Serbia’s proposed €2.6 billion Iron Gate 3 pumped-storage hydropower project is advancing through contracted technical preparation, but its long-term investment case will depend on securing a Danube operating regime that protects existing Serbian and Romanian hydropower facilities, as well as downstream nuclear cooling requirements.
Known locally as Đerdap 3, the project could deliver substantial electricity storage, balancing capacity and flexibility to Southeast Europe’s power system. However, its commercial value will depend on when water can be withdrawn, stored and released, particularly during droughts when hydropower generation, river navigation and nuclear cooling face simultaneous pressure.
Serbia’s current starting configuration envisages up to 1,800 MW of installed capacity, although the final design and investment cost remain subject to technical studies and further review. The government aims to complete project documentation by the end of 2029, with construction still dependent on additional technical, environmental, bilateral and financing decisions.
A consortium led by the Jaroslav Černi Water Institute, controlled by Millennium Team, has secured the assignment for planning and early technical documentation. Meanwhile, Bechtel–Enka has qualified for further negotiations under the government’s international supplier-selection process. These developments represent meaningful progress, but they do not establish financial close or an unconditional commitment to construction.
This distinction is central to determining whether Đerdap 3 is entering a credible investment cycle or facing another prolonged period of feasibility work. Contracting technical studies advances project preparation, but the decisive test will be whether the scheme remains financially viable after binding water-management restrictions, construction risks and financing costs are incorporated into its business case.
Existing hydropower and nuclear generation define the constraints
The existing Iron Gates hydropower cascade already supports significant electricity generation on both sides of the Serbian-Romanian border.
Serbia’s Elektroprivreda Srbije (EPS) operates 1,140 MW at Đerdap 1 and 270 MW at Đerdap 2. On the Romanian side, Hidroelectrica operates 1,166.4 MW at Porțile de Fier I, while Porțile de Fier II includes approximately 251.2 MW in its main plant and another 54 MW at Gogoșu. Together, these facilities represent approximately 2.9 GW of installed hydropower capacity.
Their actual output depends on natural river flows, hydraulic head, equipment availability and coordinated reservoir operations. Installed capacity alone does not guarantee equivalent electricity production, particularly during low-water periods and prolonged droughts.
Further downstream, Nuclearelectrica’s Cernavodă nuclear power station, with approximately 1.4 GW of gross capacity across its two existing reactors, relies on the Danube as a cooling-water source.
The river consequently performs several critical functions within the same regional electricity system. It supplies water for hydropower generation, supports nuclear reactor cooling and would provide the working water for Đerdap 3’s pumped-storage operations.
These functions can complement one another under favourable conditions, but they also create competing water requirements during periods of scarcity. Any new storage scheme must therefore be assessed against the needs of existing energy infrastructure, rather than as an isolated electricity project.
The 2026 drought has highlighted these vulnerabilities. In October, both Cernavodă units were maintained in a safe shutdown state, with their return dependent on improved Danube water levels. Nuclear safety remained protected, but electricity production was interrupted.
Nuclearelectrica’s revised forecast for 2026 reduced expected annual electricity deliveries to the grid from 9.080 TWh to 6.743 TWh. The reduction also reflects an earlier transformer-related outage, meaning that the entire decline cannot be attributed to drought conditions alone.
Đerdap 3 did not cause these interruptions because the project has not yet been built. The events nevertheless demonstrate the operating vulnerabilities against which its future effects must be assessed, particularly if water availability becomes a binding constraint across the regional energy system.
Pumped storage changes the timing of water availability
Đerdap 3 is conceived as a pumped-storage hydropower development connected to the existing Iron Gate I reservoir. Under EPS’s established concept, water would be taken from the Danube near river kilometre 1007, pumped into elevated reservoirs at Pesača and Brodica, and subsequently returned through turbines when electricity is needed.
An earlier full-development concept envisaged 2,400 MW of capacity, approximately 578 million cubic metres of reservoir water and an energy equivalent of around 484 GWh. These legacy figures must be distinguished from the current 1,800 MW starting configuration, which remains subject to design development.
Pumped storage allows electricity to be shifted between periods of lower and higher market value. The plant consumes electricity to pump water uphill and recovers part of that energy during generation. Its commercial performance depends on storage duration, charging opportunities, electricity-price spreads and operational flexibility.
The project’s water balance will also differ between initial reservoir filling and routine operation. During initial filling, water is retained in newly constructed upper reservoirs. The resulting downstream effect would depend on the speed of filling, prevailing river conditions and the availability of compensating releases. A staged filling programme during high-flow periods could have substantially different consequences from rapid filling during a drought.
Routine pumped-storage cycling presents a different situation because water pumped into the upper reservoirs is subsequently returned to the river system. Over a completed cycle, lasting water losses would mainly arise from evaporation, seepage and other consumptive uses, rather than from the electricity lost through the pumping and generation process.
However, limited net annual water consumption would not prove that the project has no downstream impact. Temporary retention can still matter when river levels, navigation depths or agreed minimum releases are close to their operating limits. The timing of water withdrawals may therefore be as important as the total annual volume.
Cernavodă is located approximately 700 river kilometres downstream of the conceptual Đerdap 3 intake. Between the two locations are existing reservoirs, coordinated releases, tributaries and river branches, meaning that a withdrawal at Đerdap 3 would not necessarily translate into an immediate or identical reduction in flow at the nuclear station.
A credible assessment requires hydraulic modelling across the entire affected river system, including the timing and distribution of water reaching Cernavodă’s cooling-water supply route. The analysis must account for cumulative effects, reservoir operations and different drought scenarios, rather than focusing exclusively on water levels near the proposed Serbian powerhouse.
Romania is already addressing local cooling-water constraints through dredging and river-channel interventions, while also considering the potential requirements of an expanded Cernavodă nuclear station in the future. This reinforces the need for transboundary water modelling and shows why a Serbian assessment confined to the immediate project site would be insufficient.
Drought support could create strategic value, but reserves are finite
EPS has also presented Đerdap 3 as a potential mechanism for supporting downstream Danube water levels during droughts. Water stored in advance could theoretically be released during periods of low river flow, helping improve downstream conditions and potentially supporting nuclear cooling-water availability.
The underlying principle is physically plausible. Water retained before a drought could later be released, contributing to both electricity generation and downstream river flow. However, the practical benefit would depend on the volume available, the timing of release, downstream routing and whether the water had actually been reserved for this purpose.
For illustration, a reserve of 100 million cubic metres could provide an additional flow of 100 cubic metres per second for approximately 11.6 days, or 500 cubic metres per second for around 2.3 days, before accounting for losses and downstream routing. These figures are illustrative calculations, not forecasts of Đerdap 3’s future drought-support capacity.
The example demonstrates why installed megawatts alone cannot measure the project’s ability to support the Danube during a prolonged drought. The relevant factors are usable storage volume, release capacity, the duration of the shortage and the extent to which additional water reaches the locations where it is required.
A credible operating agreement would need to establish minimum drought-reserve volumes, release triggers, institutional responsibilities and compensation arrangements. It would also have to define when downstream water support takes priority over commercial electricity trading and other Serbian strategic requirements.
The central issue is that the same stored water cannot simultaneously be treated as fully available for unrestricted market operations, Serbia’s strategic reserve and guaranteed Romanian drought support. These uses must be reconciled through explicit operating rules and clearly defined priorities.
The most challenging scenario would be a prolonged regional drought combined with high electricity prices. Hydropower output could fall, nuclear cooling could become constrained and wholesale electricity prices could rise. Đerdap 3 might then generate electricity from previously stored water, but replenishing its reservoirs would require both electricity for pumping and permission to withdraw water under the prevailing hydrological conditions.
This creates a direct relationship between water availability and project revenue. Restrictions introduced to protect downstream users could limit reservoir replenishment precisely when electricity prices and the wider power system’s need for flexibility are at their highest.
Romania has important leverage over the operating regime
Romania’s strongest position stems from its rights and obligations within the existing shared Iron Gates system. The bilateral arrangements governing the cascade require each side to avoid actions that interfere with the other’s use of the installations. They also provide for agreement on certain modifications that could change the operating parameters of the jointly managed hydropower system.
The implications for Đerdap 3 will depend on its final configuration. If the project requires changes to shared infrastructure or agreed reservoir operations, Serbia cannot assume that domestic authorisations alone will be sufficient. The precise scope of Romania’s approval rights will depend on the relevant bilateral provisions and the technical effects of the selected design.
Romania could withhold agreement for a configuration that materially alters the functioning of the shared system, where such agreement is required under the applicable arrangements. Existing operating rules also restrict the freedom to retain water, including through minimum discharge requirements and limits on daily water-level variations in the relevant downstream reach.
These restrictions mean that Đerdap 3 could not simply retain water whenever electricity prices made pumping attractive. Commercial dispatch would have to operate within the applicable water-management rules, potentially reducing the number of hours available for profitable charging and generation.
Environmental assessment introduces a separate set of obligations. Projects with the potential for significant transboundary environmental impacts may require notification, assessment and consultation with neighbouring states under the applicable legal framework. This does not automatically give Romania a veto over every Serbian project authorisation, but environmental objections and unresolved hydraulic impacts could delay approval, require redesign or impose operating conditions.
Romania could therefore influence the project without demanding its outright cancellation. Conditions covering initial reservoir filling, pumping schedules, water-level fluctuations, drought reserves, minimum releases or compensation could materially alter the project’s viable configuration and commercial value.
For lenders and investors, an unresolved operating regime would represent a major project-financing risk. A technically feasible power plant may still prove commercially unattractive if the number of hours in which it can operate profitably is significantly reduced by cross-border restrictions and hydrological limitations.
Jaroslav Černi and Bechtel–Enka shape the preparation phase
The Jaroslav Černi Water Institute will play an influential role in defining the hydrological assumptions, environmental baseline, technical alternatives and cost framework underpinning subsequent investment decisions. Its findings will help establish which project configuration is technically feasible and economically defensible.
The contracted assignment includes a pre-feasibility study with general design, alongside a special-purpose spatial plan and strategic environmental assessment. The work must address potential cross-border impacts and the need to agree key parameters with neighbouring countries, making Romanian operating requirements and Danube hydrology central considerations.
Millennium Team controls the institute through its ownership structure. That relationship does not, by itself, demonstrate that the technical work is biased or that the project outcome has been predetermined. Nevertheless, it reinforces the case for independent technical and financial scrutiny, given that an institute within a wider construction group is helping define a project that could later generate substantial construction contracts.
A bankable development process should independently verify the preferred design, hydrological assumptions, reservoir requirements, construction costs and projected electricity-market revenues. It should also compare the proposed scheme with smaller, phased or alternative storage configurations to determine whether a less complex option could deliver better risk-adjusted returns.
The studies must be capable of supporting a decision not to proceed if the project fails to meet economic, environmental or operational requirements once realistic constraints are incorporated. A feasibility process that assumes unrestricted operation from the outset would risk overstating both generation potential and investment returns.
Meanwhile, Bechtel–Enka has qualified for further negotiations concerning additional project and technical documentation. This gives the consortium a strong prospective position in the preparation process, but qualification is not the same as a final construction award or a binding commitment to build.
Coordination between the Černi-led assignment and any subsequent Bechtel–Enka work will be critical. Reservoir design, geotechnical assumptions, Romanian operating conditions, grid requirements and cost contingencies must be based on a consistent technical framework. Otherwise, later revisions could result in redesign costs, delays and contractual claims.
An experienced international contractor could strengthen execution capacity and help manage complex engineering interfaces. However, construction expertise cannot resolve unresolved water-management rights or establish commercial viability on its own. Those questions must be settled before the project can be considered ready for investment.
Financing depends on the restricted operating scenario
The commercial rationale for Đerdap 3 rests on absorbing lower-value electricity and selling electricity when market prices are higher, alongside potential revenues from balancing and ancillary services. Yet pumped storage incurs efficiency losses, meaning that the selling price must cover more than the original cost of charging electricity.
Under an illustrative 80% round-trip efficiency, electricity purchased at €50/MWh would create an effective charging cost of €62.50 per MWh subsequently discharged, before operating expenses, network charges, financing costs and capital recovery. The project would need sufficient price spreads and service revenues to cover these additional costs.
The plant could also influence the market in which it operates. A large pumped-storage facility may increase prices during charging periods and reduce peak prices during generation, potentially narrowing the price spreads on which its original business case depends. A robust feasibility study therefore cannot simply apply historical price differences to all projected future output.
Hydrology adds another layer of uncertainty. The financial model must evaluate river flows, nuclear availability, renewable generation, electricity prices and transmission constraints together. A severe drought could simultaneously reduce hydropower output, constrain nuclear generation and push wholesale electricity prices higher.
Such conditions are precisely when flexible storage capacity could have its greatest system value. However, if water-management rules prevent or restrict reservoir replenishment, Đerdap 3 might be unable to capture the full commercial opportunity created by high prices and increased demand for balancing services.
The decisive financial assessment must therefore be based on a constrained operating scenario, rather than an unrestricted technical maximum. Investors need to understand the project’s expected revenues after water restrictions, environmental obligations, downtime, efficiency losses and financing costs are taken into account.
Beyond feasibility studies, but not yet an investment decision
Đerdap 3 has moved beyond a purely political announcement, with contracted technical preparation, government involvement, Serbian-Romanian engagement and interest from major engineering and construction groups. These developments make the project more credible than an early-stage concept without an established preparation process.
However, progress in project preparation is not equivalent to a financed construction project. The remaining challenges include a consistent final design, independently reviewed hydrological modelling, a clearly defined operating regime with Romania, realistic capital-cost estimates and a financing structure based on the restrictions the plant will actually face.
Serbia can commission studies, advance domestic procedures and negotiate with contractors, while Romania can materially influence the conditions governing the shared Danube system. Ultimately, river hydrology will determine how much water is available and when it can be used.
Whether Đerdap 3 becomes a strategic regional electricity-storage asset or another prolonged feasibility exercise will depend on how much commercial value remains after these three factors—Serbia’s investment decisions, Romania’s operating requirements and the Danube’s physical limits—are incorporated into a single, credible investment case.
