ACER Opinion 14-2026 on the Romanian National Resource Adequacy Assessment

No 14/2026

OPINION

on the Romanian National Resource Adequacy Assessment

9 July 2026

A C E R O P I N I O N N O 1 4 / 2 0 2 6

Executive summary

Under Regulation (EU) 2019/943, the European Network of Transmission System Operators for Electricity (ENTSO-E) conducts an annual European resource adequacy assessment (ERAA). Member States may perform national resource adequacy assessments (NRAAs) to capture national developments / specificities. Where, under the central reference scenario, an NRAA identifies risks not reflected in the ERAA, it must be submitted for ACER’s opinion, along with an explanation about the differences between the national and European assessments. Transelectrica – the Transmission System Operator for Romania – conducted an NRAA. The NRAA complements ERAA 2025 edition by reflecting recent developments in Romanian electricity system.

Key Findings of the NRAA’s central reference scenario

• Adequacy risk in 2027, with annual loss of load expectation (LOLE) of 143.55 hours – higher than in ERAA 2025 (7.59 hours in 2028) and above the 13-hour reliability standard; primarily driven by lower resource capacity. • Adequacy risk in 2035, with LOLE of 6.69 hours – above ERAA 2025 estimates (0.2 hours) and exceeding the 2-hour reliability standard. The risk is primarily driven by local transmission grid bottlenecks, conservative battery deployment, the absence of price-sensitive demand response and expected delay in new nuclear capacity.

ACER found the following differences justified

• The following simplifications adopted in the NRAA – the regional scope, the fewer Monte Carlo simulations, and the omission of curtailment sharing – reduce modelling complexity while keeping an acceptable quality of the assessment and enabling detailed modelling of the Romanian transmission grid. • Lower renewable capacities – reflect recent trends.

ACER found the following differences not justified

• The partially implemented economic viability assessment, which is necessary to assess whether capacity is financially viable and, therefore, likely to be built or maintained. • The accelerated coal phase-out does not align with slower gas and wind capacity deployments. • The delayed commissioning of new nuclear capacity does not align with recent timelines. • The conservative battery deployment assumptions that project no additional battery installations after 2030, despite ongoing renewable energy expansion and favourable policies. • The absence of demand side flexibility, despite rapid demand growth in price-sensitive sectors and introduction of new demand side flexibility regulatory measures.

Key recommendations for the central reference scenario

ACER highlights that the unjustified differences could result in mis-estimation of security of supply and have significant impact on adequacy results. To address these key issues and inconsistencies, ACER recommends to: • Revise thermal capacity assumptions to reflect recent trends and the most likely projections. • Fully implement the economic viability assessment to support decision-making on future resource capacity mix and assumptions for new non-fossil technologies. • Consider the impact of national policies that promote demand response and storage and align projections with other studies (i.e. ENTSO-E Seasonal Outlooks). • Apply most likely electrification assumptions with the pace of renewable deployment.

A C E R O P I N I O N N O 1 4 / 2 0 2 6

1. Background

1 Article 20 of Regulation (EU) 2019/943 (Electricity Regulation) requires Member States to monitor resource adequacy within their territory based on the European Resource Adequacy Assessment (ERAA). It also allows them to complement ERAA by conducting a National Resource Adequacy Assessment (NRAA) according to Article 24 of the Electricity Regulation. Either of the two assessments can identify resource adequacy concerns, but both assessments must be based on the ERAA methodology.

2 Where resource adequacy concerns are identified in the central reference scenario (CRS), they should first be addressed by eliminating regulatory distortions or market failures, and, where it does not suffice, by introducing capacity mechanisms.

3 According to Article 24(3) of the Electricity Regulation, where an NRAA identifies an adequacy concern that was not identified in ERAA, it must include reasons for the divergence between the two assessments. The Member State must publish the NRAA and submit it to ACER for an opinion. ACER assesses, on a case-by-case basis, whether the differences between the two assessments are justified. ACER primarily considers the differences identified and reasoned by the Member State, but may also identify further differences, if they also have a material impact on the results.

4 The national body responsible for preparing the NRAA, which can be a transmission system operator (TSO) or a different body designated by the Member State, must take into account ACER's opinion, and, where necessary, amend the assessment. Where the body decides not to take ACER's opinion fully into account, it must publish a report with detailed reasons.

5 Romania is currently phasing out coal-fired power plants while expanding renewables and gasfired generation as a transitional fuel, in line with its National Recovery and Resilience Plan (NRRP). According to the Annex to the Council Implementing Decision, the indicative timeline for full coal decommissioning is 2032, with a partial phase-out reducing the coal- and lignite-fired capacity to about 1 GW by 2026. The Romanian National Energy and Climate Plan (NECP) foresees a faster phase-out pathway.

6 Due to delays in the commissioning of new gas-fired replacement capacity, the Romanian government notified the European Commission in December 2025 of a three-year extension of the electricity system restructuring plan, until the end of 2029. This, in turn, may further delay the phase-out of coal capacities as per press release.

7 Against this background, the Romanian Ministry of Energy has considered the NRAA framework as an evidence-based method of assessing whether the delayed gas-fired capacity entry along with the planned pace of coal phase-out could affect security of supply in Romanian electricity system in 2027 and 2030. Thus, Transelectrica, the Romanian TSO, carried out an NRAA which was submitted to ACER on 27 April 2026 by the Romanian Ministry of Energy.

8 This Opinion evaluates whether the differences between the Romanian NRAA and ERAA 2025 edition are justified. It is addressed to the Romanian Ministry of Energy and is issued in accordance with Article 9(2) of Regulation (EU) 2019/942 and Article 24(3) of the Electricity Regulation.

2. Summary of the Romanian NRAA CRS

9 The aim of the Romanian NRAA is to complement ERAA by capturing recent market trends in demand and resource deployment within the Romanian electricity system.

10 The adequacy results of the CRS in the NRAA are shown in Table 1. The NRAA assesses the target years 2027, 2030 and 2035. For 2027 and 2035, the NRAA identifies new adequacy A C E R O P I N I O N N O 1 4 / 2 0 2 6 concerns that were not detected in the last ERAA. The modelled Loss of Load Expectation (LOLE) in the NRAA exceeds the applicable reliability standard (RS) of 13 hours in 2027 and 2 hours in 2035.

11 RS changes across target years in the NRAA are driven by changes in the inputs used to estimate the RS. RS is estimated as a ratio of the cost of new entry (CONE) of a reference technology to the value of lost load (VOLL). While VOLL increases slightly in each successive target year, the reference technology changes from battery to demand side response (DSR) in 2035. Because DSR has a significantly lower CONE than battery, this shift results in a substantially lower RS in 2035.

12 The divergence in the LOLE results is due to differences in the modelling assumptions and input data used in the CRS for the NRAA and ERAA. Compared with ERAA, the key factor driving the high LOLE values in the NRAA is the difference in resource mix volumes provided by Transelectrica.

13 Figure 1 shows the main differences in resource capacities between the NRAA and ERAA 2025 edition under the CRS for the first target years. ERAA assumes substantially higher battery storage, wind and solar capacities than the NRAA, contributing to the absence of adequacy concerns in 2028. In ERAA, the economic viability assessment (EVA) adjusts initial capacity assumptions based on economic feasibility and provides the updated resource mix as input to the economic dispatch adequacy model.

14 ERAA starts with 850 MW of lignite-fired capacity and around 4 GW of gas-fired capacity in the pre-EVA phase (assuming the timely commissioning of new gas capacities). During the EVA A C E R O P I N I O N N O 1 4 / 2 0 2 6 phase, most lignite and half of the gas capacity are assessed as economically unviable and therefore decommissioned due to overcapacity.

15 ERAA assumes around 1 GW higher hydro capacity than the NRAA, where hydro capacity was reduced to reflect historically observed lower generation levels. Meanwhile nuclear capacity assumptions are consistent between the NRAA and ERAA. Approximately half of the current Romanian nuclear fleet is assumed to undergo refurbishment works in 2027.

16 ERAA assumes higher solar and wind capacities than the NRAA. However, the contribution of these technologies during scarcity depends on weather conditions and may be limited. In particular, solar generation provides low contribution during scarcity periods (around 2% from installed capacity ). Solar generation peaks around midday while scarcity often occurs in the evening when solar irradiance is low or absent. Hence the diverging assumption on solar capacity is not assessed in this opinion.

17 The differences between the NRAA and ERAA assumptions are discussed in detail and assessed by ACER in the following section.

3. ACER’s assessment: NRAA vs ERAA

18 The CRS differences between the NRAA and ERAA fall into two categories: (i) methodological and (ii) input data. Each difference is listed in Table 2, along with its rationale, and potential impact on LOLE. A detailed description of these differences is provided below.

3.1. Regional scope

19 Description and rationale: The Romanian NRAA, compared to the pan-European ERAA, has a smaller geographic scope limited to Romania, Bulgaria, Hungary, and Serbia. The entire Romanian transmission grid with 400 kV, 220 kV and most of 110 kV lines is represented in the NRAA model, while ERAA models the Romanian electricity system as a single node.

20 The transmission grid model for the assessment target years is updated in line with the Transelectrica’s 10-year Development Plan. This plan focuses on expansion and modernization A C E R O P I N I O N N O 1 4 / 2 0 2 6 of Romania's high-voltage electricity grid to accommodate growing electricity demand and increasing renewable generation.

21 Demand and renewable generation profiles are allocated across transmission grid nodes based on historical data and ongoing grid connection projects. Where uncertain, capacities are distributed in a way that minimises the overall grid loading. This allocation is then calibrated against historical observations to capture seasonal and daily transmission grid loading patterns.

22 The outcome of the extended model was an increased LOLE in the CRS for the target year 2035 compared to the single node reduced model, reflecting the impact of internal transmission grid bottlenecks on system adequacy.

23 Bulgaria, Hungary, and Serbia are each represented as single node with demand, resource, and cross-border net transfer capacity (NTC) assumptions aligned with ERAA 2024.

24 Hourly power exchanges with Moldova and Ukraine are based on hourly physical flows from 2024 obtained from the Transparency Platform maintained by ENTSO-E and are scaled proportionally to the projected increase in cross-border transmission capacities for the modelled target years. This approach is in line with the ERAA methodology, where non-explicitly modelled zones are represented by fixed time series of energy exchanges through interconnections. Cross-border exchanges with other non-modelled regions outside the geographic scope are not modelled.

25 Relevance: Reducing the geographical scope in electricity system modelling reduces complexity and modelling time. This enables faster and more resource-efficient analyses, allowing for a deeper focus on a specific region. The inclusion of the internal transmission grid introduces additional modelling detail, which improves the accuracy of adequacy assessments by capturing potential local grid bottlenecks.

26 Assessment: The reduced spatial coverage in the Romanian NRAA includes the key bidding zones relevant to Romania – Bulgaria, Hungary, and Serbia. The NTC approach is appropriate for cross-border exchanges of those countries with Romania, except for Hungary. Romania and Hungary are both part of the Core Capacity Calculation Region (Core CCR), where the flowbased market coupling (FBMC) modelling approach is applied in ERAA. Despite this simplification, the regional scope adopted in the NRAA is considered justified.

27 ACER recommends considering cross-border exchanges with non-modelled Member States outside the defined geographic scope, as imports from these regions may contribute to security of supply during scarcity situations. In particular, ACER recommends implementing the FBMC approach for exchanges with the Core CCR, as Romania is a net electricity importer from Hungary as per ENTSO-E Transparency Platform.

28 ACER welcomes Transelectrica’s distinction between adequacy risks caused by resource scarcity and those arising from internal grid constraints, as the Romanian transmission grid is fully represented in the NRAA. While modelling the internal grid provides a more realistic assessment and may reveal local bottlenecks, such constraints alone do not indicate a market failure requiring a capacity mechanism. Internal grid-related restrictions are structural issues that cannot be resolved through a capacity mechanism and should instead be addressed through transmission network development planning and other measures aimed at improving efficient use of network capacity.

3.2. Lower number of Monte Carlo runs

29 Description and rationale: In ERAA, each of the 36 weather scenarios (WSs) for a given target year is simulated 15 times, with each simulation applying a different random distribution of forced outages across the generation units and interconnectors between bidding zones. The NRAA also uses the 36 WSs from ERAA 2024 but limits the analysis to a single but distinct forced outage simulation per WS. A C E R O P I N I O N N O 1 4 / 2 0 2 6

30 As per the ERAA methodology, sufficient Monte Carlo (MC) runs are achieved when the coefficient of variation of the modelled energy not served (ENS) does not change more than a predefined threshold with each additional MC run, signalling acceptable modelling accuracy. This reduced number of MC runs in the NRAA reflects a balance between modelling accuracy and run time, as further described below.

31 Relevance: Multiple weather scenarios and forced outage runs deliver robust adequacy results accounting for uncertainty, capturing a wide range of potential system states and ensuring a robust analysis.

32 Assessment: In the NRAA, the relative change in the coefficient of variation of ENS between th th consecutive MC runs ranges from 1% to 4% between the 25 and 36 MC runs. For computationally intensive adequacy models, representing the entire transmission grid, this convergence could be considered acceptable. Although more MC runs would further improve convergence, the observed variation indicates that the ENS results are approaching stability and that further simulations may provide limited incremental improvements relative to the associated computational effort. Provided that the overall adequacy conclusions remain robust, the adopted modelling approach can therefore be considered justified.

3.3. No curtailment sharing

33 Description: The optional curtailment sharing feature, while present in ERAA, is not included in the NRAA to reduce modelling complexity.

34 Relevance: Curtailment sharing is embedded in the EUPHEMIA algorithm used for day-ahead electricity trade among EU bidding zones. It evenly distributes resource shortages across interconnected bidding zones whose domestic capacity is insufficient to meet local demand. In electricity system modelling, this feature is relevant in multi-bidding-zone simulations to ensure accurate LOLE results under conditions of simultaneous scarcity.

35 Assessment: Curtailment sharing can have a significant impact on LOLE results. In ERAA 2025 , its application roughly doubled the average LOLE per bidding zone, as presented in ACER’s ERAA 2025 Decision. In Romania, the inclusion of curtailment sharing had a significant impact on adequacy outcomes in 2028, increasing LOLE by 4.4 hours to a total of 7.6 hours. In contrast, its effect was much more limited in 2030 and 2035, with LOLE increasing by only 0.2 to 1.4 hours. Nevertheless, in ERAA, the RS in Romania was not exceeded in any target year, solely due to the application of the curtailment sharing feature.

36 Although curtailment sharing has a notable effect on adequacy outcomes in 2028, its omission in the NRAA is considered justified. Curtailment sharing is a feature of the flow-based market coupling approach in the EUPHEMIA algorithm; however, the flow-based approach itself is not modelled in the NRAA.

3.4. Partial economic viability assessment

37 Description and rationale: In the NRAA, the Economic Viability Assessment (EVA) functions as an economic screening layer applied after the Economic Dispatch (ED) modelling phase. It means that the EVA is used solely to evaluate the financial viability of the predefined initial resource capacity mix, rather than to influence or adjust that mix during the modelling process itself. Consequently, the EVA findings in the NRAA are indicative only, as they do not modify the initial capacity projections.

38 Relevance: The objective of the EVA is to assess whether existing and prospective power resources are financially sustainable within an energy-only market. It serves as a proxy for A C E R O P I N I O N N O 1 4 / 2 0 2 6 investor behaviour, helping identify a feasible future resource mix. In doing so, it provides a more realistic basis for evaluating future supply adequacy.

39 Assessment: The EVA approach in the NRAA is not fully consistent with the ERAA methodology, as it currently applies the EVA only as a post-processing economic screening of a predefined capacity mix. Article 6(5) of the methodology indicates that the EVA is more than an evaluation tool. It shapes the initial capacity mix by determining whether to keep, retire, mothball, re-enter, renew, or add capacity resources based on their economic viability. Without full implementation of EVA, the resource adequacy assessment risks misestimating the security of supply and potentially leading to suboptimal policy decisions. Therefore, limiting EVA use to a postprocessing screening in the NRAA is not justified.

40 ACER recommends including the EVA modelling phase into decision-making on future resource capacity mix, based on economic viability of resources within the CRS of the NRAA assessment.

41 ACER recommends that the assessment ensures strong consistency between the EVA and ED models. Attention should be paid to the selection of weather scenarios (WSs) in the EVA. These scenarios should capture the overall climatic variability in Romania present across the complete ED WS set. This will help ensure that the EVA model identifies the consistent level of investment. In addition, the day-ahead market revenues observed under the selected EVA WSs should be representative of revenues across the entire ED WS set, ensuring alignment between the two models. Also, technical constraints should be aligned in the two models.

42 ACER recommends that the assessment considers incorporating revenue stacking modelling within the EVA, with particular relevance for battery storage, while remaining applicable to all generation and demand-side resources. In addition to revenues from day-ahead market trading, potential income streams may come from intraday markets and the provision of ancillary system services.

3.5. Lower hydro capacity

43 Description and rationale: Compared to ERAA 2025, the NRAA CRS assumes a lower run-ofriver hydro capacity in Romania – about 1 GW less across all target years. This adjustment lowers peak hydro generation to align with historical observations over the past 15 years.

44 Relevance: Accurately capturing the historical generation patterns is essential for reflecting actual system behaviour. Hydro availability is highly dependent on hydrological conditions, which can vary significantly from year to year due to changes in water inflow.

45 Assessment: The adjustment of hydro capacity in the NRAA, grounded in historical observations, is justified as it supports more realistic modelling of hydro generation.

46 ACER recommends using the actual hydropower capacity in the NRAA and limiting generation through availability factors, as in ERAA. This preserves the real capacity in the model while aligning generation with historical levels, avoiding the impression that hydropower is being decommissioned.

3.6. Lower wind capacity

47 Description and rationale: Compared to ERAA 2025, the NRAA CRS assumes 3 GW less onshore wind capacity in Romania in the first target year as depicted in Figure 2. This update is based on revised input provided by Transelectrica. A C E R O P I N I O N N O 1 4 / 2 0 2 6 8 7 6 Transparency Platform 2026 5 Summer Outlook 2026 4 NRAA Target Year 2027 3 ERAA post-EVA Target Year 2028 2 ERAA pre-EVA Target Year 2028 1 0

48 Relevance: Accurately capturing historical capacity deployment trends is essential for making realistic future capacity projections. Overestimating future wind capacity may lead to an overly optimistic assessment of available generation resources and system adequacy, particularly during periods when renewable output is expected to contribute significantly to meeting demand.

49 Assessment: The revised onshore wind capacity trajectory in the NRAA is more consistent with the current pace of wind deployment in Romania, which is 3 GW in 2026, according to the ENTSO-E Transparency Platform and the ENTSO-E Summer Outlook 2026. The projected increase to 4.4 GW by 2027 represents a credible deployment rate on the path toward the 2030 NECP target of 7.4 GW. Aligning projected wind capacity to the current deployment trends supports more realistic modelling and is therefore justified.

3.7. Conservative battery deployment

50 Description and rationale: Compared to ERAA 2025, the NRAA CRS assumes lower battery capacity as depicted in Figure 3. The NRAA battery capacity is 0.5 GW (in 2027) compared to 1.4 GW (in 2028) in ERAA, and 1.2 GW vs 2 GW in 2030. Battery capacity remains unchanged beyond 2030 in both assessments, assuming no new battery installations after 2030. 2 Transparency Platform 2026 Summer Outlook 2026 1 NRAA Target Year 2027 ERAA post-EVA Target Year 2028 ERAA pre-EVA Target Year 2028 0

51 The 2027 battery capacity projection in the NRAA is close to the current capacity in Romania (493 MW) as per ENTSO-E Transparency Platform. Further battery capacity expansion could be expected due to the M€150 Romanian State aid scheme for electricity storage to be granted before 31 December 2030, which may bring extra 2174 MWh . A C E R O P I N I O N N O 1 4 / 2 0 2 6

52 Relevance: The conservative assumptions on the expected storage development may contribute to delaying decommissioning of fossil-fuelled generators. This, in turn, may impact the assessment of the indicative objectives for non-fossil flexibility that Member States are expected to establish under the Electricity Market Design Regulation (EU) 2024/1747.

53 Assessment: The Romanian Regulatory Authority (ANRE) has drafted a new regulation eliminating double tariffs for electricity stored and reintroduced into the grid, improving economic viability and investment attractiveness of energy storage projects in Romania. Previously, electricity stored and later re-injected into the grid was subject to network charges twice, creating a significant barrier to investment.

54 The removal of this double charging along with the dedicated support scheme sends a strong signal to investors that Romania is committed to supporting energy storage as a key component of the energy transition. Accordingly, conservative assumptions regarding storage deployment in Romania are not justified.

55 ACER recommends aligning battery capacity projections with other studies, i.e. ENTSO-E Seasonal Outlooks for the shorter term. In the Summer Outlook 2026, the projected battery capacity is 600 MW, whereas the NRAA projects 500 MW for the target year 2027. For the longer term, it would be beneficial to consider the impact of the favourable tariffication and the planned support scheme, as this could facilitate greater storage deployment.

3.8. Delayed entry of new gas-fired capacity

56 Description and rationale: Compared with ERAA, the NRAA CRS assumes delayed commissioning of the new fossil gas-fired capacity in Romania. In 2027, in the NRAA, the gasfired capacity is assumed to be 2.2 GW which is close to the current capacity as per ENTSO-E Transparency Platform (see Figure 4). Although ERAA starts with nearly 4 GW gas capacity in 2028 (pre-EVA), half of this capacity is decommissioned during the EVA phase due to a lack of economic viability (assuming a higher hydro, renewable and battery capacity mix in ERAA). 5 Transparency Platform 2026 4 Summer Outlook 2026 3 NRAA Target Year 2027 2 ERAA post-EVA Target Year 2028 1 ERAA pre-EVA Target Year 2028 0

57 By 2030, gas-fired capacity in the NRAA increases to 3.6 GW (1.7 GW less compared to the pre- EVA capacity in ERAA). By 2035 all planned gas-fired capacity of 5.4 GW is online in the NRAA as in the pre-EVA ERAA.

58 Relevance: Delays in the commissioning large, highly available thermal power plants can significantly increase the risk of resource shortages and should be accurately considered in electricity system planning. A C E R O P I N I O N N O 1 4 / 2 0 2 6

59 Assessment: Two major projects – 460 MW CCGT plant in Turceni and an 825 MW CCGT plant in Ișalnița – are facing construction delays. On 30 January 2026, the second tender for the construction of Ișalnița was cancelled after no successful offers were received before the submission deadline.

60 According to the Commission’s press release, Romania’s originally planned commissioning date of 2026 is postponed to 2029. These delays could also trigger the postponement of the phaseout of existing lignite-fired capacity, which may remain in operation longer than previously expected. The Turceni and Ișalnița plants are assumed to be operational by 2035 under the NRAA, which is reasonable given implementation delays.

61 For other planned gas-fired capacity, Romanian Government communications indicate strong support for accelerating the strategic Mintia project. The government expects the 1.7 GW power plant to begin generating electricity from September 2026 and emphasizes the need to accelerate grid connection works to ensure timely completion. Additionally, a manufacturer’s press release indicated that the Mintia plant may begin simple-cycle operation prior to combined-cycle completion. If commissioned as planned, Mintia would almost double Romania’s gas-fired capacity by the end of 2026 and make a substantial contribution to the capacity available in 2027.

62 However, following a project developer’s request to extend the Mintia project completion deadline, the Romanian Ministry of Energy agreed to postpone the commissioning date until the end of 2027. This supports the NRAA CRS assumption of a delayed entry into operation of new gas-fired capacity and is therefore justified.

63 ACER recommends considering in the assessment a phased Mintia construction in which the open-cycle gas turbines could enter operation and generate electricity before full completion of the steam cycle.

3.9. Lower lignite capacity

64 Description and rationale: As hard coal capacity is relatively low in both assessments – 135 MW in ERAA and 70 MW in the NRAA CRS until 2030, and zero thereafter – most differences relate to lignite-fired generation capacity. Compared to ERAA 2025, the NRAA CRS assumes an earlier decommissioning of lignite-fired generation, with just 155 MW of lignite capacity remaining available in the first target year (2027), as shown in Figure 5. By contrast, the ERAA 2025 pre- EVA assumption is that 854 MW of lignite-fired capacity remains available until 2030. However, this capacity is removed in the EVA phase due to lack of profitability, so it does not reach the economic dispatch adequacy model. 2 Transparency Platform 2026 Summer Outlook 2026 1 NRAA Target Year 2027 ERAA post-EVA Target Year 2028 0 ERAA pre-EVA Target Year 2028

65 ERAA pre-EVA capacity reflects the originally planned phase-out schedule for coal- and lignitefired generation under Romania's National Recovery and Resilience Plan (NRRP) . The NRRP gives a two-step phase-out process, (i) a partial decommissioning of coal- and lignite-fired A C E R O P I N I O N N O 1 4 / 2 0 2 6 capacity, reducing total installed capacity to approximately 1 GW by 2026; and (ii) a complete phase-out by 2032.

66 The NRAA, however, follows the Romanian National Energy and Climate Plan (NECP), which foresees a faster phase-out pathway. Under the NECP, the remaining coal and lignite capacity that could otherwise have remained operational until 2032 is already decommissioned (or placed on standby) at the end of 2025.

67 Relevance: In the CRS, aligning the timing of thermal capacity retirements with electricity demand and resource capacity developments in the rest of the system is important, as thermal resources provide highly available and dispatchable capacity during scarcity periods due to their relatively high derating factors.

68 Assessment: The CRS in the NRAA builds the ‘Trends & Projections scenario’ as defined in Article 3 of the amended ERAA methodology by incorporating recent developments in slower wind deployment and delays in the commissioning of new gas-fired capacity. Meanwhile, the accelerated lignite phase-out assumption in the CRS is not consistent with this scenario, as it does not reflect the slower pace of the Romania’s energy transition due to the delayed availability of replacement capacity. The slower decommissioning of lignite is instead portrayed in the NRAA in 2027 as a sensitivity (SA1).

69 Although the lignite phase-out trajectory in the NRAA CRS mirrors the timeline set out in the NECP, it does not consider the flexibility option in the NECP, which allows for keeping about 1 GW lignite capacity on standby. Maintaining this additional lignite capacity as shown in SA1 would be consistent with the ‘Trends & Projections scenario’, in the same way as the other updated capacity assumptions reflected in the NRAA CRS.

70 The accelerated lignite phase-out assumption – while considered a valid assumption by ACER – would be more appropriately addressed as part of a sensitivity scenario, aimed at assessing the impact of an earlier lignite exit under a slower-than-expected deployment of new wind and gas capacity. Therefore, applying an accelerated lignite phase-out assumption within a scenario characterised by a slower energy transition cannot be considered justified in the NRAA CRS.

3.10. Delayed entry of new nuclear capacity

71 Description and rationale: While ERAA 2025 assumes that a new nuclear power plant with about 1.8 GW of capacity is commissioned in Romania by the final target year (2035), the NRAA CRS omits units 3 and 4 at the Cernavodă power plant (2x700 MW) and the Small Modular Reactor (462 MW) at the Doicești plant due to uncertainty surrounding the development timelines. These NRAA assumptions increase adequacy concern in Romania only in 2035.

72 Relevance: Overly conservative assumptions on commissioning dates may lead to an overstated adequacy risk and could distort the identification and timing of potential system needs.

73 Assessment: According to Nuclearelectrica, commissioning of Cernavodă Unit 3 is planned for 2030 and Unit 4 for 2031, consistent with ERAA projections. However, the Romanian Ministry of Energy anticipates potential delays, with commercial operation now expected only in mid-2035. Thus, excluding the full capacity of Cernavodă units 3 and 4 from the CRS for the entire target year is not justified. A C E R O P I N I O N N O 1 4 / 2 0 2 6

74 ACER recommends that the assessment considers the availability of capacity from Cernavodă units 3 and 4 from mid-2035 as per the latest available commercial operation date. Adding capacity during the modelled year can significantly affect adequacy results, as most scarcity hours in the Romanian electricity system occur in December, according to ERAA modelling results.

3.11. No demand-side flexibility

75 Description and rationale: Annual electricity demand in the NRAA CRS, as provided by Transelectrica, is higher than in ERAA 2025 across all target years. In the NRAA, demand in 2027 is 4% higher (+2 TWh) than the 2028 target-year demand in ERAA; by 2030, the gap increases to 11% (+7 TWh), and by 2035 to 20% (+13 TWh).

76 Demand increase in the NRAA is primarily driven by the updated national assumptions regarding additional electricity consumption from electric vehicles (EVs), heat pumps (HPs) and data centres, which are not reflected in ERAA. In the NRAA model, EVs and HPs follow predefined consumption profiles, while data centres contribute continuously to the baseload. These additional loads are assumed to be non-price-sensitive, meaning their consumption does not decrease during scarcity periods. Although DSR is considered as a reference technology in the estimation of the RS for 2035, it is not explicitly modelled in the NRAA.

77 Relevance: Accurately capturing the most probable future activation potential of the demand-side flexibility resources – based on recent historical developments and supported by national policy trends – helps prevent overinvestment in peaking generation capacity.

78 Assessment: The NRAA incorporates updated electricity consumption projections that reflect recent national developments towards electrification and expected expansion of energy-intensive new consumers, such as data centres, allowing to better capture the future demand in Romania.

79 From ENTSO-E’s Transparency Platform, the maximum historical peak electricity demand is 9.3 GW in Romania (January 2026), which is close to the 2027 domestic consumption peak of 9 GW projected in the NRAA. However, the NRAA considers an additional 3TWh from EVs, HPs and data centres in 2027 which, combined, may increase the projected peak demand by up to 1.3 GW (1 GW above the historical maximum). The absence of demand flexibility assumptions for this additional demand may lead to overestimated adequacy risks.

80 There is a clear policy direction supporting the development of demand-side flexibility in Romania. Romania’s draft Regulation on demand flexibility introduces demand‑side flexibility as a market‑based product that Transelectrica can procure during system stress via daily auctions. The Regulation aims to establish rules for market‑based procurement and the use of the demand flexibility services.

81 Despite well projected demand volume, the absence of demand-side flexibility in a renewableresource-heavy electricity system is likely to underestimate the full future flexibility potential and lead to an overestimation of (i) system risks and (ii) the need for additional generation resources. In this context, considering that DSR was selected as the reference technology to calculate RS in 2035, and in light of the above-mentioned Regulation, the exclusion of demand flexibility from the NRAA is not justified.

82 ACER recommends that the assessment – including inputs to forthcoming ERAAs – considers price-sensitive demand flexibility in the context of rapid demand growth and high renewable energy deployment. Such approach should capture emerging sources of flexible demand, as well as the effects of national policies and incentives aimed at activating additional flexibility. A C E R O P I N I O N N O 1 4 / 2 0 2 6

83 ACER recommends aligning demand projections with generation resources. In the NRAA demand and resources develop in opposite directions – high demand projections followed by conservative resource planning.

4. Conclusions and recommendations

84 Table 3 presents ACER’s assessment of whether the differences in the Romanian NRAA CRS, compared with ERAA 2025 CRS, are justified.

85 The modelling differences between the NRAA and ERAA 2025 assessments contribute to divergent estimates of adequacy results. ACER considers that 6 out of the 11 differences are justified. However, it finds that the omission of fully implemented EVA, the conservative assumptions regarding battery deployment, the early coal phase-out, the absence of demand flexibility, and the delayed entry of nuclear capacity are not justified.

86 The main driver of adequacy concern in Romania in the first target year (2027) is (i) slower-thanexpected wind capacity deployment and (ii) missing thermal capacity – the combined impact of the early coal phase-out and delayed entry of the 1.7 GW gas-fired power plant at Mintia.

87 ACER notes that the NRAA includes the sensitivity scenario SA1 in which the only difference compared with the CRS in 2027 is that the 854 MW of available lignite-fired capacity remains in operation (i.e., is not decommissioned). The sensitivity shows that keeping this capacity in operation reduces the LOLE in the 2027 target year to 11.34 hours, thereby bringing it below the reliability standard of 13 hours. This sensitivity in 2027 appears to better reflect the most likely scenario, as it aligns the coal phase-out timeline with the actual pace of Romania’s delayed energy transition and it should, therefore, be captured in the NRAA CRS assumptions.

88 ACER also acknowledges the value of including additional sensitivity scenarios in the NRAA. These sensitivities assess a range of possible assumptions, distinct from the CRS, by considering alternative combinations of generation resource developments across all target years (e.g. higher battery deployment, the commissioning of new nuclear capacity by 2035, and different timelines for new gas-fired capacity additions), as well as lower and higher electrification rates. Through these sensitivity analyses, the NRAA provides recommendations on how to securely transition the Romanian electricity system towards future national planning objectives, while accounting for uncertainties and potential variations in resource availability. A C E R O P I N I O N N O 1 4 / 2 0 2 6

89 Other differences – the conservative battery deployment and the absence of price-sensitive demand response – mainly affect adequacy outcomes in the later target years, 2030 and 2035, when significant demand response potential and additional battery deployment are expected.

90 ACER flags that the delay in commissioning 1.8 GW of new nuclear capacity would increase the adequacy concern in Romania only from 2035 onwards. However, this delay remains subject to uncertainty and should be reassessed once greater clarity on project development timelines becomes available.

91 Finally, ACER notes the importance of providing updated information on available generation capacity and flexibility resources for consideration in upcoming ERAA editions, to ensure that adequacy projections reflect the most recent information available.

92 By addressing the recommendations in this opinion, which are also summarised in Table 4, ACER considers that the NRAA would be on a clear path to aligning with ERAA towards a robust and realistic adequacy assessment of the Romanian electricity system. This Opinion is addressed to the Romanian Ministry of Energy. Done at Ljubljana, on 9 July 2026. — SIGNED — V. ZULEGER, ACER Director ad interim