility needs This section quantifies the flexibility needs associated with unexpected variations of the residual load. In other words, the need for market players and the transmission system operator to cover unexpected variations in demand and generation, including prediction errors relating to renewable generation and periods during which generation and transmission assets are subject to forced outages. These flexibility needs are represented through three categories, as outlined below: • Very fast flexibility that needs to be activated within 5 minutes before real-time. It is activated to deal with real-time variations in generation and demand. These needs are covered by the TSO’s fast aFRR reserves as well as market actor actions within the imbalance settlement period of 15 minutes. Uncovered needs translate into area control errors (FRCE – Frequency Restoration Control Errors). FNA report for Belgium 2026, version approved by CREG

• Fast flexibility that needs to be activated within 15 minutes before real-time. It is activated to deal with real-time forecasting errors related to generation and demand (when compared with the last intraday forecasts received) and to deal with the forced outages of generation and transmission assets. These needs are generally covered by the TSO’s slower mFRR reserves as well as market actor actions, mainly in the balancing market but also the intraday market time frame. Uncovered needs are covered with exceptional balancing measures or translate as well into area control errors. • Slow flexibility needs to be activated within a few hours before real-time. It is activated in order to deal with intraday forecast updates, as well as forced outages that last longer than a few hours. These needs are covered by market actor actions in the intraday market time frame. The relation between the different needs was already shown in Section 2 (Figure 12). Total short-term flexibility needs capture the overall requirement to manage uncertainty between the day-ahead and real-time time frames. They reflect the combined effect of fast and slow flexibility needs, recognizing that these are not entirely cumulative because the largest intraday forecast errors do not occur at the same time as the largest last forecast errors. In this sense, very fast flexibility needs are part of the fast flexibility needs, while the fast flexibility needs are part of the total flexibility needs (defined by fast and slow flexibility needs together). Figure 50 shows the resulting up- and downward flexibility needs for the CC and CT scenario in 2030 and 2035 assessed at the 99.9 percentile across all system conditions. Note that the AF25 versus BAT+ scenarios do not impact the short-term needs, as these needs are driven by forecast errors of renewable generation and demand, as well as forced outages of conventional units. The additional battery capacity considered in the BAT+ scenario affects the availability of flexibility resources to meet these needs, but does not change the underlying uncertainty that generates them. • Total flexibility needs amount up to 4650 MW (upward) and 3400 MW (downward) to manage uncertainty between day-ahead and real-time, of which • Slow flexibility needs amount up to 4000 MW (upward) and 3000 MW (downward) requiring activations up to 5 hours before real-time • Fast flexibility needs amount up to 2700 MW (upward) and 2400 MW (downward) requiring activations up to 15 minutes before real-time; of which • Very fast flexibility, amount up to 400 MW (upward) and 450 MW (downward), requiring activations up to 5 minutes before real-time. The flexibility needs in the CT scenario are slightly lower, up to 160 MW lower for total, 120 MW for slow and fast, and 10 MW for very fast flexibility compared to the CC scenario. This is due to lower installed wind power capacity. Note that this figure includes 2026 as a reference point (calculated similarly to 2030 and 2035) to demonstrate that total short-term flexibility needs increase by 37 – 42 % compared to 2026.

FNA report for Belgium 2026, version approved by CREG

Figure 50: Annual flexibility needs in 2030 and 2035 across all system conditions for the CC and CT scenarios

Figure 51 illustrates the relationship between flexibility needs and system conditions, more specifically the residual load. The results indicate that absolute forecast errors and therefore flexibility needs, increase during periods of low residual load and decrease as residual load increases. Low residual load typically coincides with high renewable penetration and low demand, conditions under which system uncertainty is dominated by the intrinsic variability of weather-dependent generation. As a result, higher renewable generation shares lead to larger forecast errors in both upward and downward directions. In contrast, higher residual load reflects a greater reliance on dispatchable generation, which reduces forecast error risks and thus flexibility needs. This relationship is implemented for the calculations in this study and the impact on the flexibility needs is shown by Figure 52 for the CC scenario, while the CT scenario shows very similar results. The comparison aims to illustrate how flexibility needs vary under different system conditions. It should not be interpreted as evidence that renewable generation alone is the source of flexibility requirements, as several other uncertainty drivers are included in the assessment. More information on modelling the relationship between flexibility needs and system conditions can be found in Annex B.1.3

Figure 51: Flexibility needs in 2030 for the CC scenario linked to various levels of the system residual load

FNA report for Belgium 2026, version approved by CREG

Figure 52: Flexibility needs during High RES / Low Load and Low RES / High Load system conditions quantified as the 10% lowest and 10% highest residual load respectively in the CC scenario

6.1.2. Duration of forecast errors Some technologies which can provide fast flexibility such as battery storage and demand response face constraints in terms of the duration during which this flexibility can be delivered. For this reason, they are also referred to as energy-limited technologies. As flexibility providers with slower response times may only replace the fast flexibility providers after an activation time of up to 5 hours (e.g. after the activation of a thermal unit) it is important to assess the maximum duration times of large forecasting errors. Figure 53 shows the probability of long-duration forecasting errors for 2030 and 2035 in the CC scenario, while the CT scenario shows similar results. The assessment is based on the day-ahead forecast errors (used to calculate the total flexibility needs), i.e. data which determines the prediction risk in the flexibility needs (extrapolation of the sum of the observed forecast errors).

Figure 53: Probability of facing day-ahead forecasting errors larger than 500 MW or 1000 MW with 1 up to 4 hour durations or more than 4 hours in the CC scenario.

Positive forecast errors (shortages) which are larger than 1,000 MW and last for longer than 4 hours occur only 3.4% of time in 2035. Negative forecasting errors (excess) which are larger than 1000 MW and last for longer than 4 hours occur only 1.7% of the time in 2035. The probability of facing a longlasting excess following a prediction error which is larger than 1000 MW increases between 2030 and 2035, but its frequency remains limited around 1-2% of the time. This confirms the potential of energy limited technologies such as battery storage and demand response for covering short-term FNA report for Belgium 2026, version approved by CREG

flexibility needs. However, when looking at forecast errors with a smaller magnitude, but still exceeding 500 MW these can have longer durations.

6.1.3. Sensitivity on the yearly forecast improvements Historical residual forecast errors are adjusted to reflect expected improvements in forecast accuracy over time. The general results in this report account for an average cumulative improvement factor of 1% per year. To understand the impact of this assumption, a sensitivity analysis is conducted on the CC scenario, considering both: • the absence of forecast improvements (0% yearly improvement) and; • a higher improvement rate (2% yearly improvement). Figure 54 shows that changing the yearly improvement factor by such percentage point affects the flexibility needs. By 2035, it can result in an increase or decrease of up to 425 MW (upward) and 325 MW (downward) for total flexibility needs. For slow flexibility needs, the impact reaches 375 MW (upward) and 300 MW (downward), while for fast flexibility needs the variation amounts to 250 MW (upward) and 250 MW (downward). The corresponding increase or decrease in very fast flexibility needs is approximately 50 MW. As expected, the effects of this sensitivity analysis are larger in 2035 compared to 2030. The sensitivity analysis highlights that the quantified flexibility needs are sensitive to assumptions regarding future forecast improvements. Actual flexibility needs may therefore differ from the values presented in this assessment depending on future developments in forecasting methodologies and data availability. Additionally, it highlights the importance of achieving future forecast improvements, as they directly impact the required level of flexibility needed to cover prediction errors and forced outages.

Figure 54: Annual flexibility needs in 2030 and 2035 (CC) with a sensitivity on the yearly forecast improvement

6.2. Characterisation of short-term flexibility margins The residual margins allow the available flexibility related to each individual unit (such as gas-fired power plants and pumped hydro storage) or aggregated technologies (such as industry demand-side response and electric vehicles) to be assessed for each hour, based on the results of the economic dispatch simulations and the technical ability to provide short-term flexibility. More information on the method can be found in Annex B.2. Figure 55 (top left) shows the average available fast upward flexibility margins across all hours in 2030 and 2035 for the CC-AF25 scenario. Given that AF25 only includes battery capacities secured through CRM auctions up to and including CRM 2024, flexibility remains relatively diversified across FNA report for Belgium 2026, version approved by CREG

technologies. Where pumped hydro and thermal flexibility generally relate to existing flexibility today, part of industry flexibility, large-scale batteries and end-users (electric vehicles, heat pumps and residential batteries) are expected to play a larger role towards 2030 and 2035. The top three contributors in 2035 (on average) to upward fast flexibility are industry flexibility and large-scale storage through batteries and pumped hydro. In 2035, industry is on average becoming the largest contributor with 27% or 1300 MW of upward fast flexibility of which; around 800 MW from existing demand side response (DSR), 400 MW from new large-scale loads (mainly e-boilers and e-ovens), and the remainder from electrolysers. For downward fast flexibility in Figure 55 (top right) the average margins across all hours in 2030 and 2035 (CC-AF25) show that flexibility from thermal assets decreases while end-user flexibility and flexibility from large-scale RES gain in importance following higher installed capacities in the scenario. The top 3 contributors in 2035 (on average) to downward fast flexibility are large-scale RES, thermal and battery flexibility. In 2035 large-scale renewables are on average the largest contributor with 24% or 1300 MW mainly driven by offshore wind. In Figure 55 (bottom) the CT-AF25 scenario shows lower flexibility from end-users and industry for both upward and downward fast margins compared to the CC-AF25 scenario. For thermal downward fast flexibility, both in relative and in absolute terms flexibility is higher in the CT-AF25 than the CC-AF25 scenario due to the increased running hours, where thermal assets can ramp down to provide downwards flexibility. Furthermore, note that very fast and total flexibility margins in both the CC and CT scenario show similar trends as all new short-term flexibility from industry (with the exception of electric arc furnaces), end-users and large-scale batteries is by definition fast.

Figure 55: Average up and downward fast flexibility margins for CC-AF25 (top) and CT-AF25 (bottom) per flexible load and generation technology delivered per hour FNA report for Belgium 2026, version approved by CREG

Subsequently, as illustrated in Figure 56 (top) total average contributions to short-term flexibility are higher in the CC-BAT+ scenario than in the CC-AF25 scenario, reflecting the additional battery deployment potential identified in Adeqflex’25. As a result, large-scale batteries account for more than 50% of the average flexibility margins in both upward and downward directions. The additional battery capacities considered in BAT+ significantly increase the overall flexibility margins compared to AF25, where only batteries secured through CRM auctions up to and including CRM 2024 are included. Consistent with the dominant role of batteries described above, the remaining upward flexibility in the CC-BAT+ scenario in both 2030 and 2035 is more evenly distributed, with pumped hydro, industry, and thermal sources each contributing on average between 10% and 15%, while end‑user flexibility remains limited at around 5%. For downward flexibility in 2035, large-scale RES and thermal sources represent the second- and thirdlargest contributors, respectively, whereas industry and end‑user flexibility only provide more limited contributions of up to 7%. Finally, the CT-BAT+ scenario follows similar patterns to the CT‑AF25 scenario, with lower end‑user flexibility and higher thermal downward flexibility due to increased running hours.

Figure 56: Average up and downward fast flexibility margins for CC-BAT+ (top) and CT-BAT+ (bottom) per flexible load and generation technology delivered per hour

FNA report for Belgium 2026, version approved by CREG

Although average fast flexibility is observed to be relatively large in AF25 and BAT+ scenarios, the average contribution only provides a partial view of the importance of the contribution. Available flexibility means are found to vary greatly over time, following the economic dispatch of the units, as well as their technical constraints (e.g. lower levels of flexibility provided by heat pumps during the night). In fact, available flexibility across all hours of the ‘Monte Carlo’ years, as well as the system conditions under which this flexibility is available, are important. As seen in Figure 57 for the CC scenario in 2030, periods of low availability (corresponding to lower percentiles) of fast flexibility resources, such as batteries and pumped hydro storage, can significantly limit the system’s ability to respond to short‑term flexibility needs. As a result, even when average flexibility levels appear sufficient, time-dependent scarcity of these resources may lead to situations where flexibility needs cannot be fully met. This will directly contribute to the uncovered short‑term flexibility discussed in Section 6.3. Nevertheless, and similar to the average margins, Figure 57 shows that the BAT+ scenario provides significantly higher flexibility margins across the full distribution compared to the AF25 scenario. While periods of low availability remain possible, the additional battery capacities considered in BAT+ significantly increase the available flexibility.

Figure 57 Distribution of up and downward hourly fast flexibility margins in 2030 (CC-AF25 & CC-BAT+) per flexible load and generation technology

FNA report for Belgium 2026, version approved by CREG

While the results up till now focused on flexibility provided by domestic resources, cross‑border exchanges have an important role in providing flexibility. Cross‑border flexibility is captured by the available margin between scheduled imports and exports from the economic dispatch simulations and the theoretical maximum import and export capacities. This available margin is subsequently adjusted to account for observed transmission capacity and contractual reserve sharing minima. For fast cross‑border flexibility, observations indicate that under normal price conditions approximately 2500 MW of import capacity and 900 MW of export capacity are available following network constraints, whereas under extreme price conditions the available flexibility is reduced to only 250 MW of import capacity and 350 MW of export capacity, reflecting contractual reserve sharing minima. These results highlight that the contribution of cross‑border flexibility may vary significantly depending on market and system conditions. While cross‑border exchanges can provide substantial flexibility under normal conditions, their availability may be considerably reduced during stressed system situations, when flexibility needs are typically highest.

6.3. Uncovered short-term flexibility needs Uncovered short-term flexibility needs are calculated as the difference between the short-term flexibility needs (Section 6.1) and the residual margin (6.2). It is calculated per hour and per Monte Carlo year for each target year. The results distinguish between up- and downward uncovered short-term flexibility needs and further differentiate these across very fast, fast and total flexibility categories. As explained in the previous sections, the comparison between AF25 and BAT+ should be interpreted in light of the different battery assumptions retained in both scenarios. AF25 only includes battery capacities secured through CRM auctions up to and including CRM 2024, whereas BAT+ reflects the additional battery deployment potential identified in Adeqflex’25. Differences in uncovered short-term flexibility needs between both scenarios are therefore largely driven by the additional battery margins considered in BAT+.

6.3.1. Uncovered short-term upward flexibility needs Figure 58 represents the hours at risk in terms of coverage level (percentage of hours of the year the flexibility needs are covered) for all scenarios and target years studied, with and without accounting cross-border flexibility. During uncovered situations, insufficient upward flexibility is present in the system to cover largest forecast errors and forced outages.

FNA report for Belgium 2026, version approved by CREG

100

Figure 58: Coverage level with and without cross-border (XB) for all scenarios in 2030 and 2035

Firstly, the coverage shows that in both BAT+ scenarios (CC and CT), short-term upward flexibility needs are fully covered for all types of flexibility in 2030 and 2035, with or without considering crossborder contributions. This result reflects the higher domestic flexibility margins available in the BAT+ scenario, mainly driven by additional battery deployment potential identified in Adeqflex’25. Hence, the BAT+ scenario is not analysed further, as no uncovered flexibility needs are observed. For the AF25 scenarios, which only include battery capacities secured through CRM auctions up to and including CRM 2024 similar conclusions can be drawn when considering cross-border contributions. However, when not accounting for cross-border contributions, coverage levels below 100% are observed: • For Current Commitments – AF25: • 96% in 2030 for very fast flexibility, increased to 99% in 2035; • 85% in 2030 for fast flexibility, decreased to 83% in 2035; • 100% in 2030 and 2035 for total flexibility • For Constrained Transition – AF25: • 93% in 2030 for very fast flexibility, increased to 96% in 2035; • 84% in 2030 for fast flexibility, decreased to 78% in 2035; • 100 % in 2030 and 2035 for total flexibility In this AF25 scenario, fast flexibility consequently shows 1300 – 1400 hours at risk in 2030 and even 1500 – 1900 hours at risk in 2035. Figure 59 illustrates the volume at risk of uncovered fast flexibility without cross‑border contributions for 2030 and 2035 under the CC scenario. Note that the uncovered fast-flexibility volume shown is calculated after already accounting for the portion of the FRR requirements linked to the dimensioning FNA report for Belgium 2026, version approved by CREG

101

incident (1030 MW). Therefore, the volume highlighted in the figure represents the additional need for fast flexibility that remains after this component has been considered. According to the reserve capacity projections in Section 6.4.1 an increase in dynamic reserve capacity is expected to provide additional volume, as shown in Figure 59. At least when delivered with contracted reserve capacity or another balancing means assumed to be available for largest part of the time. This contribution will address part of the uncovered fast flexibility needs. In addition, cross-border flexibility through contracted reserve sharing (currently determined at 250 MW for upward) can further reduce the uncovered fast upward flexibility. As the availability of this contracted sharing volume is not guaranteed, the effective contribution may, however, be lower than the contracted sharing volume under certain system conditions, for example due to simultaneous reserve needs in participating countries or limitations in available cross-border transmission capacity. After accounting for dynamic reserve capacity and reserve sharing, remaining uncovered fast flexibility is 150 MW (2030) or 200 MW (2035) in the AF25 scenario. These residual volumes should be interpreted in light of the underlying AF25 assumptions, which only include battery capacities secured through CRM auctions up to and including CRM 2024.

Figure 59 Volume at risk of uncovered fast flexibility without cross-border contributions for 2030 and 2035 under the CCAF25 scenario

Figure 60 presents the fast upward flexibility margins of different technologies available in 2030 under the CC-AF25 scenario. The available flexibility margin is shown both as the average over all moments of the year and as the average during periods when uncovered fast flexibility occurs. The most significant difference between these two situations lies in the contribution from storage technologies. While such technologies can make a substantial contribution under normal operating conditions, they may already be committed in other markets during specific system conditions. As a result, their availability to deliver fast upward flexibility cannot always be guaranteed without reservations of capacity. These observations relate to the AF25 scenario, which only includes battery capacities secured through CRM auctions up to and including CRM 2024.

FNA report for Belgium 2026, version approved by CREG

102

Figure 60: Fast upward available flexibility margins in 2030 under the CC-AF25 scenario

6.3.2. Uncovered short-term downward flexibility needs Similar to the previous section, uncovered short-term downward flexibility is expressed in terms of coverage level. The coverage level in Figure 61 indicates which percentage of the time the flexibility margins are equal to or exceed the required short-term needs for very fast, fast and total flexibility. The results show that the additional battery deployment potential considered in BAT+, compared to AF25 which only includes battery capacities secured through CRM auctions up to and including CRM 2024, significantly improves coverage levels for downward flexibility needs. With contributions from cross-border flexibility the coverage levels for: • Very fast flexibility activated within 5 minutes before real-time reaches coverage levels of 100% in 2035 across both scenarios showing sufficient margins are available to cover largest forecast errors and forced outages. • Fast flexibility activated 15 minutes before real-time managing last forecast errors, reaches: • 94% (CC-AF25) and 91 % (CT-AF25) coverage in 2035 • 99% (CC-BAT+) and 98% (CT-BAT+) coverage in 2035 • Total flexibility capturing the overall requirement of short-term flexibility from day-ahead to realtime where fast margins (up to 15 minutes) and slow margins (up to 5 hours) can react in 2035 shows: • 91% (CC-AF25) and 88 % (CT-AF25) coverage in 2035 • 98% (CC-BAT+) and 96% (CT-BAT+) coverage in 2035 Cross-border flexibility reduces periods with uncovered needs through remaining export capacity, limited by the expected availability of liquidity and transmission capacity during normal conditions (up to 900 MW) and reserve sharing agreements (up to 350 MW) during low price conditions. Without the participation of cross-border flexibility, coverage levels for total flexibility are reduced to 70% and 62% CC-AF25 and CT-AF25 respectively. These coverage levels increase to 83% (CC-BAT+) and 78% (CTBAT+) in the high battery potential scenario, illustrating the contribution of the additional battery deployment potential identified in Adeqflex’25 beyond the battery capacities already secured through CRM auctions up to and including CRM 2024.

FNA report for Belgium 2026, version approved by CREG

103

Figure 61: Coverage level for the downward very fast, fast and total flexibility for 2030 and 2035 with and without crossborder (XB) in the CC/CT-AF25 scenario (top) and CC/CT-BAT+ scenario (bottom)

➢

Hours and volumes at risk of the uncovered short-term needs

When translating the coverage levels, including cross-border flexibility, into hours the system would be at risk, a clear distinction can be observed between the AF25 and BAT+ scenarios. The additional battery potential considered in BAT+, compared to AF25 which only includes battery capacities secured through CRM auctions up to and including CRM 2024, substantially reduces the frequency of uncovered downward flexibility needs. In Figure 62 (top) for the AF25 scenarios: • Very fast flexibility shows no risk hours in accordance with coverage levels of 100% • Fast flexibility shows risk hours that reach up to 550 hours (CC) and 790 hours (CT) in 2035 • Total flexibility shows risk hours that reach up to 800 hours (CC) and 1040 hours (CT) in 2035 In Figure 62 (bottom) for the BAT+ scenarios : • Very fast flexibility shows no risk hours in accordance with coverage levels of 100% • Fast flexibility shows risk hours that reach up to 100 hours (CC) and 200 hours (CT) in 2035 • Total flexibility shows risk hours that reach up to 210 hours (CC) and 350 hours (CT) in 2035 This reduction is particularly visible for fast and total flexibility, where the additional battery margins in BAT+ significantly decrease the number of hours during which available flexibility is insufficient to cover the largest forecast errors and forced outages.

FNA report for Belgium 2026, version approved by CREG

104

Figure 62: Total and fast flexibility hours at risk in 2030 and 2035 in the CC/CT-AF25 scenario (top) and CC/CT-BAT+ scenario (bottom)

When considering total volumes at risk occurring 1% of the time (approximately 90 hours per year) as observed in Figure 62 (top), the total uncovered volume driven by both fast and slow shortages reaches 1.8 GW in 2030 and 2.3 GW in 2035 under the CC-AF25 scenario. In the CT-AF25 scenario, the total uncovered volumes at risk increase to 2.1 GW in 2030 and up to 2. GW in 2035 at the 1% level, reflecting higher total uncovered flexibility needs. At even more extreme levels (0.1% percentile), the uncovered volume at risk rises to values comparable to the maximum flexibility needs identified in Section 6.1. A similar assessment is presented in Figure 63 for the BAT+ scenario. Volumes at risk amount to around 1.0 GW in the CC‑BAT+ scenario for both 2030 and 2035, while they increase further to approximately 1.2 GW and 1.4 GW in the CT‑BAT+ scenario. Compared to AF25, the BAT+ scenarios show a marked reduction in the volumes at risk across the probability distribution. This reflects the contribution of the additional battery potential identified in Adeqflex’25, although extreme events remain possible under specific system conditions. Extreme levels (0.1% percentile) follow a similar pattern to the AF25 scenario, although they remain below the maximum flexibility needs identified in Section 6.1. The reliability threshold of 1% (around 90 hours) is not defined in the European FNA methodology nor in Belgian legislation and should therefore be interpreted as a reference point rather than a target. However, it does align with reserve dimensioning criteria. Furthermore, note that the uncovered hours at risk represent situations where downward flexibility margins are insufficient to cover the largest forecast errors and forced outages. As these results are based on extreme forecast-error levels and represent situations in which available flexibility margins are insufficient, the number of hours at risk should not be interpreted as the expected duration of actual system issues. The effective occurrence

FNA report for Belgium 2026, version approved by CREG

105

of such events depends on the simultaneous occurrence of large forecast errors and the corresponding system conditions.

Figure 63: Probability distribution of the uncovered total and fast downward flexibility needs for the CC/CT-AF25 & BAT+ scenarios

➢

Daily and seasonal patterns of the uncovered short-term needs

Figure 64, shows the daily uncovered total and fast flexibility computed as the hourly average across all days for 2030 and 2035. Across all scenarios, the results show a peak around midday. These patterns mainly occur during hours characterised by high solar generation and limited downward flexibility availability. Due to the increase in solar generation between 2030 and 2035, the average uncovered volume at risk increases towards 2035 both for total and fast flexibility in the CC-AF25 and CT-AF25 scenarios. In the CC-BAT+ and CT-BAT+ scenarios, which consider the additional battery potential identified in Adeqflex’25, similar daily patterns are observed, although with lower volumes.

FNA report for Belgium 2026, version approved by CREG

106

Figure 64: Hourly average downward total and fast uncovered short-term flexibility volume at risk distribution per target year for the CC/CT-AF25 (top) & BAT+ (bottom) scenarios

In Figure 65, similar results can be seen for the hourly average total volumes at risk across summer, spring, autumn and winter months. Across CC-AF25 and CT-AF25 scenarios, it is observed that the hourly average uncovered volumes at risk are larger during summer and spring 2035, while winter and autumn show the lowest peaks. For the CC-BAT+ and CT-BAT+ scenarios in Figure 66, similar patterns are observed with lower magnitudes due to the additional battery potential identified in Adeqflex’25 considered in these scenarios.

FNA report for Belgium 2026, version approved by CREG

107

Figure 65: Hourly average of downward total uncovered short-term flexibility volume at risk distribution per target year per season for CC-AF25 and CT-AF25.

Figure 66 Hourly average downward total uncovered short-term flexibility volume at risk distribution per target year per season for CC-BAT+ and CT-BAT+

FNA report for Belgium 2026, version approved by CREG

108

➢

Correlation of the uncovered short-term needs with system conditions

Assessing hourly total uncovered volumes at risk on a day-by-day basis, uncovered volumes, similarly, mainly arise during spring and summer periods. More specifically, they coincide with conditions characterised by low system load, limited wind generation, and high solar infeed. The limited impact of wind generation should be interpreted in light of the modelling assumptions, including the fact that the flexibility of large wind parks, in particular offshore wind, is already taken into account. Uncovered volumes at risk most frequently occur during spring and summer weekends, when reduced demand can amplify situations with high renewable generation and limited downward flexibility availability. This temporal clustering is clearly visible in the heatmap shown in Figure 67 for the CC-AF25 scenario in 2030, which highlights recurring patterns of elevated uncovered downward flexibility needs during spring and summer. Similar patterns are observed in the CT and BAT+ scenarios, although the magnitude of uncovered volumes is lower in BAT+ due to the additional battery potential identified in Adeqflex’25 considered in that scenario.

Figure 67: Representation of the average hourly short-term total uncovered volumes at risk per day and per target year for the CC-AF25 scenario, average over all Monte Carlo years

Building on the preceding analysis of hourly uncovered flexibility volumes, as illustrated by the heatmaps, the correlation analysis in Table 9 provides further support for these observed patterns. In particular, total uncovered short‑term flexibility shows a positive correlation with solar generation across all scenarios and time horizons. The correlation is, however, noticeably lower in the BAT+ scenarios (0.39–0.50) than in the AF25 scenarios (0.63–0.72), indicating that the additional battery potential identified in Adeqflex’25 considered in BAT+ reduces the relationship between periods of high solar generation and uncovered downward flexibility needs. Nevertheless, high solar infeed remains an important driver of uncovered downward flexibility needs across all scenarios. By contrast, offshore wind exhibits a small but systematic negative correlation (around −0.09 to −0.14). This reflects the fact that large offshore wind parks can provide downward flexibility through curtailment when system pressure arises. Under such conditions, offshore wind curtailment can effectively mitigate system strain and thereby reduce the occurrence and magnitude of uncovered flexibility needs. FNA report for Belgium 2026, version approved by CREG

109

Correlations with onshore wind and load remain close to zero, indicating a limited direct relationship with uncovered short‑term flexibility in the analysed scenarios. Table 9: Correlation with the total uncovered short-term flexibility needs and solar generation, onshore wind, offshore wind and the load for CC/CT-AF25 and CC/CT-BAT+ scenarios.

➢

Durations of the uncovered short-term needs

In addition to the magnitude of uncovered short-term flexibility needs, the duration of these events provides further insight into their underlying drivers. As shown by the P50 and P99 statistics in Table 10, half of the uncovered flexibility events for the AF25 scenarios have a duration shorter than 4 to 5 hours, while most events (P99) typically remain below 9 to 10 hours. In the BAT+ scenarios, which include the additional battery potential identified in Adeqflex’25, durations are further reduced, with P50 values of 2 to 3 hours and P99 values limited to around 5 hours. This reduction suggests that the additional battery potential not only decreases the magnitude of uncovered flexibility needs, but also shortens their duration. This limited duration is consistent with the strong relationship observed between uncovered flexibility needs and periods of high solar generation. The shorter durations observed in BAT+ further indicate that the additional battery potential identified in Adeqflex’25 reduces the persistence of such events. As a result, uncovered events tend to follow a pronounced daily pattern, in line with the daily structures observed in the heatmaps discussed in the previous paragraphs. Importantly note that these durations only translate into system issues when adverse forecast errors, assessed at the P99.9 level for shortterm flexibility needs, persist simultaneously over extended periods. However, as demonstrated in Section 6.1, such long-duration events combined with large forecast errors are unlikely, as their probability of occurrence remains low. As such, the duration profile of the uncovered short‑term flexibility needs suggests that short‑duration flexibility resources, such as battery storage and demand‑side response, may be well suited to address these events. The shorter event durations observed in BAT+ further support the contribution that additional battery flexibility could make in reducing uncovered needs.

FNA report for Belgium 2026, version approved by CREG

110

Table 10: Duration of total uncovered short-term flexibility events 50th and 99th percentile for CC/CT-AF25 and CC/CT-BAT+ scenarios

6.3.3. Energy from uncovered downward short-term flexibility events Figure 68 illustrates the effective activation frequency associated with the uncovered downward flexibility volumes identified in the previous sections. This is calculated by means of combining the hours at risk with the effective probability of facing a forecast error of a magnitude exceeding the available flexibility margins. Note that the activation frequency in Figure 69, expressed in hours per year, is an average over all Monte Carlo years. The results show a clear difference between the AF25 and BAT+ scenarios. While significant uncovered volumes remain possible in AF25, these volumes are associated with increasingly infrequent activations. Compared to the AF25 scenarios, which only include battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios, which incorporate the additional battery potential identified in Adeqflex’25, show a significant reduction in both the magnitude and the frequency of uncovered flexibility events. As a result, higher uncovered volumes are associated with substantially lower activation frequencies than in AF25.

FNA report for Belgium 2026, version approved by CREG

111

Figure 68: Average frequency of activation over all Monte Carlo years from short-term flexibility uncovered volumes at risk in the CC-AF25 (top) and CC-BAT+ (bottom) scenarios

Building on these activation frequencies, the total energy linked to uncovered moments (energy not managed by available flexibility) and covered moments (periods where large-scale renewable assets are curtailed to provide the required flexibility) in Figure 69 is summarized below. Note that the total of the covered and uncovered needs are used in the finetuning RES integration needs in accordance with Article 10(5). During uncovered moments: • CC/CT-AF25 scenario: 20-57 GWh/year • CC/CT-BAT+ scenario: 2-5 GWh/year During covered moments: • CC/CT-AF25 scenario: 21-45 GWh/year • CC/CT-BAT+ scenario: 8-17 GWh/year Compared to the AF25 scenarios, which only include battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios lead to a substantial reduction in both the frequency of activation and the associated uncovered energy volumes. This indicates that the additional battery potential identified in Adeqflex’25 not only reduces the magnitude of uncovered downward flexibility needs, but also significantly lowers the probability that such events occur. The apparent contrast between high instantaneous power needs from the short-term volumes at risk and modest energy volumes linked to the activation frequencies, is explained by the probabilistic nature of forecast errors and forced outages. Short‑term flexibility needs are evaluated at the P99.9 forecast‑error level, corresponding to extreme but infrequent events. As a result, although large downward flexibility capacities may be required to cope with the uncovered volumes at risk, the associated energy volumes remain limited because these situations occur only during a small number of hours.

FNA report for Belgium 2026, version approved by CREG

112

Figure 69: Curtailed energy in GWh per target year during covered and uncovered short-term flexibility moments for CC/CTAF25 and CC/CT-BAT+ scenarios.

This becomes even clearer when considering the average uncovered energy per hour, illustrated in Figure 70 for CC/CT-AF25 and CC/T_BAT+ scenarios. In AF25 scenarios, average values remain negligible outside daytime hours and peak around midday, with maximum hourly averages of 15– 23 MWh/h in 2030 and 23–39 MWh/h in 2035. These patterns are consistent with the relationship previously observed between uncovered flexibility needs and periods of high solar generation. In the BAT+ scenarios (Figure 71), significantly lower uncovered energy volumes are observed compared to the AF25 scenarios, which only include battery capacities secured through CRM auctions up to and including CRM 2024. This suggests that the additional battery potential identified in Adeqflex’25 substantially reduces the occurrence and magnitude of residual uncovered downward flexibility needs. Despite the presence of high uncovered power requirements during these periods, the corresponding energy per hour remains modest. This confirms that uncovered short‑term flexibility needs are primarily a power‑driven challenge rather than an energy‑driven one. The relatively limited energy volumes involved, combined with the shorter event durations observed in the BAT+ scenarios, suggest that short-duration flexibility resources such as battery storage and demand response may be well suited to address these events, provided sufficient flexibility is available at the relevant moments.

FNA report for Belgium 2026, version approved by CREG

113

Figure 70: Hourly average curtailed energy from uncovered short-term flexibility moments per target year for CC/CT-AF25 scenarios

Figure 71: Hourly average curtailed energy from uncovered short-term flexibility moments per target year for CC/CT-BAT+ scenarios

6.3.4. Sensitivities on uncovered downward short-term flexibility This section and Figure 72 present the results of a sensitivity analysis for the CC and CT scenarios in 2030 and 2035, illustrating how variations in system drivers influence the level of uncovered total short‑term flexibility. The sensitivities assess the impact of changes in end‑user participation, cross‑border reserve sharing, and renewable energy source flexibility on the uncovered flexibility (previously represented by the P1 percentile level in Figure 63). Overall, compared to the AF25 scenarios, which only include battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios not only lead to lower uncovered shortterm flexibility needs, but also reduce the sensitivity of the results to several uncertainty drivers. This indicates that the additional battery potential identified in Adeqflex’25 improves the robustness of the system against variations in renewable flexibility reliability, cross-border contributions and end-user participation. The first sensitivity evaluates the impact of different levels of end‑user participation in short‑term flexibility markets. The low variant assumes no participation of end‑users in short‑term markets, while the high variant assumes 100% participation of end‑user assets assumed in the scenarios to react to day-ahead market prices. The impact of this sensitivity is relatively limited in 2030. In both the CC and CT scenarios, the difference between low and high end‑user participation remains within a narrow range (e.g. 1.7–1.9 GW for 2030 CC-AF25 and 0.8–1.0 GW for 2030 CC-BAT+). By contrast, by 2035 the spread widens, with results ranging from 2.1 – 2.4 GW for CC-AF25 and 0.5 – 1.2 GW for CC-BAT+. This indicates that the availability and activation of flexible end‑users become increasingly relevant in later years. High participation of end‑users in short‑term markets alleviates system needs, while the absence of such participation leads to an increase of uncovered short-term flexibility needs. The results also indicate that, even when additional battery potential is considered in the BAT+ scenarios, end-user participation remains an important contributor to reducing residual uncovered flexibility needs.

FNA report for Belgium 2026, version approved by CREG

114

The second sensitivity represents the absence of 350 MW export capacity through cross‑border reserve sharing assumed to be not available during extreme prices. This sensitivity translates to an increase of roughly 200–400 MW in uncovered short‑term flexibility depending on the scenario. Consequently, these observations underline the contribution of cross-border flexibility in reducing uncovered short-term flexibility needs. However, compared to the AF25 scenarios, the BAT+ scenarios show that part of this sensitivity is mitigated by the additional battery potential identified in Adeqflex’25, which reduces the reliance on cross-border flexibility to cover residual downward flexibility needs. Subsequently, the third sensitivity captures uncertainty in renewable energy source reliability to provide downward flexibility by applying a ±10 percent-point variation to the reliability of RES to provide flexibility, reflecting uncertainties related to forecasting errors. The result shows that when RES generation is 10% less reliable to provide downward flexibility (due to forecast errors), the uncovered flexibility needs increase and vice versa.

Figure 72: End-user, cross-border and RES availability sensitivities on the 1-percentile total uncovered short-term needs in CC/CT-AF25 (top) and CC/CT-BAT+ (bottom) scenarios

Finally, a fourth sensitivity is analysed, representing a ± 20 percent change in the availability of existing downward market response. No observable impact is found, mainly due to the limited volume of existing downward response (less than 300 MW) and because this flexibility is already fully activated during low-price periods. Consequently, variations in the availability of existing downward market response have only a negligible impact on the quantified uncovered short-term flexibility needs. As this sensitivity does not materially affect the results, it is not shown in the figure.

6.4. Flexible resources needed to meet the uncovered needs Article 16 of the FNA methodology requires TSOs to further assess the capability of different flexibility resources to address uncovered flexibility needs, including short-term flexibility needs. Two pathways can be identified to address the remaining needs. 1. A first pathway consists of deploying additional non-fossil flexibility resources to achieve full RES integration. In the AF25 scenarios, additional non fossil flexibility, in particular battery deployment beyond the capacities existing or contracted through CRM auctions up to 2024, FNA report for Belgium 2026, version approved by CREG

115

remains a potential option to reduce the remaining uncovered needs. The impact of such additional battery deployment is illustrated through the BAT+ scenarios, which reflect the battery potential identified in Adeqflex’25. The comparison between the AF25 and BAT+ scenarios therefore already provides an indication of the contribution that additional non fossil flexibility may make in reducing the identified needs and is not further explored in this section. 2. A second pathway consists of accepting RES curtailment and relying on renewable flexibility to manage the resulting periods of excess generation. Renewable flexibility can contribute to covering downward flexibility needs but is not expected to play a significant role in upward flexibility. The remainder of this section therefore focuses on assessing the capability of renewable assets to contribute to the remaining uncovered downward short-term needs. Without additional non-fossil fuel flexibility resources, these uncovered flexibility needs can be met with (1) reserve capacity needs or (2) unlocking additional renewable flexibility. The first is subject to an existing framework with a dimensioning following European legislation and national legislation and is discussed in Section 6.4.1. The need for the second is discussed in Section 6.4.2. Note that the scenarios already account for all other flexibility resources participating in short-term flexibility through the scenarios: • Conventional flexibility, including scenarios with additional batteries up to 4.5 GW in 2030 and 5.7 GW and 2035. • End-user flexibility, up to already ambitious levels of electric vehicle, heat pump and home batteries participation as explained in Section 3. • Cross-border flexibility, through intra-day and balancing markets, only constrained in practice during extreme price periods, while still maintaining minimum margins through reserve sharing agreements. • RES flexibility, through participation of all assets larger than 25 MW in the balancing market through regulatory obligations.

6.4.1. Reserve capacity needs This report contains Elia’s latest update for Belgium’s FCR and FRR reserve capacity needs projections up to 2035. These represent the flexibility needs covered by Elia through reserve capacity, in line with legislation and are determined to cover at least the dimensioning incident, referred to as the largest relevant power or network outage impacting system imbalances, and to cover at least 99% of up- and downward system imbalances. It is therefore considered as a subset of the fast flexibility needs presented in the previous sections (covering up to 99.9% of the forecast errors and forced outages). The projections are an update of the reserve capacity projections published in Adeqflex’25. The update concerns, on the one hand, an alignment with the scenarios used in the FNA, namely the CC and CT scenarios. Reserve projections are assessed only for the AF25 scenarios. Compared to AF25, which only includes battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios incorporate the additional battery potential identified in Adeqflex’25. The results would show that this additional battery potential contributes to reducing future reserve capacity needs by increasing the availability of flexibility in balancing timeframes and thereby lowering residual system imbalances. Due to the large amounts of batteries towards 2030 and 2035, and the large impact expected on the intra-day and balancing markets as demonstrated in Section 6.2, it is therefore expected that reserve needs would remain around the dimensioning incident for upward and remain around actual levels for downward. On the other hand, the update concerns a few methodological improvements compared to Adeqflex’25 as well as alignments with the FNA framework: • The use of up to five years of observed historical data instead of two; • The use of last forecast data instead of day-ahead forecast data; • A review of market performance indicators in line with latest observations. FNA report for Belgium 2026, version approved by CREG

116

The projections are based on the formal reserve dimensioning method which is based on EU legislation and NRA approved methodologies. The methodology to conduct the projections is discussed in Annex B.1.6. The projections are developed by N-SIDE, a consultant specialised in advanced analytics, using Elia’s dynamic reserve dimensioning tool which dimensions the needs from day to day. The projected increases in upward reserve capacity needs are addressed through established balancing market and procurement frameworks. As reserve requirements are periodically dimensioned and procured by the TSO, increasing reserve needs are expected to create corresponding market signals and remuneration opportunities for flexibility providers. Consequently, the reserve capacity volumes identified in this section are expected to be addressed through the existing balancing framework and do not, in themselves, constitute additional flexibility resource needs requiring specific measures to unlock flexibility. The projected increases in downward reserve capacity needs, in absence of procurement, do require additional flexibility resource needs in view of reserve capacity procurements or alternative measures such as unlocking additional renewable flexibility.

6.4.1.1. FCR needs For the fastest product FCR, used to stabilise the European grid, as explained in Annex B, projections of needs are based on a projection of the weight of Belgium’s generation and demand in the synchronous area. These are based on the latest available data, in this case in TYNDP 2026 covering 2030 and 2035. Note that in 2026, FCR needs increased following introduction of probabilistic dimensioning, increasing the total FCR needs in the Synchronous Area from 3000 MW to 3450 MW, impacting the national FCR share in a respective way. Figure 73 shows it is expected that the Belgian FCR share reduces to 91 MW in 2027 due to the impact of the nuclear phase out in 2025 on Belgian generation, and the total demand in the Synchronous Area following the integration of Ukraine and Moldavia that same year. These will affect the calculation of the FCR needs from 2027 onwards. The impact of the nuclear maintenance during summers of 2026 - 2028 is found to have a very small effect (around 1 MW) for FCR needs in 2028 to 2030.

Figure 73: Expected FCR needs up to 2035 in MW

6.4.1.2. FRR needs - methodology and assumptions Estimations regarding Elia’s future reserve capacity needs are based on the methodology presented in in Annex B.1.6 and summarized in Figure 74. FNA report for Belgium 2026, version approved by CREG

117

Figure 74: summary of the methodology for the projections on reserve needs

• In the first step, scenarios are constructed concerning market performance or the future ability of market players to balance new renewable generation assets in their portfolio. This covers (1) the ability to balance forecasting errors related to renewable generation and forced outages based on intraday re-scheduling and reactive balancing; (2) changes in system imbalances related to the existing generation mix; (3) assumptions regarding forecast improvements. • In a second step, system imbalance projections are calculated up to 2035 by upscaling historic system imbalances in light of expected forecasting errors related to the future generation mix, and in particular the increase in installed capacity of variable renewable generation. This is based on projections related to the installed wind and solar power presented in the AF25 CC and AF25 CT scenario of this study and time series of historic forecast errors. • In a third step, estimations about future FRR / aFRR / mFRR needs are made based on Elia’s current dynamic dimensioning methodologies, in line with the existing legal and regulatory framework The CC scenario is aligned with current renewable energy policies while the CT scenario assumes less onshore wind power, less electrification but also less end-user flexibility. The latter will result in lower market performance indicators. In terms of relevant assumptions on the generation fleet, it is reminded that offshore wind power capacity up to 2035 is limited to 4.4 GW (3.0 GW as from 2031 and 4,4 GW in 2032), while not assuming the commissioning of a second interconnector (Nautilus) to UK. Two nuclear units (2 GW) remain into operation up to at least 2035 while accounting maintenance periods during the summer periods before 2029. Nemo Link flows, relevant in the context of the dimensioning incident, are based on Adeqflex’25 CC & CT simulations used as reference for this study. The market performance indicators represent the ability of the market parties to balance the future capacity of renewable generation installed; (1) the forecasting accuracy improvement itself, (2) the system imbalance improvement, and (3) the market coverage. Annual improvements in forecasting accuracy are fixed at 1% in both scenarios. The system imbalance improvement represents the reduction in system imbalances in percentage terms compared with the previous year (iteratively taking into account new renewable capacity added to the system in the previous years). This represents changes in the assumed ability of market players to balance their existing portfolio. The market coverage represents the ability of market players to cover forecasting errors related to newly installed renewable capacity: it represents the percentage of forecasting error, related to this new capacity, covered by the market parties while the remaining share contributes to the increase of the system imbalance. In view of the relevance of reserve dimensioning, the methodology is based on large forecast errors, aggregated over the different technologies. The projections are based on an analysis of the historic system imbalance and market performance as starting point for the projections, extrapolated based on Elia’s expectations in a certain scenario as shown in Figure 75. FNA report for Belgium 2026, version approved by CREG

118

Figure 75: market performance indicators

All indicators are improved compared to Adeqflex’25 at least until 2030 following observations of everimproving imbalances (except during extreme conditions). This trend is not expected to continue with increasing RES at least without additional actions and indicators are assumed to fall back to similar levels as in the Adeqflex’25 report towards 2035. The CT scenario presents lower market performance following lower flexibility. • The SI improvement follows an observation of average deterioration of 0.55% over the last five year (but observed 6% over the last three years only due to a large improvement in 2025). The SI improvement increases in line with recent observations (but stabilisation after 2030) with increasing renewables; • The market coverage follows an observation of 40% (upward) of the last three years, and 45% (downward). The market coverage improves in line with recent observations (but stabilisation after 2030) with increasing renewables. Note that lower market performance following lower flexibility impacts both the ability of market players to balance existing (represented by SI improvement) and new capacity (represented trough market coverage) as less flexibility will be available in intra-day and balancing markets. The projections follow a positive evolution up to 2032, after which stabilisation or reduction is foreseen to the values presented in the Adequacy and Flexibility study. Finally, no potential impact of congestion management measures or possible effect of decreasing imbalance prices on system imbalances is explicitly accounted. 6.4.1.3. FRR needs - results In Figure 76, the projections for FRR needs show that average reserve needs are expected to: • increase in the upward direction from around 1030 MW observed in 2025 to 1190 MW in 2035 in the AF25 CC; and increase up to around 1470 MW in 2035 in the Constrained transition scenario. Until 2030, the upward FRR needs are mainly driven by the dimensioning incident. After 2030, additional RES including offshore and the assumed stabilisation of market performance drives the FRR needs up. The impact of the nuclear maintenance during summers of 2026 until 2028 is found to have limited effect due to larger Nemo Link imports (and large-scale CCGTs) which create a comparable dimensioning incident. FNA report for Belgium 2026, version approved by CREG

119

• increase in downward direction from around 940 MW observed in 2025 to around 1020 MW in the AF CC scenario; and may increase up to around 1340 MW in the Constrained transition scenario. • Until 2030, the downward FRR needs remain stable. After 2030, additional RES and market performance stabilisation drives FRR needs up in line with the upward needs. The reduction until 2030 is mainly driven by the decrease of Nemo Link exports in line with the simulations for both AF25 CC and CT scenarios. FRR needs in the CT scenario are higher following lower market performance indicators (less electrification end-user flexibility, even if this is partially compensated by less onshore wind power to balance). Due to the large amounts of batteries towards 2030 and 2035 in the BAT+ scenarios, and the large expected impact on the intra-day and balancing markets as demonstrated in Section 6.2, it is therefore expected that reserve needs in 2030 and 2035 remain at lower levels compared to the AF25 scenarios, i.e. around the dimensioning incident for upward and remain around actual levels for downward.

Figure 76: Expected FRR needs up to 2035

6.4.1.4. aFRR needs In Figure 77, the projections for aFRR needs show that average reserves needs are expected to: Increase from around 103 -107 MW observed in 2025 for upward and downward aFRR respectively, to 110 MW in 2035 in the AF25 CC scenario; and 140 – 150 MW in 2035 in the CT scenario, • Develop symmetrically for upward and downward, • Initially reduce relative to 2026 aFRR needs, in line with observed trends • Increase after 2030, driven by additional RES including offshore wind, together with assumed stabilisation of the market performance. Results are shown after accounting for the FRCE quality correction factor currently fixed at 64% of the probabilistic result. Results account stable imbalance netting volumes as currently observed in IGCC and PICASSO.

FNA report for Belgium 2026, version approved by CREG

120

Similar to FRR, aFRR needs are expected to remain stable in the BAT+ scenarios, and remain below the AF25 levels. The large-scale batteries are expected to mitigate the system imbalances, and at least when assumed their ramps do not deteriorate the system imbalance within the 15-minute periods.

Figure 77: Expected aFRR needs up to 2035

These volumes will translate into balancing capacity procurements. • aFRR up- and downward needs are expected to translate one-to-one to aFRR balancing capacity procurements. • mFRR upward needs are expected to translate into balancing capacity procurements after accounting for reserve sharing and potentially accounting non-contracted balancing energy means. No new assessments are included on reserve sharing (static values of 350 MW FRR down and 250 MW FRR up) or dynamic procurement considering the availability of noncontracted bids. This will however be further analysed in the course of 2026 and 2027 with (earliest) possible implementation towards 2028. • mFRR downward needs are expected to remain covered through a combination of reserve sharing and non-contracted balancing energy means. Elia is currently investigating whether there is a need to contract downward mFRR balancing capacity.

6.4.1.5. Relation to flexibility needs and means analysis This section analyses the capability of additional flexibility resources to reduce the remaining uncovered short-term flexibility needs. • The projected increase in upward reserve capacity needs are assumed to be managed through existing reserve procurement mechanisms and therefore do not give rise to additional uncovered flexibility needs in the present assessment. • The projected increase in downward reserve capacity needs may require additional flexibility resources. In the AF25 scenarios, additional non-fossil flexibility, in particular battery deployment beyond the capacities existing or contracted through CRM auctions up to 2024, remains a potential option to reduce these needs. The impact of such additional battery deployment is

FNA report for Belgium 2026, version approved by CREG

121

illustrated through the BAT+ scenarios, which reflect the battery potential identified in Adeqflex’25. • The comparison between the AF25 and BAT+ scenarios therefore already provides an indication of the contribution that additional non-fossil flexibility may make in reducing residual downward flexibility needs. For this reason, the remainder of this section focuses on the contribution that renewable flexibility could provide to the remaining uncovered downward short-term needs. Figure 78 further clarifies the relation between fast flexibility needs and reserve capacity (FRR needs). The fast flexibility needs represent the capacity that needs to be activated within 15 minutes before realtime and relates closest to FRR needs. The example based on the results in the FNA is for upward but the same principles apply for downward. Flexibility needs cover 99.9% of forecast errors and forced outages. The flexibility needs thus account for very extreme events (99.9% percentiles) and incorporate the reaction of the market. In contrast, reserve capacity (FRR needs) is by legislation only dimensioned to cover the dimensioning incident or 99% of system imbalances. Reactions from the market result in lower system imbalances, hence lead to a reduction of the FRR needs, and a lower percentile of the possible system imbalances need to be covered. The difference between the fast flexibility needs and FRR needs can be explained by comparing forecast errors and system imbalances (accounting for market reactions) and the chosen reliability level comparing 99.9% with 99.0% (accounting for extreme events balanced by exceptional balancing measures). Uncovered fast flexibility needs by the market or reserve capacity will result in activation of available additional capacity in other countries via the aFRR- and/or mFRR-Platforms (MARI/PICASSO) where possible but could also lead to uncovered imbalances in the grid (frequency restoration control errors). Same for aFRR, the difference between very fast flexibility and aFRR is translated into activation of additional available reserve capacity in the aFRR-Platform (PICASSO) where possible but could also lead to uncovered imbalances in the grid (frequency restoration control errors), which could lead to additional FCR activations.

Figure 78: Average upward flexibility needs and reserve needs in 2035 (Current Commitment scenario)

While flexibility needs are found to increase over time, due to forecast errors associated with increasing renewable capacity, reserve capacity needs are found to increase to a lower extent. This is on the one hand explained by the 99th percentile level dampening the effect of extreme events, as well as the FNA report for Belgium 2026, version approved by CREG

122

market coverage which is expected to still cover larger part of the increasing forecast errors (as observed during the last years).

6.4.2. Additional flexibility resource needs Without the addition of non-fossil fuel flexibility resources, the uncovered flexibility needs can be met with (1) reserve capacity needs or (2) unlocking additional renewable flexibility. While the largest part of uncovered upward short-term needs can be covered with reserve capacity, uncovered short-term downward needs would require, at least in the absence of downward procurement, measures to unlock additional flexibility. This would be in particular renewable flexibility when assuming no other flexibility would be developed on top of the flexibility accounted in the scenario. Note that the scenarios already account for all other flexibility resources participating in short-term flexibility: •

Conventional flexibility, including the BAT+ scenarios with a potential battery capacity up to 4.5 GW in 2030 and 5.7 GW in 2035.

•

End-user flexibility, up to already ambitious levels of electric vehicle, heat pump and home batteries participation as explained in Section 3.

•

Cross-border flexibility, through intra-day and balancing markets, only constrained in practice during extreme price periods, while still maintaining minimum margins through reserve sharing agreements.

•

RES flexibility, through participation of all assets larger than 25 MW in the balancing market through regulatory obligations.

As the AF25 scenarios only account for battery capacities existing or contracted through CRM auctions up to 2024, additional non-fossil flexibility, in particular battery deployment, remains a potential option to reduce the remaining uncovered downward short-term needs. By contrast, the BAT+ scenarios can be interpreted as reflecting a situation where the battery potential identified in Adeqflex’25 has already largely been unlocked. The impact of such additional non-fossil flexibility is therefore already illustrated through the comparison between the AF25 and BAT+ scenarios. For this reason, the remainder of this section focuses on the ability of renewable assets to contribute to covering the remaining uncovered downward short-term needs. It is already concluded in Section 6.3.1 that, once additional reserve capacity and reserve sharing contributions are taken into account, no additional upward flexibility resource needs are identified, particularly in the BAT+ scenarios. By contrast, uncovered downward flexibility needs were already identified in Section 6.3.2 persisting in specific situations characterised by low system load, limited wind generation, and high solar infeed. As explained in the previous section, apart from aFRR, no additional reserve capacity is foreseen at this stage to cover these flexibility needs. Figure 79 (left), recapitulates the uncovered volumes of total short-term flexibility identified in Figure 63. These represent the additional downward flexibility resources that would need to be available close to real-time under the 1% reference reliability level. Compared to AF25, which only includes battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios, which include the additional battery potential identified in Adeqflex’25, reduce the residual flexibility needs by approximately 0.8–0.9 GW in 2030 and by 1.3–1.5 GW in 2035, corresponding to a reduction of around 40%–60%. • CC 2030: 1.8 GW in the AF25 scenario, compared to 1.0 GW in the BAT+ scenario; • CT 2030: 2.1 GW in the AF25 scenario, compared to 1.2 GW in the BAT+ scenario; • CC 2035: 2.3 GW in the AF25 scenario, compared to 1.0 GW in the BAT+ scenario; • CT 2035: 2.9 GW in the AF25 scenario, compared to 1.4 GW in the BAT+ scenario.

FNA report for Belgium 2026, version approved by CREG

123

These volumes represent residual flexibility resource needs close to real-time during periods of incompressibility risk, beyond the flexibility already assumed to be available in the scenarios, including batteries, large renewable units, end-user flexibility and cross-border contributions. In practical terms, they indicate the additional flexibility that would need to be available in intraday and balancing timeframes to address short‑term forecast errors and forced outages. The higher residual flexibility needs observed in the CT scenarios reflect the lower availability of flexibility resources and lower demand levels assumed in that scenario compared to the CC scenarios.

Figure 79: Residual uncovered total short-term flexibility needs (left) and fast flexibility needs (right) in the CC/CT-AF25 and CC/CT-BAT+ scenario expressed in GW generating power at the 1% reference reliability level

Of these total short-term flexibility volumes, the fast flexibility needs shown in Figure 79 (right) are significantly lower in the BAT+ scenarios, which include the additional battery potential identified in Adeqflex’25, than in the AF25 scenarios, which only include battery capacities secured through CRM auctions up to and including CRM 2024. • CC 2030: 0.2 GW in the BAT+ scenario, compared to 1.0 GW in the AF25 scenario; • CT 2030: 0.4 GW in the BAT+ scenario, compared to 1.2 GW in the AF25 scenario; • CC 2035: 0.1 GW in the BAT+ scenario, compared to 1.4 GW in the AF25 scenario; • CT 2035: 0.7 GW in the BAT+ scenario, compared to 2.2 GW in the AF25 scenario. Compared to AF25, the additional battery potential identified in Adeqflex’25 reduces the residual fast flexibility needs by approximately 0.8 GW in 2030 and 1.3–1.5 GW in 2035, corresponding to reductions of around 65–90%, depending on the scenario. The largest effect is observed in the CC 2035 scenario, where the residual fast flexibility needs decrease from 1.4 GW to 0.1 GW. Sensitivities to assess the capability of additional renewable flexibility to cover uncovered downward needs First, a sensitivity is conducted assuming full availability of decentralised solar generation to provide downward flexibility. This is found to mitigate the identified residual flexibility needs under the 1% reference reliability level. This reflects the fact that uncovered events mainly occur during periods with solar generation. Second, a sensitivity is conducted with full availability of decentralised wind generation to provide downward flexibility, as illustrated in Figure 80. The sensitivity should be compared with the uncovered total short-term flexibility needs of Figure 79 (left). As such, it allows the contribution of decentralised wind to be assessed by comparing the residual flexibility needs before and after accounting for this flexibility source.

FNA report for Belgium 2026, version approved by CREG

124

Figure 80 Uncovered total short-term flexibility needs to unlock on decentralised assets after participation of decentralised wind in the CC/CT-AF25 and CC/CT-Bat+ scenarios

The results indicate that decentralised wind generation can contribute to reducing the identified uncovered total short-term flexibility needs, with the magnitude of the impact depending on the scenario and target year. In the AF25 scenarios, which only include battery capacities secured through CRM auctions up to and including CRM 2024, accounting for decentralised wind flexibility reduces the remaining uncovered total short-term flexibility needs: • from 1.8 GW to 1.0 GW in CC 2030; • from 2.1 GW to 1.6 GW in CT 2030; • from 2.3 GW to 1.3 GW in CC 2035; • from 2.9 GW to 2.2 GW in CT 2035. In the BAT+ scenarios, which incorporate the additional battery potential identified in Adeqflex’25, accounting for decentralised wind flexibility reduces the remaining uncovered total short-term flexibility needs: • from 1.0 GW to 0.2 GW in CC 2030; • from 1.2 GW to 0.7 GW in CT 2030; • from 1.0 GW to 0.0 GW in CC 2035; • from 1.4 GW to 0.8 GW in CT 2035. Compared to AF25, the BAT+ scenarios show substantially lower residual flexibility needs after accounting for all decentralised wind flexibility. This reflects the contribution of the additional battery potential identified in Adeqflex’25. In this scenario batteries become a dominant source of flexibility, significantly lowering the need for additional flexibility resources and improving the system’s ability to respond to short‑term variations. In the CC-BAT+ scenario, the remaining uncovered total short-term flexibility needs are reduced to only 0.2 GW in 2030 and are fully eliminated in 2035 after accounting for decentralised wind flexibility. This indicates that, under the assumptions of the assessment, the combination of the additional battery potential identified in Adeqflex’25 and decentralised wind flexibility is sufficient to address all identified uncovered total short-term flexibility needs in the CC-BAT+ scenario in 2035. In the CT-BAT+ scenario, residual needs remain after accounting for decentralised wind flexibility, amounting to 0.7 GW in 2030 and 0.8 GW in 2035. These remaining volumes reflect the lower availability of flexibility resources assumed in the CT scenario compared to the CC scenario. Where residual needs remain, particularly in the AF25 scenarios which only include battery capacities secured through CRM auctions up to and including CRM 2024, and in the more constrained CT scenarios, additional flexibility resources are required beyond decentralised wind flexibility. These FNA report for Belgium 2026, version approved by CREG

125

situations are generally associated with periods characterised by high solar generation and limited wind availability. The comparison between AF25 and BAT+ indicates that part of these residual needs could be addressed through additional battery deployment beyond the capacities already secured through CRM auctions up to and including CRM 2024. To the extent that future adequacy needs result in additional flexibility resources being developed through the CRM, these additional capacities could therefore contribute to reducing at least part of the residual flexibility needs identified in the AF25 scenarios.

FNA report for Belgium 2026, version approved by CREG

126

7. Results - Network flexibility needs The FNA method foresees and assesses how grid constraints affect the flexibility needs of the system. Its objective is to assess additional RES curtailment resulting from network limitations. In this first version of the FNA, network needs are assessed across four categories as shown in Figure 81 : TSO network needs for which all analysis out of scope for FNA 2026. An implementation plan for the FNA 2028 is presented in Section 7.1. It presents an implementation plan to study the flexibility needs following congestions in the transmission grid, referred to as TSO network needs. The implementation plan is for the next version of this report foreseen in 2028. As already presented in the implementation plan, the timing of this first version did not allow to develop the specific methods in line with the FNA method. Network needs on T/DSO connection points for which a quantitative analysis is conducted on RES curtailment on distribution level following constraints on the TSO transformer at the T/DS connection point. This analysis is conducted under Article 11 of the FNA methodology. Data and analyses provided by DSOs based on their DNDPs, i.e. latest distribution network development plans (investment plans) approved where available. An overview of the main assumptions on renewable capacity and flexibility is given in Section 3 and Annex D. The scope is limited to downward flexibility needs. Note that the analysis of the network needs on T/DSO connection point focuses only on downward flexibility as best aligned with scope of the system needs on RES integration (while upward flexibility needs, cf. on the demand side, is already covered in investment plans and related discussions).

Figure 81: scope definition of network needs in FNA 2026

DSO network needs for which a qualitative analysis is conducted on voltage and internal congestions. Quantitative analysis are out of scope for the FNA 2026 . Note that the scope of quantitative assessments may expand in the FNA 2028. This is elaborated in Section 7.3. Unavailability of flexible resources for which a quantitative analysis is conducted on TSO-level following grid constraints which is based on information available in flexible connection agreements made available in connection studies. Section 7.4 foresees an analysis of the unavailability of flexible resources due to grid constraints. As these are not accounted in the simulations or calculations, the flexibility needs resulting from the RES integration needs and Short-term flexibility needs assessment could be under-estimated. The assessment focuses on the most relevant case for this study: offtake limitations of flexible resources, and in particular large batteries, during high RES conditions. Note that the analysis of the network needs on T/DSO connection point focuses only on upward flexibility as best aligned with scope of the system needs on RES integration (while upward flexibility needs, cf. on the demand side, is already covered in investment plans and related discussions). First step in quantifying the total RES curtailment on T/DS-level following availability of data. The analyses on RES curtailment following voltage issues or internal congestions assessed qualitatively DSO scope may expand in FNA 2028. FNA report for Belgium 2026, version approved by CREG

127

7.1. TSO network needs Article 12(1) of the FNA method allows TSOs to quantify downward network needs resulting in RES generation curtailment due to transmission network constraints. This analysis is required if transmission network constraints within the bidding zone are not yet accounted in the economic dispatch simulation used for this FNA. Indeed, in such case, the impact of grid constraints would be implicitly accounted for in the calculation of the system flexibility needs. The economic dispatch model used by Elia for the FNA does not currently support accounting network constraints other than the flow-based network constraints. Furthermore, even if internal network elements are considered in the flow-based constraints, the location of the congestion inside each bidding zone cannot be assessed using simulations at bidding zone level. Such RES curtailment can actually only be determined by means of detailed network models, and the methods to identify TSO network needs in line with the FNA are still to be developed. As the identification of the TSO network needs is identified in the FNA method as a voluntary task, yet considered relevant to obtain an integrated view on flexibility needs, Elia is required to integrate this part as from the next version of the FNA. In line with the objectives of the FNA, the TSO network needs should be understood as flexibility needs arising from grid constraints within the bidding zone. Their assessment aims to identify situations where network constraints may limit the efficient integration of renewable generation or require additional flexibility to maintain secure system operation. The assessment of TSO network needs is expected to be further developed in future editions of the FNA. The methodology, modelling assumptions, level of granularity and applicable scenarios remain to be defined through the FNA implementation process, in accordance with the ACER methodology and the applicable legal framework. Any detailed methodology would be submitted to the CREG for approval before implementation. The cost and benefit analysis referred to in Article 12(5) of the FNA method requires to justify the use of flexible solutions as an alternative to grid expansion. Based on the system needs and the TSO network needs identified in a given area, insights can be derived regarding the interaction between flexibility solutions and network reinforcement options. These insights can be captured in the next federal development plan (FDP 2032-2042). In that sense, it is important that the objective of the FNA remains the identification of flexibility needs associated with the integration of renewable generation and the maintenance of system security. While it can assess flexible solutions alongside network reinforcement options, its objective lays not in identifying grid investment needs. This remains exclusively the objective of the federal development plan. This is similar on how DSO network needs in the FNA can serve as input of the DSO distribution capacity investment. Note that the network needs analysis in that sense further integrates the different studies in Belgium. However, for development plans new scenarios are created, which may or may not be in line with adequacy/flexibility study scenarios as these follow a different approval process and timings. Currently there is a best effort of alignment, but differences may and will exist (due to constraints during different timelines and different processes). This has to be taken into account when TSO, and also DSOs use the FNA as an input in the next development plans. As illustrated in Figure 82, the FNA is therefore to be seen as an input for the federal and regional development plans. It may identify areas where network constraints warrant further assessment in the context of the federal and regional development plans., with the updated set of scenarios at the time of the elaboration of those development plans. In addition, information on RES curtailment or unavailability of generation, storage or demand assets can serve as an input for the next adequacy study.

FNA report for Belgium 2026, version approved by CREG

128

Figure 82: overview of the different electricity system studies in Belgium

7.2. Network needs on T/DSO connection points This section is the result of an assessment performed by the DSOs (Fluvius, Ores, Resa, Sibelga, Arewal). The implementation of the FNA methodology is described in Annex D and was developed in concertation with Elia.

7.2.1. Context The RES integration needs in Section 4, and the system security needs in Section 5 and Section 6 are determined by the expected RES curtailment on system level. The FNA method foresees to consider that additional RES curtailment may occur following grid constraints in the distribution grid. The FNA methodology requires network-related flexibility needs to be assessed. In this first implementation, these needs are quantified through the RES curtailment on distribution level expected to occur as a consequence of transmission grid infrastructure. This section therefore identifies the expected RES curtailment in 2030 and 2035 for the different distribution grids in Belgium. As renewable generation continues to grow, the DSO grid is increasingly injecting energy into the TSO network at certain moments in time, potentially overloading the transformers on the T/DSO coupling point. The RES curtailment in the distribution grid to avoid such overloading is calculated by each concerned DSO individually using harmonized calculation assumptions. The method is presented in Annex D but is based on an assessment of the local injections exceeding the available capacity of the transformer on the T/DSO point of coupling during representative days : • Local injection: is based on the load profile in the DSO grid is based on the DSO’s own distribution network development plan (DNDP). Note that no demand side flexibility (including batteries) was taken into account as (i) DNDPs do not quantify available downward flexibility of end-users or batteries, and (ii) currently no congestion markets exist in DSO grids. These assumptions can be seen as conservative and result in an over-estimation of the flexibility needs • Available capacity: the T/DSO coupling point power threshold for injection is defined as 90% of the sum of current active transformers present at the concerned substation, in line with current congestion management practices. Confirmed grid reinforcements have been taken into account. Fluvius also took into account grid reinforcements it considered probable. • Representative days: the same set of representative days based on load and RES generation conditions are used for quantifying excess generation.

FNA report for Belgium 2026, version approved by CREG

129

For each of these T/DSO coupling points and for each representative day, the Network Needs (Curtailment) to prevent congestion are calculated. Note that – since no available flexibility means are accessible for activation by the DSOs (at least to cover for grid related issues) – currently all Network Needs would need to result in RES curtailment. Such curtailment is possible today in Flanders by means of technical measures (tele-control steering) on RES installations above 400 kW, above 500 kW in Brussels and 250 kW in Wallonia. Smaller installations are currently not directly controllable by the DSO. This could lead to uncovered network needs when the needs are driven by small-scale solar power.

7.2.2. Results In this first implementation, network needs are quantified through the RES curtailment resulting from constraints at T/DSO coupling points. RES curtailment is used here as the indicator of RES curtailment (flexibility needs) at distribution level following congestions on the transformer of the transmission system operator at the T/DSO coupling point. The results presented in Table 11 show distinct patterns across the three Belgian regions. Three indicators are used to quantify network needs through the associated RES curtailment in 2030 and 2035: • Total number of coupling points with RES curtailment: based on any occurrence of RES curtailment on a T/DSO coupling point, even if very small, during at least one hour or quarterhour in the target year , • Total Network Need: Annual RES curtailment resulting from constraints at T/DSO coupling over all T/DSO coupling point per DSO following aggregation of all RES curtailment over all hours or quarter-hours in the target year, • Peak Network Need: Maximum RES curtailment power resulting from T/DSO coupling point constraints over all T/DSO coupling points per DSO following the maximum observed value on a representative day (aggregation over DSOs is made for the representative day with the largest RES curtailment). Table 11: Summary of Network Needs at T/DSO Coupling Points Region

Flanders

DSO

Fluvius

ORES

Resa

Arewal

Sibelga

240

120

424

Total number of coupling points with simulated congestion from RES production in 2030

Total Network Need in 2030, (summed over all connection points)

GWh

0,5

Peak Network Need in 2030, (summed over all connection points)

MW(3)

74(2)

Total number of coupling points with simulated congestion from RES production in 2035 (1)

Total Network Need in 2035, (summed over all coupling points)

GWh

133

Peak Network Need in 2035, (summed over all coupling points)

MW(3)

506

278

546

Total number of coupling points (1)

Wallonia

Brussels

Total

2030

2035

The total does not equal the sum of the individual DSOs as the DSO-specific count includes coupling points shared with other DSOs.

FNA report for Belgium 2026, version approved by CREG

130

Since the peak power of Fluvius and Arewal is based on the worst hour of the year and the peak power of other DSOs is based on the worst power of the representative days, the Fluvius/Arewal a result is more extreme. (See Appendix D for more information on the methodology for producing representative day results for each DSO). 3

The peak curtailment for Fluvius is on sunny days, the peak curtailment for ORES and Arewal is on windy days. The joint peak is on windy days, and therefore does not match the sum of the columns in the table.

For Flanders (DSO: Fluvius), there is a limited amount of curtailment in 2030, for a total of 0.5 GWh and a peak power of 0,1 GW distributed over 12 coupling points (5%). By 2035 there is an increase in curtailment, totalling 66 GWh and a peak power of 0,5 GW summed over 36 (15%) affected Flemish coupling points. There are only four connection points where more than 1% of annual local RES generation is curtailed. In most of these cases, this is driven by a strong increase in additional wind power towards 2035, on top of an already significant solar generation capacity. The contribution of both solar and wind generation can be observed in Figure 83. Curtailment generally occurs during periods of high renewable generation and low demand. While most events take place during daytime hours, indicating a contribution from solar generation, significant curtailment is also observed during windy conditions, including on representative days with curtailment during nighttime hours. This highlights that network needs are driven by a combination of solar and wind generation depending on the local generation mix and system conditions. Figure 84 shows the annual RES curtailment at affected connection points expressed in percentage of the RES generation behind that coupling point. In 2030, only one of the twelve affected Flemish coupling points exceeds 0.1% curtailment. In 2035, there are two Flemish coupling points where the share of lost energy exceeds 3%. At other coupling points, except the five with the highest curtailment, the network needs are largely driven by small-scale low-voltage photovoltaic installations. Fluvius foresees potential issues in a few coupling points, in managing the congestion with high contribution of smallscale low-voltage photovoltaic installations. Dependent on the impact of market evolutions and the actual realisation of the foreseen scenarios, the most appropriate technical solution will be timely implemented.

Figure 83: Daily power profiles for representative days with yearly peak curtailment power (Fluvius). Only representative days with low load are shown.

FNA report for Belgium 2026, version approved by CREG

131

Figure 84: Curtailed energy at affected coupling points, relative to RES generation behind the coupling point. Only coupling points with more than 0,1% curtailment are shown.

The coupling points with the highest curtailment are typically coupling points with a relatively low capacity, and with high amounts of new wind energy expected in the scenarios for 2030 & 2035. There are various options possible to connect these amounts of wind energy, e.g. a connection to another coupling point nearby, to optimise the balance between investment and curtailment. As the exact location of the expected wind energy is not known yet, those alternative connection options are not yet integrated in the grid model of 2030 / 2035. Therefore, for this study, all expected wind power for the scenarios 2030/2035 has been attributed to the existing coupling points. The simulated curtailment is then to be considered as an absolute maximum (and possibly an overestimation). For Wallonia (DSOs: ORES, RESA, Arewal), the network needs are fully driven by wind power and concentrated on days with low consumption. By 2030, the curtailment attains a level of 11 GWh, and increases to 67 GWh in 2035. The peak curtailment power is always lower than in Flanders, which can also be attributed to the difference that the power was computed between the DSOs 5. It is also the result of a strong increase in additional wind energy towards 2035 in Wallonia, increasing with 180% relative to 2020. This increase far exceeds the expected local growth in consumption across all voltage levels. The number of Walloon coupling points affected by congestion doubles from 7 in 2030 to 15 in 2035. The amount of coupling points with more than 1% curtailment equally increases from 3 to 9, as indicated in Figure 84. At the strongest affected coupling point, curtailment reaches 5% of annual RES production. RESA did not identify a curtailment need on the representative days defined in the methodology (Annex D). Still, when local generation in on some coupling points is exceptionally high, congestion is possible. Since these situations are rare, the corresponding curtailed energy is however negligible. For Brussels (DSO: Sibelga), no congestion at the T/DSO coupling point due to RES production is observed. While PV production is expected to double in the coming decade (from 311 in 2023 to 635 MWp in 2035), PV production will mostly be compensated by the expected increase of load due to the electrification of the mobility and heating devices in Brussels. Qualitatively, PV production in 2035 is expected to stay below 10% of the current Brussels electricity demand in terms of energy, and less than

Fluvius peak curtailment power is based on the most extreme day, not on the average representative day, whereas the peak curtailment power for Ores is based on the maximum hourly value of the generic representative day used for calculating curtailed energy.

FNA report for Belgium 2026, version approved by CREG

132

60% of the current peak load of the Brussels-Capital Region. Wind production is negligible in the Brussels-Capital Region: only 2 micro wind turbines exist today. All in all, RES production will increase significantly in Brussels and will increase the amplitude and the frequency of the electricity injection towards the TSO grid at the T/DSO coupling points. However, this increase of injection will stay in the range of acceptable operation limits of the T/DSO coupling point transformers in Brussels. As shown in Table 11 for the whole of Belgium, RES curtailments in the DSO grid following constraints at the T/DSO coupling point (referred here as network needs) amount to 11 GWh in 2030 (0.04% of Belgian RES production). They amount to 133 GWh in 2035, which is still less than 0,3% of the total RES generation in that same year. The peak power increases from 0.1 GW to 0.5 GW. Wind power is by far the main driver for curtailed energy. Note that the peak does not substantially increases when aggregating Wallonia and Flanders, which is explained by the difference in driver for RES curtailment: the peak curtailment in Belgium is driven by wind, with a lower corresponding peak curtailment in Flanders. For Belgium, most of the network needs at the T/DSO coupling points occur during windy days and are therefore (at least) driven by wind power. RES curtailment is also observed on windy days with high load, on sunny days and on some normal days 6 (resulting from the methodology), but to much smaller extent. Figure 85 shows that curtailment on windy representative days is much more pronounced than on sunny representative days. The curtailment peaks during the sunny hours reveal that solar power still contributes to the curtailment. The total yearly curtailment is very similar in both regions (Table 11). Figure 86 shows a duration curve for the aggregate network needs over all DSOs. It demonstrates that relevant RES curtailments (e.g. higher as 100 MW) are concentrated in about 350 hours per year in 2035. This accounts for 4% of the year, driven by wind power and low load.

Resulting from the methodology which classifies all weather profiles inside the thresholds of Annex D as normal representative days.

FNA report for Belgium 2026, version approved by CREG

133

Figure 85: Cumulative network needs (curtailment) on average representative days with pronounced curtailment.

Figure 86: Duration curve of total network needs for 1 year (only 1000 hours with highest curtailment shown).

7.2.3. Conclusions FNA report for Belgium 2026, version approved by CREG

134

Figure 87 shows that DSO-related curtailment remains limited when expressed as a share of total renewable generation, even by 2035, and is therefore not considered a key driver of overall system needs. The annual curtailments are almost equally distributed over Wallonia (wind driven) and Flanders (wind and solar driven). No curtailment is observed in Brussels. In 2035 these needs amount up to 14% - 37% of the RES integration needs (depending if compared to AF25, or BAT+ scenario results) irrespective of the CC / CT scenario. It is calculated as the RES curtailment on distribution system level compared to the RES integration needs (before finetuning). Expressed in percentage of total RES generation in Belgium, this amounts only to 0.3% of total RES. Curtailment is primarily concentrated at a limited number of T/DSO coupling points and occurs under specific system conditions, with peak values reaching around 550 MW. Curtailment is most pronounced during low load and high wind conditions, particularly in Wallonia. In Flanders, curtailment is also expected during windy high load conditions or sunny low load conditions. Current estimates are likely conservative, as they do not yet account for future flexibility activations or additional grid reinforcements. The impact of such measures could be assessed in future iterations of the methodology. Finally, this first FNA introduces a methodology to assess distribution-level curtailment through transmission system infrastructure constraints on the connection points between TSO and DSO. Future assessments may further refine the methodology and extend its scope as additional data and operational experience become available.

Figure 87: Annual RES curtailment following T/DSO coupling point constraints

7.3. DSO network needs This section is the result of an assessment performed by the DSOs

As decentral generation grows in distribution networks, DSO grids face new challenges. While these networks were designed to transport power from the transmission grid to large amounts of consumers, FNA report for Belgium 2026, version approved by CREG

135

power flows have become bidirectional, regularly feeding power towards upstream into the radial network layout. This comes with new challenges for DSOs: • TSO/DSO coupling point congestion. • DSO network overvoltage • DSO network congestion This first FNA quantitatively assesses the RES curtailment following constraints on the T/DSO point of coupling (5 kV – 63 kV on the DSO side), driven by the capacity of the transformer, and qualitatively assesses the other challenges. The next version of the FNA in 2028 foresees further extending the objectives and scope of the analysis.

7.3.1. Overvoltage The rapid deployment of residential photovoltaic (PV) installations has increased the occurrence of local overvoltage events in parts of the low-voltage (LV) network where high distributed generation coincides with limited network hosting capacity existing LV networks, particularly in rural distribution networks. Flanders Due to the large increase in the number and additional capacity of solar panel installations, the number of complaints about the shut down of inverters has risen remarkably since 2022, with a peak in PV complaints in the spring of 2023. Particularly in the province of Limburg, there is 20% more PV capacity per capita than in West Flanders, the second on the list. The installations are also approximately 10% larger than in the other provinces. The biggest local problems occur in the spring on sunny days. During these periods, solar generation is high due to favourable irradiation conditions, while relatively low temperatures further increase PV panel efficiency. The root causes of the voltage complaints and the shut down of inverters are often complex and multifactorial. Therefore, a thorough and rapid root cause analysis is important. The solution can also vary, and the cause must sometimes be determined on a case-by-case basis. Depending on the type of grid, the grid configuration, and the grid components, various solutions may be appropriate to remedy the problem in the short or long term: ranging from a "simple" operational measure to the installation of new cables, a new distribution substation, or other additional grid components. Fluvius has therefore rolled out a voltage complaint action plan to address and reduce inverter trips caused by voltage issues in solar panel installations. The general objective is to establish a structural approach within Fluvius to better map complaints, to resolve them effectively and sustainably, and to prevent complaints. The technological component of this action plan includes the following policy and technology choices that are being implemented or investigated: • reference voltage: the specified voltage is adjusted. • compounding: current compensation in the automatic step regulator of a high-voltage/mediumvoltage transformer. • 150 mm² aluminium bundle cable: introduced at low voltage (larger cross-section on overhead grids to reduce voltage variations). • PRC 240 mm²: this will become the minimum cross-section for underground low-voltage cables. • Pilot projects within Fluvius with technological solutions that have not been used, or have been used very little, in the past. • medium voltage/medium-voltage transformers: where the medium-voltage network is exposed to excessive voltage drops.

FNA report for Belgium 2026, version approved by CREG

136

• OLTC (On Load Tap Changer): automatic tap position regulator for medium-voltage/lowvoltage transformers, which avoids manual intervention (the transformer does not need to be de-energized). • LVR (Line Voltage Regulator): keeps the voltage between the operating limits at the end of a low-voltage network. Naturally, there are traditional operational procedures that offer a temporary or permanent solution. These actions are usually reactive: • manual tap position changes, • installing a neutral point transformer on the low-voltage grid in case of phase imbalance, • rewiring the load on the phases of the low-voltage grid so that the load is distributed more evenly, • voltage conversion from the 230 V grid to 400 V Furthermore, together with the other DSOs, Fluvius investigated which control options can be configured in inverters so that they actively contribute to voltage management on the low-voltage grid. This would be applicable to new inverters. The implementation of this action plan has contributed to a reduction in the number of reported voltage complaints in 2025 compared with previous years. Wallonia The large-scale activation of the Power Quality (PQ) function of smart meters allows continuous monitoring of voltage levels at the customer connection point. This provides objective, data driven identification of overvoltage areas. Enhanced observability supports proactive planning and more effective prioritization of corrective actions. ORES combines PQ measurements, outage statistics, compliance with design rules and load growth projections to rank LV feeders according to their risk level. This scoring approach supports targeted interventions and improves the efficiency of operational measures and investment decisions. Such measures include phase balancing, transformer tap adjustment, neutral point compensation, low-voltage voltage regulators, cable reinforcement and replacement, conversion from 230V to 400V and network reconfiguration and construction of new substations.

7.3.2. Internal DSO Congestion The strong correlation of local renewable generation and related peaks in local injection may also stress transformers, lines and cables within the DSO network when power flows exceed the thermal limits of these elements. To avoid exceeding thermal safety thresholds, DSOs may reduce the production of larger production installations that are equipped with a remote-control cabinet. Currently, no such activations happen in Belgium on a structural basis. They are only applied sporadically in the context outage management (e.g. for grid maintenance), and the related curtailment is therefore small (in fact the volumes are so small that, just the occasional testing of the telecontrol functionality accounts already for 22% of the total curtailment time). This makes it difficult (and little relevant) to extrapolate such curtailment towards 2030 and 2035. Moreover, all DSOs closely monitor their main grid elements and plan for timely grid reinforcement in their DNDP to avoid such curtailments becoming structural.

7.4. Unavailability following grid constraints When quantifying system needs in the previous sections, it should be considered that flexible resources might not always be fully available. This could result in an under-estimating flexibility needs when FNA report for Belgium 2026, version approved by CREG

137

calculated using assets which would it practice not fully available. Whereas planned and unplanned outages are accounted for in the economic dispatch simulation, grid capacity limitations hindering the free dispatch of flexible resources are not. To account for these limitations in the FNA, the RES integration needs and system security needs calculated can be corrected for the expected unavailability of flexible resources following grid constraints. RES integration needs and System security needs identified in these sections are found to be strongly driven by high renewable generation, either by their excess generation, simultaneous RES generation or generation forecast errors. Consequently, it makes most sense to study the unavailability of flexible resources specifically in such situations with high renewable generation, and specifically the unavailability of flexible offtake to absorb the renewable generation. Grid-based limitations are currently imposed in two ways: • Directly by the TSO, by overruling of an asset’s planned exchange with the grid. Two solutions exist to impose such limits: • Redispatch (“iCAROS”) addresses non-structural congestion and is applicable to units with a maximal power of 25 MW and above. It allows to start-up, shut-down and modulate assets. • Grid User Flex (“GUFlex”) addresses structural congestion and is applied to flexible connection agreements (Grid Users connecting before necessary grid reinforcements). It allows the system operator to limit an asset’s exchange with the grid to the communicated threshold value. • Indirectly, through “Return to Schedule” restrictions and balancing energy bid filtering based on the Congestion Risk Indicator (CRI). This CRI indicates if Elia takes (or is expected to take) any direct actions to manage the congestion in the considered grid area. In such case, aggravating the congestion is avoided by (i) requesting large production installations in that grid area to not deviate from schedules in the congested direction and return to their last schedule, and (ii) filtering balancing energy bids in the congested direction that include delivery points in the concerned electrical zone (in practice, balancing service providers have a best-effort obligation to reorganise their balancing energy bids in order to provide their contracted balancing capacity volume with delivery points outside the concerned electrical zone). 7.4.1.1. Available data GUFlex is designed to allow Grid Users to connect even when the needed grid reinforcements are not yet commissioned. The future impact of GUFlex will depend on the evolution of connection requests, grid reinforcements and congestion-management practices. Consequently, its future contribution to flexibility limitations remains uncertain. As GUFlex contracts are relatively new for flexible resources, historic data is still very limited in time and number of involved flexible resources, making it difficult to make any meaningful extrapolations to the future. Until 2026, FRR energy bid filtering has been applied solely to mFRR energy bids (a process is in place for aFRR energy bids but it did not need to be used in practice yet). The observed CRI-filtering of mFRRdown flexibility remains low considering the current CRI process and is not considered suitable for any extrapolation towards the FNA target years. A shortcoming of deriving unavailability from this dataset is that BSPs have a best effort obligation to reorganize their FRR bids based on the indicative CRI level to limit the lost volume to a minimum. Therefore, congestion-prone FRR energy bids (representing relevant impacted offtake volumes) may have been replaced ahead of real-time by the BSP and may not be observed in such analysis. Therefore, given the limitations of the currently available operational data, the results of completed grid connection studies for flexible resources were used as an exploratory source of information for this assessment. These studies assess potential network constraints over the relevant time horizon and provide indications regarding possible offtake limitations during periods of high renewable generation. The results presented below should therefore be interpreted as an indicative assessment only and not as a validated methodology for future FNA cycles FNA report for Belgium 2026, version approved by CREG

138

Each detailed grid connection study (or “EDS”) assesses – for 100 representative future situations (“points in time”) – the limitations required on the connection to avoid network congestion. These situations are characterised by various indicators such as the amount of renewable generation. By considering situations with high renewable generation and aggregating the results of the available flexible resource EDS, an indicative estimate of potential offtake limitations can be derived. This analysis was performed on data for all completed EDS for flexible resources 7 on the TSO grid. As the average renewable production during excess generation events is around 60% of the maximum production, ‘high renewable generation’ was therefore defined as ‘exceeding 60%’, but the calculation is equally performed for lower threshold values. The use of EDS data in this assessment should be considered exploratory, as connection studies are based on project requests that may not ultimately materialise and are not intended to represent the most likely future system development. Consequently, the results should be interpreted with appropriate caution. 7.4.1.2. Result The exploratory assessment indicates that, under the assumptions considered and for periods of high RES production, approximately 1% of the offtake capacity of the investigated flexible resources may be subject to grid-related limitations in both 2030 and 2035. This has a logical explanation. As most renewables and flexible resources are distributed over the grid, a flexible resource’s offtake will in many cases use locally generated renewable production and hence reduce the use of the overall network to transport this produced energy. Depending on the circumstances and location, offtake by flexible resources could even reduce grid loading and relieve congestions. Note that the calculated unavailabilities will decrease beyond 2035 as the remaining grid reinforcements following the structural congestion from the new connection are being completed. A sensitivity analysis reveals that offtake limitations increase as lower RES generation situations are considered. For all sensitivities calculated, the limitation to offtake power however remains below 10% of the installed flexible capacity (Figure 88) and the RES integrations or System Security needs in Section 2 and Section 3 are not fine-tuned for the unavailability of flexible resources. It is observed that for system conditions where renewables are expected to inject at least 60% of their maximum production capacity, total offtake restrictions (i.e. capacity of the assets not available for off-take due to grid constraints) of the resources investigated remains at 1% of total offtake capacity on average, in 2030 as well as in 2035. During moments of lower RES generation, offtake limitations on flexible resources become more pronounced. For instance, where renewables are expected to inject above 45% of their capacity, total offtake restrictions of the resources investigated are already at 7% of total offtake capacity on average, in 2030 as well as in 2035.

All flexible resources for which an EDS is issued are BESS.

FNA report for Belgium 2026, version approved by CREG

139

Figure 88: Unavailability of flexible resources to offtake power from the grid, based on all completed EDS studies.

The estimated limitations under high RES conditions are driven by a limited number of EDS. Out of more than 100 considered EDS, only 2 studies showed a need for offtake limitation when renewable generation exceeds 60% in 2030. Note that an EDS is a study request and does not mean the concerned project will be built. If an EDS indicates a significant unavailability of the connection, that may negatively affect the investment decision to build flexible assets in these locations. It should also be noted that EDS studies are performed following contractual engagements, meaning that most likely the actual impact on flexible resources is even smaller. As mentioned above, the analysis was performed on all completed EDS for flexible resources, totalling 6800 MW of connection requests by 2030 and 7600 MW by 2035. This difference stems from some flexible resources only being planned to be commissioned after 2030. These later projects also cause the 2035 limitation to sometimes slightly exceed the 2030 one in Figure 88. This amount of centralised flexible resources is higher than is currently accounted for in the scenarios (to be connected to the grid in the target years). Consequently, the analysis should primarily be regarded as a sensitivity exercise illustrating the potential impact of connection-related constraints rather than an estimate of future operational limitations. By considering fewer studies, the impact on a smaller capacity of flexible resources was checked. In the reference calculation (>60% RES) the limitations are already very low and stem from only 2 EDS. To obtain a more insightful sensitivity on the amount of EDS considered, it was performed on the >50% RES threshold (which corresponds to 11 EDS with offtake limitations among all EDS). When analysing a smaller set of EDS, e.g. the 2500 MW oldest active connection requests, this restricted offtake flexibility drops from 5% of total offtake capacity on average to 2% in 2030. By 2035, the offtake restriction drops further to 0% as further grid reinforcements were completed for hosting these new resources. This reveals that for additional flexible capacity ranging from 2500 to 5000 MW, the offtake limitations during moment with high RES generation would be smaller (Figure 89).

FNA report for Belgium 2026, version approved by CREG

140

Figure 89: Sensitivity of estimated offtake limitations to the number of EDS included in the assessment.

7.4.1.3. Future improvements Each EDS accounts for hosting the capacity of all previous issued EDS of other resources (flexible resources, demand and generation), while those are not expected to be all realized. This is a drawback of this analysis and the results can therefore be expected to represent an upper limit of the limited flexibility under the circumstances considered. The analysis is also based on the asset limitation results of the EDS, and does not consider the specific network congestions leading to these unavailabilities. This hampers a deeper analysis on which grid elements are most limiting for hosting additional flexible resources. A more robust assessment of grid constraints affecting flexible resources could rely on detailed network simulations based on plausible assumptions regarding the development of demand, flexible resources and network infrastructure. Such simulations would facilitate the identification of potential network bottlenecks and the assessment of possible mitigation measures. The appropriate methodology and data sources remain to be further assessed through the FNA implementation process.

FNA report for Belgium 2026, version approved by CREG

141

8. Synthesis - Guiding criteria application outcomes The FNA aims to assess the overall flexibility needs of the electricity system in a technology-neutral manner, in accordance with the European FNA methodology established by ENTSO-E and DSO Entity and approved by ACER. Its results are intended to support Member States in defining indicative national objectives for non-fossil flexibility and justify the need for support mechanisms where necessary. This synthesis therefore starts firstly with a finetuning of the RES integration needs where all system needs and network needs presented in the previous sections are combined in view of one overall flexibility need. These represent the uncovered downward flexibility needs after accounting all existing and expected flexibility through the four scenarios. In absence of an explicit RES integration target in Belgium, in particular regarding acceptable levels of RES curtailment, the assessment of the finetuned RES integration needs is carried out using a range of RES curtailment levels. This approach allows estimating the additional flexibility needs associated with different levels of acceptable RES curtailment. In this context, the analysis makes it possible to derive: • the additional non-fossil fuel flexibility required to achieve full RES integration, implying a zero curtailment; • and, conversely, the level of RES curtailment, derived from the situation without additional nonfossil flexibility. In this case, the resulting level of RES curtailment defines the required contribution from renewable flexibility. After accounting for the flexibility already available or expected from renewable generation, the remaining renewable flexibility needs can be determined. Where curtailment remains, additional RES generation capacity may be required to compensate for the curtailed energy and achieve the NECP target. The level of RES curtailment can be assessed both in terms of energy (on a yearly basis) and capacity (on an instantaneous basis). Based on these results, the assessment can provide insights into the extent to which the existing and expected response of renewable assets is sufficient to manage RES curtailment in the absence of additional flexibility, or whether additional response may need to be unlocked. These two boundary cases do not represent preferred pathways but provide a transparent framework to assess the trade-off between deploying additional non-fossil flexibility and accepting RES curtailment. Where curtailment is accepted, additional renewable response may be required to ensure secure system operation during periods of excess generation. In addition, where curtailed energy is not otherwise recovered, additional renewable generation capacity may be required to compensate for the curtailed energy and achieve the renewable energy targets set in the NECP. The FNA does not determine which pathway should be preferred and does not assess the relative economic efficiency of these options. Rather, it provides a transparent and consistent basis for assessing flexibility needs and their implications, which can then be used to inform public policy and investment decisions. In addition, the FNA identifies potential barriers in Section 9 to the development and use of flexibility, such as regulatory, market or technical constraints, thereby complementing the assessment of flexibility needs and supporting the effective deployment of flexibility solutions.

FNA report for Belgium 2026, version approved by CREG

142

Figure 90: synthesis of the flexibility needs assessment

8.1. Finetuning The methodology for finetuning the RES integration needs is described in Annex C.2 and combines the three building blocks for defining the flexibility needs: the RES integration needs, the short-term flexibility needs and ramping needs, and the network needs. The finetuning assessment is performed for four scenarios. The AF25 scenarios (CC-AF25 and CTAF25) are based on the Adeqflex’25 assumptions and only include battery capacities secured through CRM auctions up to and including CRM 2024. As Adeqflex’25 does not explicitly fill potential future adequacy gaps, these scenarios do not account for additional battery capacities that could be developed in response to future adequacy needs. The BAT+ scenarios (CC-BAT+ and CT-BAT+) therefore complement the assessment by incorporating the battery potential identified in Adeqflex’25, providing an indication of how additional storage deployment could reduce the identified flexibility needs. In a first step, the FNA Methodology (Article 8.3) requires the integration of the different system needs into one single indicator. When extracting RES generation curtailment time series in Section 4, the additional RES generation curtailment resulting from uncovered downward ramping needs and uncovered downward short-term flexibility needs pursuant to Articles 9(6) and 10(5) of the FNA methodology needs to be combined: • RES integration needs identified in Section 4 is the basis of the finetuned system needs; • Uncovered downward ramping needs identified in Section 5 are to be combined with the RES integration needs. However, as the ramping needs are found to be relatively small, and overlap entirely with RES integration needs or the short-term needs, these have no impact on the finetuning. • Uncovered downward short-term flexibility identified in Section 6 are combined with the RES integration needs. It reflects that curtailment caused by forecast errors can occur on top of the curtailment needed to manage well-predicted excess renewable generation. It accounts first for the probability of activation before making the sum with the RES integration needs. In summary, the formula for this summation is RES integration + MAX (ramping needs ; shortterm needs) where the ramping does not impact the calculation. In a second step, the FNA Methodology (Article 8.14) shall account the additional RES curtailment associated with downward needs at transmission and distribution level pursuant to and the unavailability of flexible resources pursuant Article 14 of the FNA Methodology. FNA report for Belgium 2026, version approved by CREG

143

• Network needs on TSO/DSO connection points, identified in Section 7.1, are accounted, although they remain limited compared to the RES integration needs. In the AF25 scenarios, they represent around 3% of the RES integration needs in 2030 and 14% in 2035. In the BAT+ scenarios, the corresponding shares amount to around 11% in 2030 and 37% in 2035. The higher percentage for the BAT+ scenarios is explained as the same network needs are compared with lower curtailment levels for system needs, resulting in higher percentage. In contrast, these represent around 0.3% of the annual renewable generation in all scenarios. As solving curtailment on DSO level also reduces the RES integration needs, the maximum of both is taken to avoid double counting. • TSO network needs discussed in Section 7.2 are also required by the FNA methodology but were not quantified in this first FNA and therefore not accounted. In summary, the formula for the finetuning becomes MAX(RES integration + MAX (ramping needs ; short-term needs); Network needs) where the ramping does not impact the calculation. Note that to avoid over-estimations of the total system needs following the sum of the RES integration needs with the short-term needs, following points are taken into account in line with the FNA methodology: (1) the associated probability of occurrence of the short-term needs (forecast error risk) is accounted; (2) the sum of the short-term needs and RES integration needs is determined on hourly basis, reducing the impact of events where high RES integration and short-term needs concur; (3) the short-term needs are determined dynamically, taking into account the relation between the short-term needs and the renewable generation, accounting for periods with lower prediction risks; Article 14 also requires considering the increase in uncovered system needs resulting from the unavailability of flexible resources due to network constraints. The assessment presented in Section 7.3 shows that these constraints, evaluated for transmission-connected battery assets during periods of high renewable generation, remain limited and are well below the defined finetuning threshold. Consequently, they are not further included in the RES integration or short-term flexibility needs. The resulting total flexibility needs for each scenario are presented in Figure 91. The indicator combines RES integration needs, short-term flexibility needs, ramping needs, and network needs into a single RES curtailment indicator. It represents the volume of renewable energy that would need to be curtailed in the absence of sufficient flexibility. For 2030 this equals about 0.4 – 0.5% of total RES generation in case of the BAT+ scenario with an increase up to 1.0 – 1.3 % in case of the AF25 scenario. In 2035 this increases towards 1.1% for the BAT+ scenario and 1.9 – 2.5% in case of the AF25 scenario. All scenarios see an increase compared to only taking into account RES integration needs. This curtailment can be addressed through additional flexibility resources, and/or managed through additional response from renewable generation in combination with additional renewable generation capacity to achieve the NECP target. As such, the indicator can be interpreted as the uncovered flexibility needs that remains to be addressed. It should be noted that this assessment only quantifies the curtailed energy volume (TWh). It does not provide information on the additional flexible peak power capacity (GW) required to manage these events. This power dimension is assessed separately in the following section.

FNA report for Belgium 2026, version approved by CREG

144

Figure 91: Total curtailment resulting from the different scenarios (with and without finetuning)

8.2. Guiding criteria Total finetuned flexibility needs are translated into additional flexibility requirements through the analysis of RES curtailment. In the absence of explicit curtailment targets in the Belgian NECP, different levels of curtailment are assessed, reflecting two extremes: full RES integration (zero curtailment) and the absence of additional flexibility.

8.2.1. Non-fossil fuel flexibility needs for RES integration needs The analysis is first performed assuming all RES to be integrated, implying zero curtailment. The left side of Figure 92 shows the amount of flexibility per duration type that would need to be installed to achieve full RES integration. The results show that very high amounts of flexibility are needed in this case. In 2030, the identified needs range from around 16 GW in the CC-BAT+ scenario to 19 GW in the CT-AF25 scenario. In 2035, these needs further increase and range from around 25 GW in the CTBAT+ scenario to 32 GW in the CC-AF25 scenario. It should be noted that these numbers reflect avoiding curtailment in all simulated Monte Carlo years and therefore correspond to a 100% certainty of avoiding curtailment under all simulated (weather) conditions. This is considered unrealistic, as it would require dimensioning flexibility on the most extreme event observed over all simulated Monte Carlo years. These values should therefore be interpreted as theoretical upper bounds rather than realistic deployment targets. For this reason, additional sensitivities are performed where a certain share of RES generation is allowed to be curtailed. The right side of Figure 92 shows the additional flexibility needs when allowing 0.1% of RES generation to be curtailed, this already shows a high decrease of installed capacity. The remainder of this section therefore focuses on sensitivities starting from 0.1% of RES curtailment (in energy). These sensitivities illustrate the trade-off between the level of acceptable curtailment and the amount of flexibility required. The assessment does not define the optimal level of curtailment, as this remains a policy choice. Note that the impact of the BAT+ scenario is not linear compared to the AF25 scenario, an addition of 4.2 GW of batteries in 2035 do not lead to an exact reduction of 4.2 GW in terms of flexibility needs. The reason for this is that when running an economic dispatch model, the batteries react in a European market, their charging and discharging decisions are therefore driven by price and system signals across Europe, not only by conditions in Belgium.

FNA report for Belgium 2026, version approved by CREG

145

However, when calculating the additional flexibility needs shown in Figure 92, the additional batteries are assumed to react only to Belgian system conditions. They only react based on the residual load indicator and curtailment indicator presented in this report. As a result, the additional batteries in the optimisation cannot discharge when there is no remaining positive residual load in Belgium, even if there would still be demand for flexibility elsewhere in Europe.

Figure 92: Total additional flexibility needs in case of 0.0% and 0.1% of RES curtailment allowed

Figure 93 and Figure 94 gives an overview of the amount of flexibility needed to reach different predefined levels of acceptable curtailment. For example, if 0.5% of RES generation (in energy) is allowed to be curtailed, around 5 GW of flexibility would need to be installed in 2030 and around 12 GW in 2035 in the CC-AF25 scenario, which only includes battery capacities secured through CRM auctions up to and including CRM 2024. If the full battery potential identified in Adeqflex’25 is realised, allowing a similar level of RES curtailment would no longer require additional flexibility in 2030 and would reduce the need to around 5 GW in 2035 in the CC-BAT+ scenario. Accepting curtailment also means that less renewable energy is effectively integrated into the system. To compensate for this lost renewable generation and still reach the RES target, additional renewable capacity has to be installed. This is shown on the right-hand side of Figure 93. For example, in the 2030 CC-AF25 scenario, allowing 0.5% RES curtailment would require around 0.2 GW of additional solar PV or around 0.1 GW of additional onshore wind to compensate for the curtailed energy. In the BAT+ scenarios, where the battery potential identified in Adeqflex’25 is realised, the corresponding renewable capacity requirement is significantly lower as a result of the lower curtailment levels.

FNA report for Belgium 2026, version approved by CREG

146

Figure 93: Flexibility needs for different curtailment levels for CC scenarios and additional renewable capacity required to compensate for curtailed energy and achieve the RES target

FNA report for Belgium 2026, version approved by CREG

147

Figure 94: Flexibility needs for different curtailment levels for CT scenarios and additional renewable capacity required to compensate for curtailed energy and achieve the RES target

The analysis highlights that several pathways can be used to reach Belgium’s renewable energy targets. Assuming that renewable energy is almost fully integrated, resulting in a reduction of only 0.1% in renewable energy generation, the following additional flexibility would need to be developed:

• 11 to 21 GW in the CC-AF25 scenario for 2030 and 2035 respectively FNA report for Belgium 2026, version approved by CREG

148

• 11 to 23 GW in the CT-AF25 scenario for 2030 and 2035 respectively • 9 to 18 GW in the CC-BAT+ scenario for 2030 and 2035 respectively • 9 to 17 GW in the CT-BAT+ scenario for 2030 and 2035 respectively Conversely, if RES curtailment is accepted, the curtailed energy can be compensated through relatively small additions of solar or wind capacity in order to achieve the renewable energy targets. For each scenario, the additional renewables capacity required to compensate for the curtailed energy is shown Figure 93 and Figure 94. The additional capacity is assumed to be flexible as well. Note that the values should not be summed, as they represent alternative additions of either solar PV or onshore wind capacity. This amounts to: • 0.4 to 1.3 GW for solar PV and 0.2 to 0.5 GW for onshore wind in the CC-AF25 scenario for 2030 and 2035 respectively; • 0.5 to 1.7 GW and 0.2 to 0.5 GW for onshore wind in the CT-AF25 scenario for 2030 and 2035 respectively; • 0.2 to 0.7 GW and 0.1 to 0.3 GW for onshore wind in the CC-BAT+ scenario for 2030 and 2035 respectively; • 0.2 to 0.6 GW and 0.2 to 0.2 GW for onshore wind in the CT-BAT+ scenario for 2030 and 2035 respectively. However, accepting a certain level of RES curtailment does not automatically mean that the system can safely manage this curtailment. During periods of excess generation, renewable assets must be able to respond and reduce their output when required. If insufficient renewable response is available, the excess generation may remain in the system and create system security issues. Therefore, it is also necessary to assess the additional renewable response required when curtailment is accepted.

8.2.2. Additional RES response required to cover RES curtailment Additional RES response needs are assessed under the assumption that no additional flexibility is developed on top of what is included in the scenarios (including renewable assets reacting to prices). This implies that all RES curtailment shown in Figure 91 occurs. Offshore wind is assumed to be fully responsive, while only a part of onshore wind (60%) and solar PV (5%) is assumed to react based in line with the scenarios described in Section 3.3. This is based on actual observations and can be considered as the lower bound for reactions in the market during extremely low price periods. After accounting for these expected reactions, remaining uncovered needs may still occur during periods of high renewable generation and low demand. The remaining RES response needs represent the additional renewable response that would need to be unlocked to safely manage the accepted curtailment. Figure 95 shows the results for a reliability criterion allowing only one or three hours of uncovered needs per year on average across all Monte Carlo years. The reliability threshold used to identify the additional RES response needs is not defined in the European FNA methodology nor in Belgian legislation and should therefore be interpreted as a reference point rather than a target. While adequacy assessments in Belgium rely on a LOLE standard of 3 hours per year, these are complemented with complementary measures for operational security such as with load shedding plans. As no comparable reliability standards have yet been defined for periods of excess generation, the assessment presents results under both 1‑hour and 3‑hour reliability criteria. The choice of an acceptable reliability level ultimately remains a policy decision. Additional renewable response required to limit uncovered needs to a maximum of 1 hour per year (average across all Monte Carlo years) : • 3.4 to 4.6 GW is needed in the CC-AF25 scenario in 2030 and 2035 respectively • 4.1 to 5.7 GW is needed in the CT-AF25 scenario in 2030 and 2035 respectively FNA report for Belgium 2026, version approved by CREG

149

• 1.7 to 2.9 GW is needed in the CC-BAT+ scenario in 2030 and 2035 respectively • 2.6 to 4.0 GW is needed in the CT-BAT+ scenario in 2030 and 2035 respectively Additional renewable response required to limit uncovered needs to a maximum of 3 hour per year (average across all Monte Carlo years) • 2.3 to 3.6 GW is needed in the CC-AF25 scenario in 2030 and 2035 respectively • 3.2 to 5.0 GW is needed in the CT-AF25 scenario in 2030 and 2035 respectively • 0.6 to 1.9 GW is needed in the CC-BAT+ scenario in 2030 and 2035 respectively • 1.4 to 2.8 GW is needed in the CT-BAT+ scenario in 2030 and 2035 respectively

Figure 95: Uncovered flexibility needs distribution after accounting for all expected flexibility in the scenarios

To quantify this need, the first assessment assumes that uncovered needs should occur for no more than 1 or 3 hours per year on average across all Monte Carlo years. Figure 96 summarises the resulting uncovered needs for the different scenarios under the 1-hour and 3-hour reliability criterion.

FNA report for Belgium 2026, version approved by CREG

150

Figure 96: Summary of uncovered flexibility needs required to limit uncovered needs to a maximum of 1 hour (left) and 3 hours (right) per year, after accounting for all expected flexibility in the scenarios

Since future response of renewable remains uncertain, several sensitivities are performed to assess how these uncovered needs evolve depending on the level of response provided by different renewable technologies. The first sensitivity is performed in which all decentralised onshore wind capacity is assumed to react on top of the response already expected in the scenarios. Figure 97 illustrates this for the 1- and 3-hour reliability criterion, allowing no more than 1 or 3 hours of uncovered needs per year on average across all Monte Carlo years. Assuming full participation of onshore wind and under the 1‑hour reliability criterion, the remaining uncovered needs are reduced to: • 2.7 to 3.5 GW in the CC-AF25 scenario for 2030 and 2035 respectively • 3.6 to 5.2 GW in the CT-AF25 scenario for 2030 and 2035 respectively • 0.8 to 1.7 GW in the CC-BAT+ scenario for 2030 and 2035 respectively • 2.2 to 3.5 GW in the CT-BAT+ scenario for 2030 and 2035 respectively Assuming full participation of onshore wind and under the 3-hour reliability criterion, the remaining uncovered needs are reduced to: • 1.6 GW and 2.7 GW in the CC‑AF25 scenario in 2030 and 2035 respectively; •

2.8 GW and 4.5 GW in the CT‑AF25 scenario in 2030 and 2035 respectively;

•

0.0 GW and 0.6 GW in the CC‑BAT+ scenario in 2030 and 2035 respectively;

•

1.0 GW and 2.1 GW in the CT‑BAT+ scenario in 2030 and 2035 respectively.

Figure 97 shows that, even if all onshore wind reacts, part of the uncovered need remains for the 1hour reliability criterion. This means that onshore wind response alone is not sufficient to cover all remaining needs and that additional renewable response and/or other flexibility resources remain necessary under the assumptions of the assessment. Relaxing the reliability criterion reduces the amount of additional renewable response required. Under the 3‑hour reliability criteria, the remaining uncovered needs are further reduced compared to the 1‑hour case. In some scenarios, particularly the BAT+ scenarios, the additional renewable response required becomes limited and may even disappear entirely. This illustrates that the estimated renewable response needs are highly sensitive to the selected reliability criterion. Since neither the European FNA methodology nor Belgian legislation defines an acceptable level of uncovered renewable response needs, the assessment presents results under both 1-hour and 3-hour reliability criteria. As in adequacy assessments, the selected reliability criterion has FNA report for Belgium 2026, version approved by CREG

151

a significant impact on the identified needs, with a less stringent reliability criterion reducing the amount of additional flexibility or renewable response required to manage periods of excess renewable generation. The results highlight that the remaining uncovered needs are strongly influenced both by the selected reliability criterion and by the assumed battery deployment. While the AF25 scenarios only account for battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios assume the full battery potential identified in Adeqflex’25. Relaxing the reliability criterion (e.g. from 1 hour to 3 hours of uncovered needs per year) and/or assuming higher battery deployment significantly reduces the remaining uncovered needs, and may even eliminate them in some scenarios. Similar to the choice of a LOLE reliability standard in adequacy assessments, these assumptions have an important impact on the estimated renewable response needs and ultimately reflect policy choices.

Figure 97 Uncovered flexibility needs distribution after accounting for all expected flexibility in the scenarios and the additional participation of all decentralised wind (sensitivity)

Figure 98 summarises the resulting uncovered needs for the different scenarios under the 1- and 3hour reliability criterion after accounting for all expected flexibility accounted in the scenarios and the additional participation of all decentralised wind.

FNA report for Belgium 2026, version approved by CREG

152

Figure 98: Summary of uncovered flexibility needs required to limit uncovered needs to a maximum of 1 hour (left) and 3 hours (right) per year accounting for all expected flexibility in the scenarios and the additional participation of all decentralised wind (sensitivity)

Figure 99 combines sensitivities of onshore wind with PV response for 2030. The heatmaps present the remaining uncovered flexibility needs for different combinations of onshore wind response rates (yaxis) and PV response rates (x-axis), while already accounting for offshore wind response and all other expected flexibility sources included in the scenarios. The RES response is expressed in percentage of installed capacity in 2030. The results shown in Figure 99 correspond to a reliability criterion allowing no more than 1 hour of uncovered needs per year on average across all Monte Carlo years. The top part shows CC‑AF25 and CC-BAT+ scenarios, while the bottom part shows CT-AF25 and CT‑BAT+ scenarios in 2030. Two reference points are highlighted: •

the base scenario, assuming 5% PV response and 60% onshore wind response, resulting in a remaining uncovered need of 1.7 – 3.4 GW in the CC-BAT+ and CC-AF25 scenarios respectively (identical to Figure 96); and 2.6 - 4.1 GW in the CT-BAT+ and CT-AF25 scenarios respectively

•

the sensitivity with 5% PV and 100% onshore wind response, reducing the uncovered need to 0.8 - 2.7 GW in the CC-BAT+ and CC-AF25 scenarios (identical to Figure 98); and to 2.2 – 3.6 GW in the CT-BAT+ and CT-AF25 scenarios respectively.

In addition, the sensitivity is highlighted where with approximately 60% of installed onshore wind capacity able to react: •

approximately 15 - 35% of installed PV capacity would need to react to cover uncovered needs in respectively the CC-BAT+ and CC-AF25 scenario;

•

approximately 25 - 40% of installed PV capacity would need to react to cover uncovered needs in respectively the CT-BAT+ and CT-AF25 scenarios

FNA report for Belgium 2026, version approved by CREG

153

Figure 99: Sensitivity on the impact of additional decentralised onshore wind and PV response on uncovered flexibility needs at the 1h level after accounting for all expected flexibility in the scenarios

Note that it approximately refers to the fact that dark green on this figure can still contain up to 400 – 500 MW of uncovered needs. The percentages referred to correspond to installed capacities of onshore wind and PV, while the volumes of uncovered flexibility needs or RES response needs identified in previous sections relate to the effective reaction. Note that larger non-residential installations are expected to amount up to 5 to 6 GW in respectively 2030 and 2035. Note that when comparing with installed solar installations, a derating factor should be applied accounting that only 60 – 70 % of capacity is effectively generating during production. The capacity to preserve the residential segment will depend on the speed of decision to make new existing units steerable, the development of the market and the capacity to retrofit existing units. As it is already explained that the additional RES response needs are strongly influenced by the selected reliability criterion, the results shown in Figure 100 correspond to a reliability criterion allowing no more than 3 hours of uncovered needs per year on average across all Monte Carlo years. The top part shows CC‑AF25 and CC-BAT+ scenarios, while the bottom part shows CT-AF25 and CT‑BAT+ scenarios in 2030. Two reference points are highlighted: • the base scenario, assuming 5% PV response and 60% onshore wind response, resulting in a remaining uncovered need of 0.6 – 2.3 GW in the CC-BAT+ and CC-AF25 scenarios respectively (identical to Figure 96); and 1.4 - 3.2 GW in the CT-BAT+ and CT-AF25 scenarios respectively • the sensitivity with 5% PV and 100% onshore wind response, reducing the uncovered need to 0.0 - 1.6 GW in the CC-BAT+ and CC-AF25 scenarios (identical to Figure 98); and to 1.0 – 2.8 GW in the CT-BAT+ and CT-AF25 scenarios respectively.

FNA report for Belgium 2026, version approved by CREG

154

In addition, the sensitivity is highlighted where with approximately 60% of installed onshore wind capacity able to react: • approximately 5 - 20% of installed PV capacity would need to react to cover uncovered needs in respectively the CC-BAT+ and CC-AF25 scenarios; • approximately 15 - 30% of installed PV capacity would need to react to cover uncovered needs in in respectively the CT-BAT+ and CT-AF25 scenarios The results in this section indicate that the reduction of the remaining uncovered needs is not driven by a single technology, but rather by a combination of (1) renewable response and/or (2) additional non‑fossil flexibility (such as the assumed battery deployment in the BAT+ scenarios). They further show that the magnitude of this reduction is strongly influenced by (3) the selected reliability criterion. While the AF25 scenarios only account for battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios assume the full battery potential identified in Adeqflex’25. Increasing the response of onshore wind and PV generation, relaxing the reliability criterion, and/or assuming higher battery deployment all contribute to reducing the remaining uncovered needs and, in some cases, eliminating them entirely. These assumptions therefore have a significant impact on the estimated renewable response needs and ultimately reflect policy choices.

Figure 100: Sensitivity on the impact of additional decentralised onshore wind and PV response on uncovered flexibility needs at the 3h level after accounting for all expected flexibility in the scenarios

8.2.3. Sensitivity on nuclear and end-user flexibility The results presented above are based on assumptions regarding the future generation mix and the availability of flexibility in the system. Since both remain uncertain, additional sensitivities are assessed to understand how changes in these assumptions affect the remaining uncovered needs after FNA report for Belgium 2026, version approved by CREG

155

accounting for expected renewable response and flexibility resources. In particular, this section considers the impact of higher nuclear capacity and lower end-user flexibility. Figure 104 shows the results of these sensitivities compared to the main scenarios. It assumes full reaction of wind power and a reaction of 5% of the installed PV fleet.

Figure 101: Remaining uncovered renewable response needs for the different scenarios and sensitivities at the 1h level with a PV response of 0% of the installed capacity

The nuclear sensitivity shows that additional nuclear capacity increases the remaining uncovered needs. Adding 1 GW of nuclear generation lowers the residual load across almost all hours. As a result, during periods with high renewable generation and low demand, the system reaches excess generation conditions more often and with larger volumes, increasing the need for renewable response and/or additional flexibility. As nuclear generation structurally shifts the residual load downward, the additional excess generation is not limited to a few isolated hours, but affects the entire residual load profile. The sensitivity indicates that adding 1 GW of nuclear capacity increases the remaining uncovered needs by at least around 1 GW. This should however be interpreted as a lower bound, as it assumes that market signals lead to an efficient battery reaction. In particular, prices need to correctly reflect the timing and magnitude of the residual load peaks, including sufficiently negative prices and an appropriate negative merit-order effect. If these price signals do not adequately reflect periods of highest curtailment risk, batteries may start charging too early and become fully charged before the actual peak excess occurs. In that case, batteries would no longer be available when they are most needed, potentially increasing the remaining uncovered needs by more than the additional nuclear capacity itself. A second sensitivity concerns end-user flexibility. If less end-user flexibility is available, the remaining uncovered needs also increase. End-user flexibility from electric vehicles, heat pumps and small-scale batteries can absorb part of the renewable surplus by shifting demand towards periods with high renewable generation. If this flexibility develops more slowly, or if fewer end-users react to market or local price signals, less surplus generation can be absorbed by demand. Consequently, more excess generation remains in the system and must be managed through additional renewable response and/or non‑fossil flexibility.

FNA report for Belgium 2026, version approved by CREG

156

8.2.4. Qualitative assessment of the economically efficient balance The present FNA adopts a technology-neutral approach and therefore quantifies the implications associated with different boundary cases rather than identifying an economically optimal outcome. In the absence of an explicit RES curtailment target in Belgium, future editions of the FNA could be complemented by a quantitative social welfare assessment aimed at identifying the economically efficient balance between additional flexibility deployment, acceptable RES curtailment and the associated additional renewable generation capacity. Such an assessment could help identify the economically justified level of flexibility and curtailment and provide additional input for the definition of indicative flexibility objectives. The present FNA nevertheless provides a qualitative assessment of where the economically efficient balance may lie between additional flexibility deployment, acceptable RES curtailment and the associated additional renewable generation capacity. This first FNA does not include a quantitative social welfare assessment comparing the costs and benefits of developing additional non-fossil flexibility with those associated with accepting a certain level of RES curtailment and deploying additional renewable generation capacity to compensate for the curtailed energy. It therefore does not identify the economically optimal balance between these options. Nevertheless, the quantitative results provide useful indications of where this balance may lie. In the BAT+ scenarios, which incorporate the battery potential identified in AdeqFlex'25, RES curtailment remains limited to approximately 0.4–0.5% of annual wind and solar generation in 2030 and around 1.1% in 2035, without additional non-fossil flexibility beyond the resources already included in those scenarios. These results suggest that the economically efficient outcome is likely to be close to a situation in which no significant additional non-fossil flexibility needs to be developed beyond the battery potential identified in AdeqFlex'25. However, this qualitative conclusion should be confirmed through a quantitative social welfare assessment in a future FNA, taking into account the costs of additional flexibility, RES curtailment and the additional renewable generation capacity required to compensate for curtailed energy. Within the scope and assumptions of this first FNA, the results do not indicate a need for a dedicated flexibility support mechanism in addition to existing market arrangements and adequacy mechanisms, including the CRM. Priority should instead be given to ensuring that the battery potential identified in AdeqFlex'25 can effectively materialise, notably by preventing grid connection procedures and constraints from becoming barriers to the development of economically viable battery projects. Measures aimed at strengthening renewable participation in response to market signals should also be pursued. Their effectiveness will be important in determining the residual flexibility needs and the level of curtailment that can be safely managed. Demand-side response may provide an additional contribution and should be considered alongside storage and renewable response as a source of nonfossil flexibility.

FNA report for Belgium 2026, version approved by CREG

157

9. Barriers for flexibility and contribution of digitalisation Unlocking large-scale flexibility requires coordinated efforts across the entire energy ecosystem. Since earlier publications on Belgian flexibility, several general key enablers have progressed and continue to evolve. Further progress on the key enablers will depend on stronger collaboration among all stakeholders in the areas summarized in this section. For the purpose of evaluating the barriers for flexibility, ACER identifies six categories which are discussed accordingly in this section: 1. Lack of proper legal framework for market access to new entrants and small actors 2. Lack of enablers and incentives to provide flexibility 3. Restrictive requirements to provide balancing services 4. Restrictive requirements to provide congestion management 5. Complex, lengthy, and discriminatory administrative requirements 6. Lack of regulatory incentives to system operators to consider non-wire alternatives This list of barriers is not exhaustive and ACER encourages Member States to identify and assess any additional barriers they consider relevant to the development of non-fossil flexibility. Figure 102 provides a high level overview of the barriers identified by ACER, and potential enablers, complemented with an additional barrier identified by Elia on infrastructure and interoperability.

Figure 102: Enablers for the unlocking flexibility

FNA report for Belgium 2026, version approved by CREG

158

Barrier 1 : Lack of proper legal framework for market access to new entrants and small actors INDICATOR 1.1 MAIN ROLES AND RESPONSIBILITIES OF NEW ACTORS NOT DEFINED All necessary primary and secondary legislation (except for low voltage levels where valorisation of explicit flexibility on the DA/ID markets through a flexibility service provider is not yet allowed) is in place to ensure that final customers are entitled to act as active customers and exercise the following rights: a. Consume and store electricity generated within their premises located within confined boundaries or self-generated or shared electricity within other premises, b. Operate directly or through aggregation, c.

Sell self-generated electricity, including through power purchase agreements,

d. When they own an energy storage facility, they have the right to a grid connection within a reasonable time after the request, provided that all necessary conditions are fulfilled. Provider (FSP) that can be performed by either an aggregator or the end consumer, meaning there is no need for a separate definition of the aggregator role in legislation. The transfer of Energy Framework enables each final customer to valorise their flexibility via the FSP of their choice without the consent of their energy supplier. The model provides for financial compensation between the supplier and the FSP based on the total volumes transferred. In this way, final consumers are not subject to undue payments or other undue contractual restrictions. Energy storage facilities on TSO-level are permitted to be combined with facilities producing renewable energy, electric vehicle recharging points, final customer demand facilities connected to the same grid access point. In case more than one balance responsible party is active on or behind an Access point, a perimeter correction is foreseen to ensure : • All BRPs have access to the Day-Ahead market /or Intra-day market. • aFRR / mFRR can be offered by BSPs for each Delivery Point. In the particular case of FCR, there is no perimeter correction, as it's considered a capacity product (average volume zero). FCR can be offered by BSPs for each point with a BRP • Redispatching is offered by Scheduling Agents for each point with a BRP. Regarding activation for congestion management purposes in the framework of a flexible connection agreement, the framework is being adapted to include a BRP perimeter correction. INDICATOR 1.2 MARKET ACCESS RESTRICTED DUE TO LACK OF LEGAL ELIGIBILITY All listed market participants can participate in all markets across DA, ID, FCR, aFRR, mFRR and TSO capacity markets. Restrictions apply only on which assets can participate. Low voltage participation to DA/ID is not yet allowed. Elia is currently in the process of updating the ToE DA/ID framework, in order to better facilitate independent aggregation for participation to Day-Ahead and Intraday markets. Whereas the current framework focuses solely on industrial demand response, the redesign will ensure all relevant asset types can valorise their flexibility independently from the Supplier, via an FSP, on DA and ID markets. This is done notably via the introduction of additional baselines, combinability of participation in DA/ID markets and aFRR during the same quarter-hour and allowing the FSP a more complete optimization on the assets. This new design is currently estimated to go-live end of 2027.

FNA report for Belgium 2026, version approved by CREG

159

What concerns 'direct' participation, note that participation is always only possible through a BRP, BSP, or FSP intermediary, or if the actor assumes this role himself. INDICATOR 1.3 OWNERSHIP RESTRICTIONS FOR ENERGY STORAGE AND ELECTRO -MOBILITY The Electricity Directive prohibits TSOs from owning, developing, managing or operating energy storage facilities. There are no derogations in Belgium on TSO-level. INDICATOR 1.4 RESTRICTIONS ON TRADE IN DAY -AHEAD AND INTRADAY MARKETS DA/ID markets have a minimum bids size of 0.1 MW, and a time interval of 15 minutes, which is aligned with the Imbalance Settlement Period. Belgium operates under a Multiple NEMO Arrangement (MNA) for its electricity market. NEMOs already enable market participants since the go live of 15 minute time intervals in SDAC and SIDC to trade in energy in time intervals which are at least as short as the imbalance settlement period for the DA and ID markets

Barrier 2 : Lack of enablers and incentives to provide flexibility It is required to empower end users by continuing to enhance end-user behaviour, without adding complexity as this is essential to keep it convenient and comprehensible for end consumers. For a majority of residential end users comfort, stability and low energy cost are the most important concerns. INDICATOR 2.1: LACK OF ADEQUATE METERING SYSTEMS The roll-out of smart metering is still ongoing in Belgium. Figure 103 (left), shows that at the end of 2025, total EANS under smart meter amounts up to 3.7 million (representing 60%). Beyond smart meters, submeters are also used for a wide variety of TSO services and products, including DA/ID as well as FCR, aFRR and mFRR. Figure 103 (right) shows the number of dynamic contracts in Flanders and since recently also in Wallonia and number of smart meters with unlocked ‘smart’ capabilities. It is shown that the number of households with SMR38 enabled meters and dynamic contracts is growing, but very slowly. Adeqflex’25 already indicated this slower-than-expected growth which led to a revision of the scenarios on uptake of end user flexibility.

SMR3 enables the grid operator to automatically read the consumption data on the smart meter with a quarterly hourly granularity and send it to the grid user's energy supplier

FNA report for Belgium 2026, version approved by CREG

160

Figure 103: Indicators on smart meters installed and Dynamic Contracts / SMR3 (based on information available information through regulators of distribution system operators).

On the other hand, there are positive indications that market parties have started offering simple, transparent and innovative contract models that activate end user flexibility and could increase general awareness and knowledge around flexibility. These ‘contracts’ referring to Happy hours or Sunny Sundays contracts focus on increasing offtake, or reducing injections, during negative price periods. Yet, limited information is available on the numbers of consumers under such contract and the impact on the electricity system. At the same time, there is still a large untapped flexibility potential in electric vehicles and heat pumps which is currently not unlocked or whose dispatch is not yet driven by market signals. Figure 103 (right), shows that dynamic contracts are growing very slowly in number and amount up to 32000 EANs (around 1% of installed digital meters). More consumers activate their SMR3 (190000 EANs, around 6% of installed digital meters) which is nevertheless also a possible indicator for consumers’ willingness to engage in flexibility. As only a small fraction is driven by market prices, the dispatch of these resources may not be fully aligned with the flexibility needs at system level. INDICATOR 2.2: ABSENCE OF PRICE SIGNALS On system level, it is needed to continue to provide the right price incentives. A key feature of this facilitation is, next to other relevant price signals (such as day-ahead and intraday markets) the RealTime Price, which is a price forecast and an evolution of the imbalance price. These price signals should enable suppliers to offer new types of energy contracts to consumers, such as dynamic or fixed timeof-use contracts, which we see emerging in the market. Innovative offers should also be supported by incentive-based non-static grid tariffs that promote flexibility and encourage optimal energy consumption at times when it’s most beneficial for both consumers and the grid. A consistent price signal between energy and grid tariff elements will make flexibility more attractive and should also provide incentives to relieve the network, yet cost-reflective tariffs are equally important. A capacity-based DSO grid tariff in Flanders and a Time-of-use DSO grid tariff in Wallonia were recently introduced. A time-of-use DSO grid tariff is currently in development in Brussels. At TSO level, the new tariff methodology will be in force from 1st January 2028. The new methodology:

FNA report for Belgium 2026, version approved by CREG

161

• Foresees the extension of pricing based on available capacity to all users directly connected to the transmission network, is intended to encourage network users to smooth out their consumption and feed-in profiles and introduces a reduction factor for voluntary flexible connection capacities. • Phases-out the exemption from transmission tariff granted to storage to encourage developers of electricity storage projects to opt for voluntary flexible connections. • Foresees the application of the tariff for the annual peak only to the annual peak net power drawn from the grid during the peak tariff period (currently November to March, on weekdays excluding weekends, from 5 pm to 8 pm, but subject to change based on updated analysis of the grid load) in order to introduce incentives for efficient network use. DSOs identify the absence of time-based capacity tariffs for low- and medium-voltage grids as a barrier. The introduction of such tariffs, whereby peaks during off-peak hours are discounted by a certain factor (as in Flanders), could provide stronger incentives for flexibility.

Barrier 3 : Restrictive requirements to provide balancing services Next to supporting market parties in their (innovative) offerings, it is also important to enable grid users to access flexibility services, further facilitating growth of decentral flexibility. INDICATOR 3.1: NON-MARKET -BASED BALANCING PRODUCTS Procurement of these services is market-based and, aside from FRR capacity auctions which are settled “pay-as-bid”, completely in line with the EU target model in terms of volumes, validity periods, asymmetric products, settlement rules and non-contracted energy bids, procurement timeline and capacity contract periods. INDICATOR 3.2 : RESTRICTIONS IN THE PREQUALIFICATION OF RESERVE PROVIDING GROUPS (RPG S) To allow delivery of these services independently of the BRP and supplier: • Industrial clients connected to the Transmission Grid are able to benefit from the Multiple BRP service, which allows the grid user to assign a BRP to a submeter different from the BRP of the Access Point (Head Meter); currently implemented via a Letter of Intent. As soon as the regional legislation is in place, Elia will take the necessary step to further roll-out the Multiple BRP service to the local transport grids as well. • In parallel, the DSOs are facilitating the data exchange of the supply split enabler. • The Transfer of Energy (ToE) framework enables grid users to leverage their flexibility through a Flexibility Service Provider (FSP), independently of their energy supplier. ToE is open for all voltage levels, for aFRR and mFRR; as well as ToE DA/ID for high and medium voltage. Elia and the DSO are further investigating improvements to the ToE framework, notably by studying improvements to the ToE DA/ID product (cf. indicator 1.2) . INDICATOR 3.3 : LARGE MINIMUM ELIGIBLE CAPACITY The minimum capacity (MW) required by each TSO in the prequalification process for the different local or standard and specific balancing products used for FCR, aFRR and mFRR is 1 MW while aggregation is allowed. INDICATOR 3.5 – 3.9 UNREGULATED DURATION OR LONG PREQUALIFICATION PROCESS To further facilitate prequalification and delivery, possible actions considered are FNA report for Belgium 2026, version approved by CREG

162

• Transitioning from locally implemented gateway communication requirements towards more scalable, centrally coordinated or platform-based communication frameworks. • Streamlining onboarding and prequalification procedures with a view to reducing administrative burden, implementation costs, and overall process complexity. This includes the omission of prequalification tests to add new Delivery Points to the active pool of a BSP and the omission of re-certification following the change of baselining methodology. • Assessing potential approaches for BSPs/FSPs to efficiently manage and deliver multiple flexibility products that address distinct system requirements across different voltage levels. INDICATOR 3.5 – 3.9 : DEVIATIONS FROM EU TARGET MODEL No specific products not in line with EU target model requirements exist in Belgium (for FCR, aFRR capacity, aFRR energy, mFRR capacity, mFRR energy). Only deviation from the EU target model concerns price settlement where aFRR and mFRR capacity auctions use pay-as-bid where the target model is marginal pricing. FCR and aFRR and mFRR energy auctions use pay-as-clear. INDICATOR 3.10 : TOO SHORT PREDEFINED MINIMUM DURATION BETWEEN CONSECUTIVE ACTIVATIONS In terms of the shortest and the longest minimum duration (minutes) between end of deactivation and following activation for local or standard and specific balancing energy products for aFRR and mFRR. Activations can be consecutive without a minimum duration between end of deactivation and start of activation INDICATOR 3.11 : LONG PROCUREMENT LEAD TIME AND LONG DURATION OF BALANCING CAPACITY CONTRACTS

No regulatory derogations exist on time between contracting and provision of balancing capacity, on contractual period of capacity contracts or the contractual period of balancing capacity contracted more than one day-ahead. All balancing capacity is contracted within one day before provision, for all balancing reserves and products.

Barrier 4 : Restrictive requirements to provide TSO congestion management INDICATOR 4.1 : UNJUSTIFIED LACK OF MARKET -BASED TSO CONGESTION MANAGEMENT The main congestion management measures taken by TSOs to address physical congestions is redispatching. Re-dispatching is implemented as a non-market based service. There are several reasons for this: • Re-dispatching is mainly used for high voltage grid congestion issues currently. Currently this service is offered by a limited number of service providers. The high voltage grid is subdivided in multiple electrical zones (zones where the copper plate hypothesis holds, congestion occurs between zones). In multiple electrical zones only one market party offers redispatching, while in every zone the maximal number of unique parties offering redispatching is equal to 4, meaning the Herfindahl-Hirschman Index (HHI) is always ≥0.25 with the best case assumption, indicating a concentrated market in accordance with guidelines from the European commission. • When a congestion risk is identified (this is a rare occurrence currently in Belgium) it is local and predictable, which would lead to strategic bidding under a market‑based scheme as it is most often related to specific planned grid outages impacting only a small area, with limited market participants able to impact the congestion. Even when multiple market participants could impact the congestion, the level of predictability is sufficiently high that it would lead to strategic bidding. • In combination of freedom of dispatch, where market parties are able to trade in intraday without constraints until 45 min before RT (until 30 min before RT from April 2027 on), even if a congestion risk has been identified, transparency obligations (CRI, outage publication), and FNA report for Belgium 2026, version approved by CREG

163

highly predictable local congestion would structurally enable gaming, undermining market integrity under a market‑based redispatch design. Currently non-market based re-dispatching is obligatory only to technical facilities (production and storage) ≥ 25 MW. Demand and smaller units do not have an obligation to participate. Activation is always after the day-ahead market, and during intraday market timeframe, but typically as late as possible (after ID GCT). Since the level of competition and predictability is applicable to all electrical zones, there is no distinction between grid areas currently. A re-assessment of the reasons for non-market dispatch will be realised, accounting where relevant for results of TSO network needs assessments in the next version of the FNA. In parallel Elia is investigating alternative remuneration schemes. If market‑based redispatching (or a hybrid/variant) becomes feasible, an implementation plan will be prepared and submitted to regulators. In parallel Elia is investigating alternative remuneration schemes. If market‑based redispatching (or a hybrid/variant) becomes feasible, an implementation plan will be prepared and submitted to regulators Initiation is either regulatory‑driven or triggered by changes in market or system conditions identified by Elia. The final approval remains with the national regulatory authorities. Adaptation is needed to both federal and regional grid codes, to be developed and enacted by the competent regulatory authorities

Barrier 5 : Complex, lengthy, and discriminatory administrative requirements INDICATOR 5.1 INEFFICIENT GRID CONNECTION PROCEDURES On TSO level, Elia publishes online up-to-date information on available grid capacity for connection. This is done via the publication of a monthly updated Grid Hosting Capacity map, taking into account the existing and planned infrastructure as well as the existing, reserved and allocated capacity. Note that hosting capacity is pre-reserved for low voltage Electrical Vehicle, Heat Pump and other low voltage demand to prevent their development from being blocked by other sectors. Next to these hosting capacity limitations, Elia observes a steep increase in grid connection requests to the TSO network (for flexible resources, as well as demand and generation). The First-In First-Out principle for dealing with such requests also increased the requests for less mature projects which are submitted to obtain a better place in the connection queue. This dynamic has increased the throughput time to obtain a connection permit (i.e. the result of a detailed grid connection study). In search of the best remediation, Elia is looking into solutions for disincentivizing immature connection requests, increasing the processing capacity to handle requests and distributing the limited grid capacity to generation, storage and consumption in line with national and regional targets (through stakeholder dialogue in a dedicated Task Force Local Redistribution of Injections and Offtakes). INDICATOR 5.2 DISPROPORTIONATE PROCEDURES AND REQUIREMENTS FOR MARKET ACCESS As regards procedures and requirements to access wholesale electricity markets and local markets for congestion management, there are currently no local markets for congestion management at TSO level, in the sense that participation to congestion management products are mandatory and remunerated at a cost-based price. Elia is continuously working on facilitating access to small BRPs, a.o. via supply split, multiple BRP, bank guarantees reductions, etc. In addition, the transfer of energy framework is being extended across balancing products and different voltage levels (i.e. up to the residential level). Elia is currently conducting a study to improve products facilitating entry into the DA/ID markets for aggregators.

Barrier 6 : Lack of regulatory incentives to system operators to consider non-wire alternatives INDICATOR 6.1 LACK OF INCENTIVES FOR SOS TO CONSIDER NON -WIRE ALTERNATIVES FNA report for Belgium 2026, version approved by CREG

164

On TSO level, the TOTEX methodology is under development. In addition, a robust regulatory framework is needed to apply the use of flexibility for managing grid constraints. On DSO level, solving congestion with flexibility faces specific challenges requiring a specific framework: • DSOs have limited real-time monitoring over their low-voltage networks making it difficult to observe, forecast and activate the DSO Network Needs in real-time. • Grid user behavior can be subject to high uncertainty, making it difficult to accurately forecast congestion, in distribution as well as transmission grids. • Network needs can often only be solved by a limited number of flexible assets. Costs and availability of explicit flexibility may therefore be hard to assess upfront when trading off flexibility against grid reinforcement. • Collaboration between TSO and DSO should also be further specified, especially when solving all type of congestion by activating flexible resources connected to DSO grids. • lack of observability of local constraints towards market parties is also a barrier for investing in, aggregating and operating distributed flexibility. Note that DSOs are working on increasing transparency on local needs9. • Multiplicity of DSO-specific market platforms for congestion management may impose an additional barrier for developing flexibility.

9.1. Digitalisation The FNA method requests to assess in addition to barriers the contribution of digitalisation of electricity transmission and distribution networks in flexibility. Two main priorities are identified. The first priority for digitalisation is to deploy enabling infrastructure, by ensuring the full deployment of smart meters across all regions in Belgium. This is the most important enabler for digitalisation. All DSOs have started the roll-out, but adoption rate vary between the regions. In 2025, over 3 million smart meters were already rolled out in Belgium. Consumers with the highest flexibility potential, such as those with electric vehicles, solar panels, or home batteries should be prioritised to facilitate a fast unlocking of the residential flexibility. Care should be taken that this does not compromise the rollout efficiency. For most flexibility services, private submeters are already allowed when the measurement device meets the technical requirements. The second priority is to standardise and ensure the interoperability of flexible assets by making these assets ‘flex-ready’ so that they can contribute and deliver flexibility to the system and to avoid lock-in and ensure interoperability across all vendors.

E.g. “Waar zoeken we marktflexibiliteit?”, Fluvius, marktflexibiliteit

https://partner.fluvius.be/nl/flexibility-service-provider/waar-zoeken-we-

FNA report for Belgium 2026, version approved by CREG

165

10. Conclusions 10.1. Key findings The first Belgian Flexibility Needs Assessment confirms that managing excess generation during periods of high renewable generation will become the main flexibility challenge. This will complement the traditional focus on scarcity situations in adequacy. The assessment assumes that renewable generation capacity is expected to roughly double by 2035, while the available flexibility potential from batteries and end-users such as demand response, electric vehicles and heat pumps is assumed to triple, or even quadruple, at least if properly unlocked. The actual deployment of both flexibility resources is uncertain and is therefore assessed through different scenarios. The analysis indicates that accepting relatively low levels of renewable energy curtailment is sufficient to significantly reduce the additional flexible resource required for RES integration. Across the assessed scenarios, curtailment levels between approximately 0.5% and 2.5% of annual wind and solar generation are associated with substantial reductions in the identified flexibility needs. Most curtailment occurs during spring and summer periods with high renewable output and relatively low demand, typically during hours characterised by high solar generation. In the absence of specific targets on RES curtailment in Belgium, the assessment explores the trade‑off between increasing flexibility to integrate renewable generation and accepting limited levels of renewable curtailment. The results show that: • achieving near‑complete RES integration would require significant amounts of additional non‑fossil flexibility, ranging from around 9 to 11 GW in 2030 and from 18 to 23 GW in 2035 depending on the scenario; • accepting limited levels of renewable curtailment can substantially reduce these flexibility requirements, while the curtailed energy can be compensated by relatively modest additions of onshore wind or solar PV capacity: 0.2 to 0.5 GW of onshore wind or 0.6 to 1.7 GW of solar capacity by 2035. The present FNA does not include a quantitative social welfare assessment comparing the costs and benefits of developing additional flexibility, accepting RES curtailment and deploying additional renewable generation capacity. Nevertheless, the results provide a qualitative indication of where such an economically efficient balance may lie. In particular, the BAT+ scenarios, which incorporate the battery potential identified in Adeqflex'25, already result in relatively limited RES curtailment levels (around 0.4–0.5% of annual wind and solar generation in 2030 and around 1.1% in 2035). This suggests that the economically efficient outcome may be close to a situation in which the battery potential identified in Adeqflex'25 is effectively realised, while accepting limited levels of RES curtailment and compensating curtailed energy through relatively modest additions of renewable generation capacity. At the same time, the system security assessment confirms that both upward and downward shortterm flexibility needs increase over time as the system becomes more driven by weather variations and forecast errors. While upward needs can be managed via the balancing market and reserve procurements, downward needs demonstrate unresolved excess energy-generation persists linked to PV. The assessment further shows that the network needs analysed in this first edition of the FNA — namely DSO/TSO coupling point constraints and limitations affecting flexible resources — have a relatively limited impact on the overall results. While local bottlenecks can occur and should continue to be monitored, the quantified impact on annual renewable curtailment remains limited compared to the broader system-level flexibility needs. The key finding of the FNA is that sufficient renewable response and non‑fossil flexibility resources go hand in hand and will become increasingly important to safely manage periods of excess generation. The assessment shows that the amount of additional flexibility resources needed

FNA report for Belgium 2026, version approved by CREG

166

and renewable response required depends strongly on the level of battery deployment assumed, and the effective participation of renewable resources. While the AF25 scenarios only account for battery capacities secured through CRM auctions up to and including CRM 2024, the BAT+ scenarios assume the full battery potential identified in Adeqflex’25 (based on feasibility studies and connection requests). The results show that higher battery deployment can significantly reduce the remaining uncovered needs associated with excess renewable generation. Under the 1‑hour reliability criterion, justified in absence of symmetric emergency measures such for shortages, between approximately 0.8 and 3.6 GW of remaining uncovered needs may persist in 2030 depending on the scenario. This is after even assuming full participation of onshore wind. The results also show that the remaining uncovered needs depend on a combination of renewable response and/or additional non‑fossil flexibility. The magnitude of the uncovered flexibility needs or remaining renewable response needs is strongly influenced by the assumed battery deployment and reaction of solar on market prices. Any additional reaction of solar in the market on top of the 5% of total installed capacity assumed to be flexible in the scenarios will proportionally reduce the uncovered needs. It should be noted that non-residential PV installations are expected to amount to approximately 5 GW in 2030 and 6 GW in 2035. When comparing the identified needs with installed solar capacity, a derating factor should be applied, as only around 60–70% of installed PV capacity is typically generating during periods of high solar production. The residual need to be covered by the residential segment would be zero if sufficient participation of larger PV installations in the market is ensured and residential PV installations increasingly respond to market price signals. The assessment also demonstrates that the estimated needs rely strongly on the selected reliability criterion. As no reliability criterion for periods of excess generation is defined in either the European FNA methodology or Belgian legislation, both criteria should be interpreted as reference points rather than targets. The 1‑hour criterion is aligned with system security needs and is justified due to the lack of measures to manage excess generation (compared with load shedding plans used to manage shortages). The 3‑hour criterion is aligned with the reliability standard applied in Belgium's adequacy framework although no such support mechanisms exist for downward flexibility today. Consequently, the renewable response needs identified under the 1‑hour criterion should be regarded as cautious, while the 3‑hour criterion results in significantly lower requirements. These thresholds can be used in further discussions on solutions to attain acceptable reliability levels while the choice of an acceptable reliability level ultimately remains a policy decision. Sensitivities on nuclear extensions and lower end‑user flexibility further illustrate that the identified needs depend strongly on assumptions regarding future generation and flexibility deployment. Higher nuclear generation and lower end-user flexibility participation increase the amount of excess generation that needs to be managed and can therefore lead to higher renewable response and flexibility needs. Overall, the FNA concludes that Belgium’s main challenge is not the volume of renewable energy to be integrated but ensuring that sufficient renewable response and non‑fossil flexibility are available to safely and efficiently manage periods of excess generation. The results indicate that this objective can largely be supported through the effective realisation of the battery potential identified in AdeqFlex’25, further strengthening renewable participation in response to market signals and unlocking additional demand-side flexibility. Within the scope and assumptions of this first FNA, the results do not indicate a need for a dedicated flexibility support mechanism in addition to existing market arrangements and adequacy mechanisms, including the CRM. Continued development of market-based flexibility and the deployment of additional flexibility resources, where economically justified, may further contribute to reducing the remaining needs.

10.2. Next steps Policy makers are required to identify non-fossil flexibility targets. The results of this assessment allow a trade-off to be made between such non-fossil fuel flexibility targets and acceptable levels of RES curtailment including renewable response requirements. The needs identified in this report vary significantly depending on assumptions regarding battery deployment, renewable response and the selected reliability criterion. The results should therefore FNA report for Belgium 2026, version approved by CREG

167

support policy discussions on the appropriate balance between non‑fossil flexibility deployment, acceptable levels of RES curtailment and the resulting renewable response requirements, in order to ensure system security while achieving renewable energy objectives. Belgium currently has no explicit target regarding acceptable levels of RES curtailment. For this reason, the assessment evaluates a range of curtailment levels and quantifies the associated flexibility and renewable response needs. The analysis highlights that different pathways can be followed to achieve renewable energy objectives. The required level of non‑fossil flexibility depends on the acceptable level of RES curtailment and the expected deployment of battery storage, but also the expected levels of end user flexibility and the selected reliability criterion. In this context, the FNA provides a framework to support policy decisions regarding the balance between deploying additional non‑fossil flexibility, accepting different levels of RES curtailment and the resulting (additional) renewable response requirements. These choices fall within the scope of public policy. TSOs and DSOs should consider where useful the findings of the FNA in future network development plans. In this first edition of the FNA, no direct inputs are generated for network development plans. At distribution level, RES curtailment related to TSO/DSO coupling point constraints is found to be limited and is often already identified through existing planning processes. At transmission level, a more detailed assessment of network-related flexibility needs is foreseen in future editions of the FNA and may provide inputs for future network development plans. The analyses carried out in this first edition indicate that the impact of network constraints on the identified flexibility needs remains limited compared to the broader system-level flexibility challenge. In particular, battery offtake constraints during periods of high renewable generation were found to have a relatively limited impact on the overall results. This first FNA represents an important step towards a comprehensive assessment of flexibility needs in Belgium. Several areas for further methodological development have already been identified and could be addressed in future editions of the FNA. This will be investigated and confirmed first in the implementation plan for the FNA 2028, notably: •

•

FNA report for Belgium 2026, version approved by CREG

168

•

Social welfare assessment: assess the relevance of introducing a quantitative social welfare assessment aimed at identifying the economically efficient balance between additional flexibility deployment, acceptable RES curtailment, renewable response and the associated additional renewable generation capacity. Such an assessment could provide additional support for the definition of indicative flexibility objectives and curtailment targets.

FNA report for Belgium 2026, version approved by CREG

169

11. References Adeqflex’25 https://www.elia.be/nl/elektriciteitsmarkt-en-systeem/adequacy/adequacystudies Assumptions and methods for the Belgian FNA approved by CREG: https://www.creg.be/nl/openbare-raadplegingen/prd3184 European Legal framework: Regulation (EU) 2019/943: https://eur-lex.europa.eu/eli/reg/2019/943/oj/eng European Regulatory framework: FNA method approved by ACER: https://www.acer.europa.eu/news/acer-approves-eu-wide-methodology-assess-national-electricityflexibility-needs Belgian Implementation plan approved by CREG: https://www.creg.be/nl/publicaties/beslissing-b3152

FNA report for Belgium 2026, version approved by CREG

170

Glossary Adequacy

Ensuring sufficient generation capacity is available to meet peak demand

Adequacy gap

Shortfall between available generation and required demand to meet the LOLE criterium.

Downward Flexibility

Flexibility linked to decreasing injection into the network or increasing demand from the network;

Economic Dispatch (Simulation)

An interconnected power system simulation of generation and storage dispatch by minimizing costs subject to constraints.

Fast flexibility

Flexible capacity able to react to aggregated D-1 or shorter error variations (15 to 60 minutes before real-time);

Flexibility

Ability of a power system to adjust generation or consumption to variability and uncertainty

Gap filling

Addition of capacity to cover adequacy gap

GUFlex

Agreement allowing grid operator to limit grid connection capacity usage under constraints.

Incompressibility gap

Additional flexibility required to manage excess generation

Incompressibility Type A

Needs under perfect forecast conditions (predictable excess generation).

Incompressibility Type B

Needs driven by forecast errors or unexpected events.

Monte Carlo year (mcyear)

Simulated meteorological year used for probabilistic analysis

Network (flexibility) needs

Flexibility needed to adjust for grid availability, by means of preventing or solving congestion or voltage issues, across relevant timeframe;

Ramping (flexibility) needs

Quantity of flexibility needs to handle variations of the residual load assuming perfect forecast conditions

RES Integration Needs

Quantity of flexibility required for the Member State to achieve its annual RES integration target;

Short-Term (flexibility) needs

Quantity of flexibility required to handle unexpected variations of the residual load such as forecast errors or forced outage of assets during the intra-day or balancing timeframe; Flexible capacity able to react to aggregated D-1 or shorter horizons forecast error (60 minutes to 24 hours before real-time);

Slow flexibility Summer Outlook

Annual report by Elia assessing short-term system risks for the upcoming spring and summer.

System Needs

Quantity of flexibility needed by the electricity system to adjust to the variability of generation and consumption patterns, across relevant market timeframe;

Uncovered (flexibility) needs

Flexibility requirements not met by existing or planned resources.

Upward flexibility

Flexibility linked to increasing injection into the network or decreasing demand from the network;

Very fast flexibility

Flexible capacity able to react to aggregated D-1 or shorter horizons forecast error variations (5 to 15 minutes before real-time).

FNA report for Belgium 2026, version approved by CREG

171

Annex: Analysis of uncovered renewable response needs before any renewable reaction. To support the assessment of potential solutions, including discussions on additional needs for market reaction of PV, the analysis presented in Figure 104 of the report has been complemented by the assessment shown in Figure 101, which assumes no photovoltaic (PV) response or other renewable reaction. This additional perspective provides an estimate of the total flexibility that would need to be unlocked from PV generation before accounting for any existing market-driven or operational reactions of PV. As such, it enables a clearer quantification of the residual flexibility requirement attributable to PV and helps identify the scale of response that would be needed to address the observed adequacy challenges.

Figure 104: Remaining uncovered renewable response needs for the different scenarios and sensitivities at the 1h level with a no PV response

FNA report for Belgium 2026, version approved by CREG

172