Solar Industry Romania / investment case study / battery storage

Romanian BESS bankability after the spread trade.

The investment case is no longer “does Romania have enough volatility for batteries?” The relevant question is whether a project still services capital after more batteries compete for the same evening spreads and balancing products. The answer is yes for a narrow class of projects: secure bidirectional grid rights, disciplined 2–4 hour sizing, bankable equipment and integration, revenue models built from dispatch rather than headline spreads, and financing structures that do not require today’s balancing economics to persist unchanged.

Executive decision

Finance the rights and dispatchability, not the spread chart.

Romania now has observable daylight price compression, negative quarter-hours, large noon-to-evening system reversals and visible evening storage dispatch. It also has rapidly expanding operating BESS, large authorised pipelines and non-recourse financing precedents. Those facts make storage investable and simultaneously make naïve merchant underwriting obsolete.

Underwriting default 2h

Use a two-hour system as the default financing reference unless the third or fourth hour has separately demonstrated marginal value. Romania's disclosed large-scale lender precedents are overwhelmingly 2h.

Bankable merchant route Energy

Energy arbitrage can support base-case underwriting, but only through PT15 dispatch modelling after efficiency, degradation, market costs, SOC restrictions and competition. A daily Base-minus-daylight spread is not revenue.

Revenue requiring the largest haircut Balancing

Treat balancing and ancillary revenue as upside unless the project has qualification evidence, product-level historical data, a saturation curve and an optimiser back-test. New BESS and controllable load are direct competitors.

Strongest downside structure Floor + merchant

A fixed or minimum availability/tolling component plus merchant optimisation is the strongest structure where available. Fully merchant financing exists, but current Romanian precedents materially use public risk-sharing or strong institutional lenders.

Investment rule. Do not approve a Romanian BESS because the current spread is wide. Approve it only when grid rights, dispatch simulation, technical warranty, optimiser capability and downside debt service remain coherent after separately stressing energy-arbitrage revenue and balancing revenue by at least 50%. The 50% haircut is an underwriting stress, not a forecast probability.

90-day market diagnostic / 24 Jun–21 Sep 2026 RON/MWh unless stated / daily PZU references

PZU Base / mean777.2490 delivery days

08:00–17:00 proxy / mean422.51Unweighted daylight proxy

Mean daylight discount49.5%Derived from daily proxy/Base ratios

Negative PT15 intervals1511.75% of 8,640 quarter-hours

Scope and premise challenge

The bankability problem is saturation, not proof of use-case.

The empirical record already demonstrates a use-case for storage. The harder problem is deciding which portion of that value can survive entry by more batteries, more flexible demand and stronger cross-border/grid flexibility over a debt tenor.

The 90-day PZU record shows recurring daylight discounts and repeated negative PT15 prices. Retained Transelectrica observations show Romania moving from midday export to heavy evening import while the reported storage category rises from near-zero midday output to several hundred megawatts after sunset. These are observable system conditions, not a forecast of project revenue. [S01] [S02]

The opposite side of the thesis is equally visible. Reported operating BESS reached approximately 1,084.72 MW / 2,310 MWh by 1 September 2026, compared with a Transelectrica-reported 494 MW / 913.68 MWh installed storage reference on 5 December 2025. Because the two values come through different reporting routes and may not be perfectly definition-identical, the implied growth should be treated as a directional maturity bridge rather than an audited like-for-like series. Even on that conservative interpretation, competition has changed quickly. [S03] [S04]

Financing has also moved beyond theory. EBRD disclosures cover a fully merchant 88 MW / 176 MWh Buciumi project financed with non-recourse senior debt and expected InvestEU first-loss cover; a 127 MW / 254 MWh Scornicești project using non-recourse financing and InvestEU support; and the 342 MWp Părău 2 solar project with a 150 MW / 300 MWh co-located BESS under signed project finance. IFC has signed into an Aukera portfolio designed around 0.95 GW / 1.9 GWh of two-hour standalone storage. These transactions prove that merchant and hybrid Romanian BESS can reach institutional financing. They do not prove that an unprotected developer model using today's balancing revenue is bankable. [S09] [S10] [S11] [S12]

Decision rule. Use current Romanian financing precedents to validate structure, not to validate revenue assumptions. Replicate the qualities lenders actually financed—mature grid position, proven two-hour architecture, institutional sponsor/integrator package, independent diligence and credit support—rather than copying headline merchant IRRs.

Evidence method

What this report can prove—and what it cannot.

The retained 90-day data is a strong current-market diagnostic. It is not enough by itself to produce a 10–15 year lender forecast. Missing hours remain missing; operational readings and OPCOM prices remain separate; no battery margin is inferred from price spreads.

Observed

PZU and system shape

Daily OPCOM Base, 08:00–17:00 arithmetic price proxy, PT15 minimum and negative interval counts; retained Transelectrica hourly system observations and daily arithmetic averages. Missing operational hours are not interpolated. [S01]

Derived

Discounts and reversals

Daylight/Base ratio, daylight discount, weekday/weekend comparisons, system exchange reversal and implied stock duration are calculations from reported values. They are labelled as derived and do not represent realised BESS cash flow.

Unknown / not manufactured

Asset-level realised economics

No public project-level dispatch P&L, complete balancing settlement ledger, degradation history, detailed optimizer split, exact connection curtailment history or contracted tolling price was identified. The report does not invent them.

Lender evidence gate. Do not size debt from this 90-day dataset alone. Require at least 24 complete months of PT15 day-ahead history, the relevant intraday history, at least 12 months of product-level balancing capacity/energy data under the applicable rule regime, and an independent forward market view. Prefer 36 months where available to capture seasonality and structural changes.

Data-quality note: the public operational register currently contains a single observation dated 26 Aug 2027, outside the report's 2026 analysis window. It is excluded rather than corrected or interpolated. The retained methodology states that hourly values are local Europe/Bucharest arithmetic averages of stored public observations and empty hours are absent. [S01]

Exhibit 01 / energy-market shape

Daylight energy traded at roughly half the all-day reference over the 90-day window.

This is the observable starting point for arbitrage. It demonstrates temporal price separation. It does not include a battery's actual charging intervals, discharge intervals, efficiency, state of charge, degradation, imbalance or market fees.

90-day PZU Base vs 08:00–17:00 daylight-price proxy

24 Jun–21 Sep 2026 / Romania / RON per MWh

Yellow = PZU Base. Context line = unweighted 08:00–17:00 proxy retained by Solar Industry Romania from OPCOM outcomes. Hover, touch or use keyboard arrows for exact values. [S01] [S02]

PZU Base 08:00–17:00 proxy

Observed: 90-day arithmetic mean Base = 777.24 RON/MWh; daylight proxy = 422.51 RON/MWh.

Interpretation: use the series to justify detailed dispatch modelling, not to substitute daily Base minus daylight for BESS gross margin.

Arbitrage underwriting rule. A project model that starts with “daily Base minus solar-window price” fails credit review. Rebuild revenue from tradeable PT15 intervals, constrained by actual PCS power, usable energy, SOC, efficiency, charge/import rights, export limits, market gate closure, degradation and optimizer strategy.

Exhibit 02 / captured-price pressure

The discount is structural enough to matter, but highly non-uniform.

The median daylight/Base ratio was 55.7%. The 25th percentile was approximately 34.4%; the 75th percentile approximately 68.1%. A storage model therefore needs a distribution of daily shapes rather than one average spread.

Daily daylight / Base ratio

Price cannibalisation is concentrated in specific days.

Ratio = 08:00–17:00 arithmetic price proxy divided by PZU Base. A ratio below 100% means the daylight proxy traded below the all-day reference. [S01]

Distribution: mean ratio 50.5%; median 55.7%; 25th percentile 34.4%; 10th percentile approximately 13.6%.

Decision use: calibrate charge opportunities by regime and day type. Do not assume the mean discount occurs every day or is fully capturable.

Weekdays / 64 days 38.5%

Mean daylight discount. Mean Base 850.47 RON/MWh; mean daylight proxy 531.02 RON/MWh. No negative PT15 interval was recorded in the retained 64 weekday sample.

Weekends / 26 days 76.5%

Mean daylight discount. Mean Base 597.01 RON/MWh; mean daylight proxy 155.43 RON/MWh. All 151 negative PT15 intervals in the 90-day record occurred on weekends.

Optimizer rule. Weekend and weekday operation must be modelled separately. A dispatch engine that uses one average daily charge/discharge pattern will systematically misstate both opportunity and cycling.

Exhibit 03 / negative prices

Negative prices are real, episodic and currently a weekend phenomenon in the retained window.

Negative PT15 prices occurred on 10 of 90 delivery days. They totalled 151 intervals, equivalent to 1.75% of all quarter-hours. The most negative daily PT15 minimum in the retained period was −57.44 RON/MWh on 25 July.

Negative PT15 frequency

Ten days explain all 151 negative intervals.

Bars show the count of negative quarter-hours by delivery day. Tooltip also reports each day's minimum PT15 price. The available daily table does not provide the mean price of all negative intervals, so none is manufactured. [S01]

Date	Negative PT15	Daily minimum PT15	Daylight proxy	PZU Base
28 Jun	4	−0.01 RON/MWh	84.97	500.93
4 Jul	17	−1.07	18.00	397.11
5 Jul	19	−0.08	8.53	381.53
12 Jul	22	−1.10	32.47	419.73
25 Jul	27	−57.44	−12.30	469.27
26 Jul	13	−1.01	43.16	471.56
9 Aug	11	−0.16	63.17	500.05
30 Aug	6	−0.04	89.28	656.28
6 Sep	16	−6.45	58.43	596.34
20 Sep	16	−2.47	70.18	731.27

Observed: 10 negative-price days / 90; 151 negative PT15 intervals / 8,640.

Interpretation: negative prices improve selected charging opportunities but are too episodic to be a standalone debt thesis.

Revenue rule. Do not capitalise “negative-price charging” as a fixed annual revenue stream. Model it as part of the empirical dispatch distribution. A project must remain financeable if the incidence of negative prices changes while peak spreads also compress.

Exhibits 04–05 / physical system shape

Romania can export at noon and still require heavy imports after sunset.

This is the system condition a battery can monetize. It also explains why increasing storage penetration should progressively reduce the very price and balancing dislocations that justified early projects.

16 Sep / 12:00 to 20:00 3,248 MW

Exchange reversal from 1,466 MW net export at noon to 1,782 MW net import at 20:00. Reported storage moved from 0 MW to 475 MW. [S01]

17 Sep / 12:00 to 20:00 3,165 MW

Exchange reversal from 1,626 MW export to 1,539 MW import. Solar fell from 3,284 MW to zero; storage rose from 1 MW to 495 MW. [S01]

20 Sep / 12:00 to 20:00 1,755 MW

Exchange reversal from 274 MW export to 1,481 MW import. Reported storage increased from 0 MW at noon to 538 MW at 20:00. [S01]

Representative operating day / 17 Sep 2026

Storage discharge appears as solar disappears and imports accelerate.

Hourly arithmetic averages of retained Transelectrica observations. Solar, exchange and the source's storage category are simultaneous system readings; the chart is not a causal decomposition or asset-level dispatch record. [S01]

Solar MW Exchange MW / positive import Storage MW

12:00: 3,284 MW solar; 1,626 MW net export; 1 MW reported storage.

20:00: zero solar; 1,539 MW net import; 495 MW reported storage.

Saturation rule. Do not forecast today's evening scarcity unchanged. Every additional battery that charges during surplus hours and discharges during the evening ramp tends, all else equal, to raise the low-price charging interval and reduce the high-price discharge interval. Model that endogenous compression explicitly.

Representative day regimes

Solar output and price are related through system conditions, not a one-variable rule.

The retained daily observations provide useful examples, but they do not establish solar as the sole cause of any price. Wind, hydro, thermal availability, cross-border flows, demand and outages move simultaneously.

Day	Regime	Avg solar	Solar peak	PZU Base	Daylight proxy	Ratio	Avg exchange
14 Aug	High solar / discounted daylight	1,923 MW	3,090 MW	748.94	435.36	58.1%	−141 MW export
9 Sep	High solar / discounted daylight	1,606 MW	3,393 MW	854.28	456.14	53.4%	−217 MW export
16 Aug	High solar / discounted daylight	1,490 MW	2,565 MW	717.86	399.62	55.7%	−372 MW export
12 Sep	Low solar / import-heavy	386 MW	1,285 MW	814.35	398.19	48.9%	+1,227 MW import
18 Sep	Low solar / high Base	558 MW	2,273 MW	928.91	474.47	51.1%	+1,412 MW import
6 Sep	Low daylight price / moderate solar	623 MW	2,302 MW	596.34	58.43	9.8%	+511 MW import

Forecast rule. Do not use solar generation alone to forecast BESS spreads. The independent market model must include load, wind, hydro availability, thermal/nuclear outages, cross-border capacity and the storage fleet itself.

Operating stock and saturation

The competitive fleet is scaling faster than a static revenue model can tolerate.

Public evidence spans several maturity classes. The categories below are deliberately not summed because national operating stock, project financings, authorisations and announced pipelines overlap.

05 Dec 2025 / official stock reference 494 MW

Transelectrica reported 30 storage groups and 913.68 MWh installed energy capacity. Implied nameplate E/P ratio: 1.85h. [S03]

01 Sep 2026 / reported operator-attributed stock 1,084.72 MW

Secondary reporting attributed to Transelectrica gives 2,310 MWh. Implied nameplate E/P ratio: 2.13h. [S04]

2026 standalone support scheme ≥2,174 MWh

€150m Modernisation Fund-backed competitive grant scheme for new standalone BESS. This is a policy target, not operating capacity. [S08]

Connection / construction pipeline 105 projects

ANRE reported 105 projects with storage among projects holding connection contracts and building permits. Their 7,385.9 MW figure is total approved project export capacity, not BESS MW. [S05]

Competition rule. A revenue forecast that uses historical balancing or arbitrage values without an explicit new-entry assumption fails underwriting. At minimum, run one case where operating storage power and energy materially increase during the debt tenor and merchant gross margin is reduced by 50%.

Exhibit 06 / duration precedent

Two hours is the current financing reference; four hours needs its own value proof.

Duration is not a technology preference. It is a capital-allocation decision. The relevant question is whether each incremental MWh beyond the first two hours earns enough incremental lifecycle cash flow to pay for itself under downside conditions.

Project	Structure	Power	Energy	Nameplate duration	Maturity / finance evidence
Mireasa / Monsson	Wind + PV + BESS hybrid	6 MW	24 MWh	4.0h	Commissioned Apr 2024; issuer-reported operating asset. [S15]
Florești / Nova Power & Gas	BESS facility	200 MW	400 MWh	2.0h	Commercial operation announced Dec 2025. [S13]
Gura Ialomiței phase 1 / Aukera	Standalone	150 MW	300 MWh	2.0h	Commissioned Jun 2026. [S12] [S14]
Gura Ialomiței phase 2 / Aukera	Standalone	100 MW	200 MWh	2.0h	Debt signed within IFC portfolio; construction-stage. [S12]
Scornicești / R.Power + Eiffel	Standalone / partially merchant	127 MW	254 MWh	2.0h	EBRD non-recourse / disbursing; InvestEU risk support. [S10]
Buciumi / Scatec	Standalone / fully merchant	88 MW	176 MWh	2.0h	EBRD approved non-recourse senior debt; expected InvestEU cover. [S09]
Părău 2 / Econergy	342 MWp PV + BESS	150 MW	300 MWh	2.0h	Signed club project finance; InvestEU-supported merchant exposure. [S11]
Călan / ENGIE	Standalone project	54 MW	216 MWh	4.0h	Ready-to-build acquisition; COD planned Jan 2029. [S16]

2 hours Default

Use when the stack is power-led: evening arbitrage, reserve/balancing optionality and one principal energy shift.

It has the clearest large-scale Romanian financing precedent. Reject a longer configuration unless incremental duration is separately valued.

3 hours Prove hour 3

Use when the third discharge hour repeatedly clears above its lifecycle marginal cost or when a connection/export constraint makes longer shifting valuable.

No equally strong public Romanian project-finance precedent was identified for a 3h configuration. Require asset-specific modelling.

4 hours Duty-cycle led

Use for longer solar shifting, prolonged evening coverage, a contractual discharge window or a binding grid-export constraint.

Romania has a small operating 4h hybrid precedent and a large 4h ready-to-build project, but 4h should not be justified by “future flexibility” alone.

Longer than 4h Contract first

Use only when long-duration output is explicitly monetised through contract, curtailment recovery or a demonstrable grid requirement.

Do not extrapolate a system-planning need for long-duration flexibility into merchant lithium-ion bankability without a revenue mechanism.

Incremental-duration approval test:
NPV [ incremental discharge value + curtailment recovery + contracted grid/availability value ]
> NPV [ incremental battery CAPEX + extra degradation + augmentation + losses + financing + O&M ]
under the downside case, not only under the optimiser's base case.

Design rule. Start at 2h. Add hour 3 only if hour 3 pays for itself. Add hour 4 only if hour 4 pays for itself. Reject “future-proofing” as a standalone investment rationale.

Exhibit 07 / maturity ladder

Pipeline MW is not competition until it crosses a maturity threshold—but ignoring it is equally wrong.

The correct competition model is a maturity-weighted entry schedule. Commissioned capacity competes today. Construction-stage and financed assets deserve high entry weight. Authorised or connection-stage assets deserve lower weight. Announced portfolios are scenarios, not operating stock.

Maturity	Evidence	Underwriting treatment	Do not do
Commissioned / operating	Reported national stock 1,084.72 MW / 2,310 MWh on 1 Sep; operating examples include Florești and Gura phase 1. [S04] [S13]	Include 100% in present competitive supply.	Do not call authorised or financed MW operating.
Under construction / financed	Gura phase 2; Scornicești; signed Părău 2; ENGIE reports 85 MW / 170 MWh already under construction in Romania. [S10] [S11] [S12] [S16]	Include as near-term entry with project-specific COD and delay sensitivity.	Do not assume every financed asset commissions exactly on schedule.
Approved / credit process	Buciumi 88 MW / 176 MWh approved by EBRD as a fully merchant standalone project. [S09]	Include in strong-competition case; move to base entry schedule when financing is signed/conditions precedent are met.	Do not treat board approval as commercial operation.
Authorised	ANRE approved 596 MW of storage establishment authorisations in one February 2026 committee session, including a 534 MW storage component alongside a 550 MW PV project. [S18]	Track individually; include only with construction/financing bridge.	Do not aggregate authorisation MW directly into operating stock.
Connection + building permits	105 projects reported with storage; total approved project export 7,385.9 MW, of which 35 projects / 1,764.42 MW total project export were reported with 2026 COD plans. [S05]	Use as saturation boundary and node-by-node competing queue.	The MW is whole-project approved export, not storage MW.
Corporate / announced	Electrica reported ATRs for 17 BESS projects totalling 700 MWh; Aukera/IFC references a multi-asset portfolio up to 0.95 GW / 1.9 GWh. [S12] [S19]	Use in strategic competition scenarios after checking overlap with financed projects.	Do not sum portfolios, project lists and national queues without de-duplication.

Competition-model rule. Maintain a project-level maturity ledger. Update it at every investment-committee refresh. If the revenue consultant's supply forecast does not reconcile to this ledger, do not use the consultant's merchant forecast for debt sizing.

Exhibit 08 / revenue stack

Different revenues deserve different credit treatment.

Bankability improves when revenue sources are separated by observable market depth, contractual enforceability, technical compatibility and saturation exposure. Stacking revenues is not additive if the same MW, MWh or SOC is committed twice.

A / Energy arbitrage

Base case, but only after dispatch modelling.

Observable evidence: 90-day PZU record shows mean daylight discount of 49.5%, weekend discount of 76.5%, negative PT15 intervals and repeated evening system scarcity. [S01] [S02]
Bankable: Conservative net energy contribution derived from tradeable PT15 DA/ID prices after AC-to-AC efficiency, degradation, SOC, market fees, imbalance and connection constraints.
Upside: Exceptional negative-price charging, unusually large evening peaks, intraday volatility and optimiser alpha above independently modelled capture.
Saturation risk: High. New batteries directly arbitrage away the same low/high spread.
Downside: At minimum run −25% and −50% net-arbitrage gross-margin sensitivities independently from balancing revenue.
Increase contribution when: Two or more years of quarter-hourly back-cast, current forward curves and out-of-sample optimiser results converge.

B / Balancing and ancillary

Upside until market-depth evidence survives saturation.

Observable evidence: Romanian BESS financing disclosures expect participation in balancing/ancillary markets; Transelectrica runs balancing-capacity tenders. Demand response delivered a first 3 MW upward mFRR activation on 1 Sep. [S04] [S09] [S10]
Bankable: Only the portion supported by product qualification, executable market access, historical prices/volumes/activation and a saturation model accepted by the independent market adviser.
Upside: Any amount that assumes today's clearing prices, activation frequency or BESS share persists as capacity entry accelerates.
Saturation risk: Very high because balancing pools are finite and storage, hydro, generation and controllable demand can compete.
Downside: Set balancing contribution to zero for lender sizing if product-level evidence, qualification and optimiser back-test are unavailable; otherwise run at least a 50% revenue haircut.
Increase contribution when: Delivered asset or comparable portfolio settlements demonstrate repeatable margin after opportunity cost and SOC reservation.

C / Co-located solar optimisation

Base only where the avoided loss is measurable.

Observable evidence: Daylight discounts and negative intervals show captured-price pressure; Părău 2 demonstrates lender willingness to finance a 342 MWp PV + 150 MW / 300 MWh BESS structure. [S01] [S11]
Bankable: Measured or independently modelled recovery of otherwise curtailed, negative-value or low-value solar plus demonstrable shared grid/EPC/O&M savings.
Upside: Merchant “capture-price uplift” that depends on perfect forecasting or future negative prices.
Saturation risk: Medium to high. Storage improves solar capture but widespread storage also lifts low daylight prices and depresses evening prices.
Downside: Model zero curtailment benefit if historical or connection-study evidence does not establish curtailment. Model the common POI export cap explicitly.
Increase contribution when: Node-level studies, historical restrictions or contractual export limitations establish recurring otherwise-lost energy.

D / Grid and connection value

Zero unless monetised.

Observable evidence: ANRE's 2026 connection reforms, financial guarantees and capacity-allocation process show grid access is scarce and economically material. [S06] [S20]
Bankable: Avoided connection reinforcement cost, contracted network service, measurable curtailment reduction or additional saleable generation under a legally secure shared connection.
Upside: Generic claims that a battery “helps the grid” or may receive future congestion revenue.
Saturation risk: Node-specific. Competing projects may consume headroom before COD.
Downside: Value at zero unless a contract, connection condition or quantified avoided cost makes the benefit cash-real.
Increase contribution when: The connection study and legal documents demonstrate a direct economic saving attributable to the BESS.

E / Contracted availability or tolling

Strongest debt base when properly matched to the battery.

Observable evidence: No sufficiently transparent public Romanian standalone tolling price precedent was identified in the reviewed evidence. This route is therefore a structural underwriting framework rather than a quoted local benchmark.
Bankable: Fixed availability/capacity payment from a creditworthy counterparty, with dispatch rights, throughput limits, efficiency losses, degradation and augmentation obligations fully allocated.
Upside: Merchant sharing above the fixed floor, ancillary optimisation retained by owner or performance incentives.
Saturation risk: Lower for contracted fee during contract tenor; re-contracting risk remains.
Downside: Stress counterparty default, contract termination, excessive dispatch and merchant tail.
Increase contribution when: Contract tenor covers debt tenor, payment is not easily avoidable and warranty throughput matches contractual dispatch rights.

F / Demand response

Model it first as a competitor.

Observable evidence: Transelectrica activated an all-controllable-load balancing group for upward mFRR on 1 Sep, demonstrating that flexible consumption can compete for flexibility revenue. [S04]
Bankable: For a standalone BESS, none unless the project has a contracted aggregation or load-flexibility arrangement. For a C&I hybrid, contracted load reduction can complement a smaller battery.
Upside: Aggregation fees, shared optimisation or avoided battery sizing where controllable load can deliver part of the flexibility requirement.
Saturation risk: It is itself a saturation mechanism for balancing prices.
Downside: Assume controllable load takes a growing share of suitable upward products.
Increase contribution when: A site has a measured, repeatable and contractually available flexible-load profile with acceptable process cost.

Revenue-concentration rule. Do not permit a lender base case to depend on one merchant revenue pool unless that pool alone survives the downside debt-service test. “Stacking” balancing and arbitrage does not diversify risk if both collapse when more storage enters.

Merchant economics

Gross spread must be converted into net dispatch contribution before it enters valuation.

The single most common modelling error is to treat observable price dispersion as a cash margin. Charging volume, round-trip losses, opportunity costs and battery wear sit between the two.

Net merchant contribution =
discharge revenue
+ contracted availability / activation revenue
+ measurable curtailment or connection savings
− charging-energy cost
− round-trip conversion loss
− market / imbalance / optimiser fees
− variable O&M
− degradation economic cost
− augmentation reserve
− revenue lost because SOC or MW is reserved for another service.

Minimum discharge price for an energy cycle ≈
charge price ÷ guaranteed AC-to-AC round-trip efficiency
+ variable market cost per discharged MWh
+ degradation / throughput cost per MWh
+ opportunity value of reserved capacity.

Use AC-to-AC, point-of-connection economics.

Do not use cell efficiency, DC rack efficiency or a marketing “system efficiency” figure if the financial model settles at the grid meter. Require a guaranteed AC-to-AC efficiency curve including PCS, transformer, HVAC and auxiliary loads across representative SOC, ambient-temperature and power conditions.

Underwrite to the guaranteed floor, not the brochure midpoint. If the OEM or integrator will not guarantee the value under the project's actual dispatch envelope, the financial model must use the lower independently validated value.

Degradation is a dispatch cost, not a footnote.

Flat “2% per year” assumptions are inadequate. Require a model combining calendar ageing, cycle depth, C-rate, temperature, average SOC and throughput. Map that model to the dispatch simulation and warranty.

Augmentation must appear as dated capex, planned outage and warranty-reset logic. If the base case depends on augmentation but no procurement strategy, price allowance, space or electrical capacity exists, reject the base case.

Economic gate. Reject any model in which the battery dispatch assumptions exceed the OEM's warranted throughput, temperature, SOC or cycle envelope, or where the model uses a higher AC-to-AC efficiency than the contracted warranty floor.

Standalone versus co-location

Solar co-location is valuable when it solves a real constraint, not because solar and batteries look complementary.

Co-location can reduce connection duplication and shift low-value solar. It can also trap the BESS behind a shared export limit, restrict grid charging, create common-mode outages and force the battery to cycle when the merchant optimum would not.

Co-location materially improves economics when

There is value to recover.

Approve the co-location premium only where at least one of the following is quantified: recurring solar curtailment; negative/very low captured prices; shared connection capex; an export-right benefit; a grant/contract that rewards stored renewable energy; or a common O&M/EPC saving that survives interface risk.

Co-location does not automatically improve economics when

The POI becomes the bottleneck.

Do not assume value where solar already consumes the common export capacity during valuable discharge periods, the battery cannot grid-charge, charging origin is restricted, solar has little curtailment, or common transformers/substations create correlated outage risk.

Issue	Standalone BESS	Solar + BESS	Underwriting rule
Grid charging	Usually central to merchant flexibility.	May be unrestricted, limited or programme-dependent.	Do not assume grid charging from export rights; verify import/charging right explicitly.
Export capacity	Dedicated export can maximise dispatch freedom.	Shared POI may cap combined solar + BESS output.	Simulate net POI flow, not separate nameplate MW.
Charge energy	Buys market energy.	Can absorb low-value/curtailed solar.	Value curtailed energy at its real opportunity price, not at PZU Base.
Infrastructure	Separate substation/connection may cost more.	Potential shared transformer, land, control and O&M.	Count savings only after interface and augmentation allowances.
Availability	Fewer common-mode renewable interfaces.	Shared switchgear/POI may create correlated outage.	Model common-mode outage in availability guarantee.
Finance precedent	Buciumi, Scornicești, Aukera portfolio. [S09][S10][S12]	Părău 2 provides signed large-scale precedent. [S11]	Both structures are financeable; choose based on actual duty cycle and connection economics.

Co-location rule. Choose solar + BESS over standalone only if the NPV of quantified shared-grid, curtailment and captured-price benefits exceeds the NPV of POI congestion, operational coupling, charging restrictions and common-mode risk under the downside case.

Exhibit 09 / bankability gates

A strong revenue case cannot cure weak grid rights or an unfinanceable technical package.

The gates below are conditions precedent to an investment decision. “Expected”, “under discussion” and “to be clarified after FID” are not equivalent to satisfied.

Gate	Minimum evidence	Approval standard	Reject / defer when
Grid connection rights	Final ATR/connection agreement; node, voltage, works, milestones, securities.	Rights are valid, transferable/financeable where required and timetable matches COD.	Critical reinforcement, auction or third-party dependency is unfunded or outside base schedule.
Charging rights	Explicit approved import/consumption power and operating conditions.	Modelled grid charging is legally and technically permitted.	Model assumes bidirectional charging that documents do not grant.
Export limits	Firm MW at POI by operating state.	Financial model uses net POI cap, including co-located generation.	Battery nameplate is modelled as exportable when POI cannot deliver it.
Commissioning status	Construction schedule, energisation route, grid testing, ANRE licence path.	Clear critical path with LDs and appropriate long-stop.	Revenue start precedes realistic grid acceptance or qualification.
EPC / integration	Battery, PCS, transformer, EMS/BMS/PPC/SCADA interface matrix.	One accountable wrapper or bankable interface regime with matching LDs.	Major performance gaps sit between OEM and EPC scopes.
Cell / system warranty	Capacity, availability, defects, operating envelope.	Warranty duration and remedy align with financing tenor and dispatch.	Warranted operating conditions conflict with financial model.
Cycle / throughput warranty	MWh throughput, DoD, C-rate, SOC and temperature conditions.	Base and upside dispatch are inside warranty or incremental wear is priced.	Revenue model implicitly consumes more lifetime throughput than purchased.
Degradation	Independent calendar + cycling curve.	Dispatch-linked degradation reconciles annually to usable MWh.	Flat annual degradation assumption has no link to duty cycle.
Augmentation	Timing, capex, space, outages, procurement and warranty treatment.	Funded reserve or contracted solution preserves minimum capacity.	Augmentation is required to meet model but absent from capex/cash flow.
Round-trip efficiency	Guaranteed AC-to-AC efficiency at POI across relevant duty envelope.	Underwriting uses warranty floor including auxiliaries.	Only cell/DC efficiency is provided.
Thermal management	Design temperatures, HVAC redundancy, derating curves, alarms.	Output and warranty hold under site climate and dispatch profile.	Summer derating materially reduces modeled revenue without allowance.
Fire protection	Propagation testing, detection/suppression, separation, emergency plan, authority/insurer review.	Independent engineer and insurer accept design and emergency response.	Thermal-runaway containment or emergency-water/runoff strategy is unresolved.
EMS / BMS / PCS integration	Functional specification and responsibility matrix.	End-to-end setpoint, SOC, limits and fail-safe behaviour tested.	Optimizer cannot reliably control the asset within warranty envelope.
Telemetry / qualification	Market telemetry, metering, communications, response testing and product qualification plan.	Qualification precedes revenue assumption.	Balancing revenue begins before technical qualification in the model.
Maintenance / LTSA	Response times, spares, availability guarantee, software support, cybersecurity.	Remedies cover material lost availability.	Critical spares or software support end well before debt tenor.
Insurance	Indicative terms before FID; property, BI, liability and battery-specific exclusions reviewed.	Lender engineer and insurer agree protection design.	Uninsurable thermal-runaway/common-mode risk or prohibitive deductibles remain unresolved.
Lender technical DD	Independent engineer report covering design, grid, performance, fire, schedule and capex.	No material qualification remains open at financial close.	Core performance depends on sponsor-only assumptions.
Revenue concentration	Separate energy, balancing, contracted and solar-optimisation models.	No double-counting of MW/MWh/SOC across products.	Two revenues assume simultaneous use of the same battery capacity.
Merchant exposure	Independent forecast + back-cast + downside.	Debt survives lender-approved merchant stress.	Debt service requires unhaircut merchant consultant base case.
Balancing saturation	Market size, competing eligible MW, product rules, entrant schedule.	Material capacity entry is included.	Model holds current balancing economics constant while BESS pipeline expands.
Refinancing / debt service	Contracted debt terms, DSCR/LLCR, reserves, merchant tail and refi sensitivity.	Project meets lender-agreed covenants without relying on refinancing at a better rate.	No specified debt-service hurdle exists but leverage is presented as “bankable”.

Approval standard. Grid import/export rights, technical performance guarantees, safety/insurance acceptance and the revenue model are hard gates. Do not trade one off against another through a higher equity return.

Grid and regulatory structure

Grid rights have become a balance-sheet item.

ANRE's 2026 reforms increase the cost of speculative grid positions and explicitly apply connection rules to standalone storage. This changes the value of a mature project: a real connection position has scarcity value, while a weak ATR can consume capital without producing financeable rights.

Connection security 20%

ANRE raised the relevant financial guarantee to 20% of the connection tariff in the 2026 reform. The exact applicability must be checked against project circumstances. [S06]

2026 capacity-allocation participation guarantee €20k/MW

For the 2026 allocation procedure, the guarantee is €20,000 per MW of capacity requested. Grid development therefore consumes material development capital before COD. [S06] [S20]

Establishment-authorisation security €30/kW

ANRE's May reform introduced a €30/kW financial guarantee linked to establishment authorisation, executable in defined non-completion or withdrawal cases. [S06]

Separately, ANRE Order 56/2025 established methodological rules for exemption from regulated network tariffs for electricity stored after being drawn from the grid. That improves storage economics but does not create a charging right. The connection documents must still authorise the import power assumed by the optimizer. [S07]

ANRE's March 2026 project register is also a reminder that connection maturity and storage capacity are different metrics. The regulator reports 105 generation projects with storage among projects holding connection contracts and building permits, but the associated 7,385.9 MW is the projects' approved maximum export, not the battery power. Converting that number into “7.4 GW BESS pipeline” would be wrong. [S05]

Grid rejection rule. Reject the project if the financial case assumes a charging MW, discharging MW or operating schedule not unambiguously supported by the connection documents and network studies. Legal ambiguity over bidirectional use is not an acceptable merchant risk.

Project structures

Downside resilience is driven more by revenue and grid structure than by “standalone versus hybrid”.

Romania now has financing evidence for both standalone and co-located BESS. The strongest structures combine mature rights with a revenue floor or credit enhancement; the weakest combine speculative grid access with concentrated merchant ancillary exposure.

Structure	Strength	Primary downside	Decision rule
Standalone / 2h / partially contracted	Highest dispatch freedom plus a debt-service floor.	Contract counterparty and re-contracting risk.	Preferred where availability/tolling terms match throughput warranty.
Standalone / 2h / merchant with risk support	Clear Romanian financing precedent at Buciumi/Scornicești/Aukera.	Revenue compression; dependence on public risk-sharing in some precedents.	Finance only with independent saturation case and conservative leverage.
Solar + BESS / 2–4h	Potential shared grid and curtailment recovery.	Shared export cap and common-mode outage.	Use when node-specific co-location savings exceed coupling penalty.
Merchant / 3–4h	More energy-shift optionality.	Third/fourth MWh may earn insufficient incremental margin.	Require marginal-hour economics; do not rely on average project IRR.
Long-duration lithium / merchant	Longer scarcity coverage.	High idle capital if scarcity window remains short.	Reject without contract or robust evidence for incremental duration.
Subsidy-supported	Lower net capex and potentially stronger debt metrics.	Eligibility, delivery deadlines, clawback and operating-condition risk.	Count grant only after award and conditions are satisfied; project must still have a viable operating revenue model.
Fully merchant without risk support	Maximum upside and operational freedom.	Highest debt-service volatility.	Use lower leverage and require severe dual-revenue downside coverage.
Ancillary-heavy / grid constrained	Potential high early revenue per MW.	Balancing saturation plus inability to pivot into energy.	Weakest structure; reject if energy route cannot support a downside floor.

Structure rule. The strongest project is not necessarily co-located or standalone. It is the project with the widest executable operating envelope and the smallest dependence on one merchant revenue pool.

Financing evidence

Romania has project-finance precedent—but credit enhancement remains economically important.

Institutional financing confirms that storage can support non-recourse structures. The details also show why developers should not cite “project finance exists” as proof that any merchant model is debt-ready.

Buciumi / Scatec

Fully merchant can be financed.

EBRD describes 88 MW / 176 MWh Buciumi as a fully merchant standalone BESS to be financed through non-recourse senior debt, expected to benefit from an InvestEU first-loss risk-cover guarantee. The guarantee is not incidental; EBRD explicitly links it to the project's risk profile. [S09]

Scornicești / R.Power + Eiffel

Merchant risk is financeable when structured.

127 MW / 254 MWh; non-recourse senior debt and bridge facility, partially merchant, with InvestEU first-loss support. EBRD notes an optimisation arrangement with GEN-I for merchant revenues. [S10]

Părău 2 / Econergy

Co-location has institutional precedent.

342 MWp solar + 150 MW / 300 MWh BESS; signed club financing. EBRD's facility is part of an aggregate debt package up to €229m and benefits from InvestEU support because the project has substantial merchant exposure. [S11]

Aukera portfolio

Two-hour standalone is becoming a portfolio product.

IFC describes five standalone BESS assets designed for two-hour storage, up to 0.95 GW / 1.9 GWh, with an IFC loan up to €136m within up to €336m of total debt financing. [S12]

Leverage rule. Do not derive leverage from another project's debt amount. Use transaction precedents only to validate that the structure is financeable. Size project debt from the subject project's contracted rights, technical package, independent revenue downside and lender-specific DSCR/LLCR requirements.

Exhibit 10 / downside sensitivity matrix

A bankable BESS must survive more than one bad thing at a time.

No artificial scenario probabilities are assigned. The stresses are credit tests, not forecasts. Exact DSCR pass levels must come from the proposed lender term sheet; if no debt terms exist, leverage should not be presented as bankable.

Scenario	Stress definition	Main transmission to cash flow	Required action
1 / Base underwriting	Independent PT15 dispatch model; conservative entry schedule; warranty-floor RTE; contracted capex/O&M; no unproven grid value.	Reference debt case.	Use only after reconciliation to at least 24 months DA history, relevant ID history and product-level balancing evidence.
2 / Spread compression	Reduce net energy-arbitrage gross margin by 50% versus independent base model.	Direct EBITDA reduction; fewer economic cycles; lower optimiser value.	Project must retain lender-required debt coverage or reduce leverage before approval.
3 / Ancillary saturation	Reduce balancing/ancillary gross revenue by at least 50%; separately test zero balancing contribution for sizing.	Power-led 2h projects lose high-margin services; SOC opportunity cost falls but does not fully compensate.	If debt service fails with balancing at zero, classify project as balancing-dependent and require contractual floor or lower debt.
4 / High-renewables / low-daylight	More frequent low daylight prices but faster BESS entry also compresses the evening premium.	Charging improves while discharge value and balancing scarcity can decline.	Do not model lower charge prices without an endogenous lower discharge spread.
5 / Grid-constrained	Apply actual POI export cap, charging limit, curtailment instructions and network-outage/commissioning delays.	Battery cannot access the modeled highest-value intervals; co-located solar may consume export headroom.	Reject if returns rely on MW that cannot physically clear the POI.
6 / Strong competition	Simultaneously reduce energy-arbitrage and balancing gross margin by 50%, while retaining contracted costs and degradation.	Tests whether project value is structural or an early-mover rent.	Use as investment-committee resilience case; if equity value becomes dependent on refinancing or terminal value, resize or reject.

Debt-service rule. A project that passes energy-spread compression but fails balancing saturation is not diversified. A project that passes each stress individually but fails the combined strong-competition case requires lower leverage, a contracted floor or additional sponsor support.

Technical design

The optimiser cannot earn a revenue stream the hardware is not warranted to deliver.

Battery bankability is a contract-alignment exercise across cells, racks, PCS, EMS, BMS, transformer, SCADA, market telemetry, warranty, EPC, O&M and insurance.

Warranty must follow duty cycle.

Require a single annual operating envelope that reconciles usable energy, minimum and maximum SOC, C-rate, temperature, throughput, cycles, availability, response time, efficiency and degradation. If the optimiser's day-ahead/balancing stack operates outside that envelope, the higher revenue case is not bankable.

A “two cycles per day” marketing statement is not enough. Ask how a cycle is defined, how partial cycles are counted, what throughput ceiling applies, how high C-rate balancing activations affect warranty and what remedy applies when usable capacity falls below guarantee.

Integration risk belongs with a solvent party.

The project company should not become the de facto systems integrator because the battery OEM, PCS supplier and EPC contractor each exclude the other's interface. Require one bankable responsibility matrix and back-to-back performance tests for setpoint response, metering, SOC accuracy, fail-safe operation and remote-control availability.

Liquidated damages must map to the financial model's principal revenue drivers: COD, power, usable energy, availability and efficiency. A nominal EPC LD cap is weak protection if the project's actual loss is prolonged merchant unavailability.

Procurement rule. Do not select the lowest €/kWh offer until bids are normalised to the same usable AC MWh, guaranteed POI efficiency, warranted throughput, augmentation obligation, LTSA duration, availability guarantee and fire-engineering scope.

Safety, insurance and operability

Safety is a financing condition, not an EPC appendix.

Thermal-runaway propagation, emergency response, firewater/runoff, access, separation, ventilation, detection, suppression and cybersecurity all affect insurability and therefore debt availability.

Independent validation

Require test evidence.

Obtain cell/module/rack propagation evidence relevant to the supplied product, independent fire-engineering review and insurer acceptance before final investment approval. A manufacturer statement alone is insufficient.

Emergency design

Design for the event, not only prevention.

Confirm emergency access, isolation, responder procedure, water strategy, contaminated runoff, neighbouring equipment exposure and re-energisation protocol with the competent authorities and insurer.

Operational control

Cyber and telemetry are revenue systems.

Remote dispatch, communications redundancy, time synchronisation, meter quality, SOC accuracy and secure optimizer interfaces must be tested because an unavailable control path can remove the asset from balancing and merchant dispatch.

Safety rejection rule. Do not close financing with unresolved insurer qualifications on thermal runaway, fire separation, emergency access or business-interruption exposure. Insurance uncertainty is a project risk, not a post-COD operating item.

Minimum investment evidence

What must exist before IC or lender approval.

A serious storage approval package should allow an independent reviewer to reproduce the economics, identify each legal right and map every major revenue assumption to technical capability.

Executable grid package

Connection agreement, ATR, approved import and export MW, network works, guarantees, milestones, land/easements and commissioning sequence.

Quarter-hourly revenue model

At least 24 months historical DA data, relevant intraday data, market fees and independent forward assumptions; 90-day proxy analysis alone is insufficient.

Balancing evidence pack

At least 12 months product-level capacity/energy prices and volumes under current rules, activation statistics, qualification requirements and competitor-entry sensitivity.

Out-of-sample optimiser test

Back-test the proposed strategy against periods not used to calibrate it; show gross capture, losses, SOC violations, cycles and settlement deductions.

Independent technical model

Usable MWh, AC MW, POI efficiency, degradation and augmentation derived from the contracted system, not a generic battery.

Contract interface matrix

Identify responsibility and remedies across OEM, PCS, EMS/BMS, EPC, grid contractor, optimizer and LTSA provider.

Safety and insurance acceptance

Independent fire-engineering assessment and insurer indication with material exclusions/deductibles known before approval.

Financing case

Actual indicative debt terms, covenant levels, reserves, hedging where relevant and downside debt-service model. Do not assume leverage from precedent transactions.

Competition ledger

Operating, construction, financed, authorised and announced competitor projects separated, de-duplicated and assigned realistic COD windows.

Red-team case

Combined spread compression, balancing saturation, grid limitation and underperformance case with defined sponsor action if covenants fail.

IC gate. If the investment memo cannot show each of these items or label it as a binding condition precedent, the project is not ready for final approval.

Ten investment-committee rules

Approval should be binary on the issues that can destroy the project.

These rules translate the evidence into an investment-committee operating standard.

Do not approve without explicit bidirectional grid rights. Import/charging and export MW must match the dispatch model at the point of connection.
Use 2h as the default reference design. Require incremental NPV evidence before paying for hour 3, hour 4 or longer duration.
Underwrite energy arbitrage from PT15 dispatch, never headline daily spreads.
Treat balancing as upside unless qualification, historical market-depth and saturation evidence support inclusion. If that evidence is absent, set balancing to zero for debt sizing.
Run independent 50% haircuts on energy and balancing revenue and a combined 50% / 50% strong-competition case. These are credit stresses, not probabilities.
Do not count pipeline as operating capacity. Maintain a maturity ledger and de-duplicate project portfolios.
Approve co-location only when quantified curtailment, connection or shared-infrastructure value exceeds the shared-POI penalty.
Require warranty assumptions to reconcile exactly to dispatch. No financial cycle, throughput, SOC or temperature condition may sit outside contracted warranty without a priced remedy.
Do not approve unresolved fire, insurance or integration interfaces. They are capable of making an otherwise attractive return unfinanceable.
Do not rely on refinancing, terminal-value expansion or higher future merchant prices to restore debt coverage. The project must survive on contracted and stressed operating cash flows.

Ten lender due-diligence requirements

Credit approval requires reproducibility, not sponsor confidence.

Each item should be capable of becoming a condition precedent, covenant, reserve mechanism or lender-engineer test.

Legal opinion on connection and charging/export rights, including transferability, security package and curtailment obligations.
Independent market report using quarter-hourly data, entry/saturation assumptions and separate energy versus balancing revenue.
Independent technical report validating AC power, usable MWh, RTE, degradation, augmentation, thermal limits and project availability.
Fully reconciled EPC/OEM interface matrix with completion, performance and delay remedies sufficient for lender exposure.
Warranty-throughput reconciliation showing every modelled cycle and service remains within guaranteed operating limits.
Fire-engineering and insurer review completed before first draw for major equipment, with no material open qualifications.
Market qualification and telemetry plan with testing milestones prior to any balancing revenue appearing in the debt case.
Lifecycle augmentation reserve or contracted augmentation package sufficient to preserve the minimum required energy capacity.
Merchant downside covenant test including energy −50%, balancing −50%, combined competition, grid restriction and delayed COD.
Counterparty and optimiser diligence covering credit, termination rights, algorithmic incentives, fee structure, data ownership, cyber controls and replacement rights.

Ten developer design and procurement rules

Procure a financeable operating envelope, not nominal battery capacity.

The cheapest container price can become the most expensive project if integration, augmentation or warranty exclusions move risk back to the SPV.

Specify usable AC MWh at the grid meter, not only DC nameplate energy.
Specify guaranteed AC-to-AC RTE curves including transformers, PCS, HVAC and auxiliaries at relevant ambient conditions.
Buy throughput that matches the intended stack; do not procure an energy-arbitrage warranty and later add aggressive balancing cycles without repricing degradation.
Reserve physical, electrical and software capacity for augmentation before the site layout is frozen.
Put EMS/BMS/PCS/SCADA responsibility into one testable interface matrix with one party accountable for end-to-end performance.
Design the thermal system for Romanian summer conditions and high cycling, including redundancy and derating guarantees.
Complete fire-propagation and emergency-response design before equipment award, not after containers are ordered.
Procure communications redundancy and accurate SOC/metering as revenue-critical equipment.
Model the actual POI export/import cap during equipment sizing; do not buy MW the grid connection cannot monetize.
Compare OEM bids on lifecycle €/delivered warranted MWh, not initial €/kWh alone.

Five project rejection triggers

Conditions that should stop the deal.

These are not “risks to monitor”. They are reasons to withhold final approval until cured.

Reject 01

Ambiguous grid charging or export right

If the model requires bidirectional MW the legal/grid package does not clearly provide, reject or re-size to the documented right.

Reject 02

Debt depends on unhaircut balancing revenue

If debt coverage fails when balancing contribution is removed or materially reduced and no contractual floor exists, reject the leverage structure.

Reject 03

Dispatch exceeds warranty

If expected throughput, temperature, C-rate, SOC or degradation exceeds the contracted warranty envelope without funded augmentation, reject the technical/economic case.

Reject 04

Unresolved integration or fire/insurance issue

If material interface obligations are unallocated or insurer acceptance is conditional on redesign, do not reach FID.

Reject 05

Strong-competition case destroys debt service

If simultaneous 50% energy and balancing gross-margin stresses breach lender requirements and the project has no contractual cure, reduce debt or reject.

Five conditions that materially improve bankability

The best improvements reduce dependency on a single forecast.

Each condition either hardens revenue, widens the executable operating envelope or transfers a risk to a party that can actually manage it.

Contracted availability floor

A bankable fixed payment with tenor, credit and throughput rights aligned to debt materially improves debt sizing.

Firm bidirectional grid capacity

Unrestricted or clearly defined charging/export rights allow the optimiser to rotate between energy and balancing as markets evolve.

Standardised 2h architecture

A two-hour design with institutional EPC/OEM/LTSA precedent reduces technology, construction and revenue-risk complexity.

Measurable co-location value

Verified curtailment, shared connection cost or low-value solar can create a project-specific value stream less dependent on national balancing prices.

Grant or first-loss support after award

Modernisation Fund grants or InvestEU-style risk sharing can improve resilience, but count them only once eligibility, award and conditions are documented.

Red-team analysis

The strongest case against Romanian BESS is that successful batteries destroy early-mover rents.

The report's central recommendation fails if future storage entry is slower than expected while renewable penetration continues to deepen daylight discounts and evening scarcity. In that world, merchant BESS could outperform the conservative structure recommended here.

What would make the conservative underwriting wrong?

First: storage deployment could lag authorised pipelines because grid security, financing, equipment, development guarantees or project attrition materially slow commissioning. If operating MW grows slowly while solar expands quickly, arbitrage and balancing spreads could persist longer.

Second: system stress can grow faster than storage. Large renewable additions, thermal/nuclear outages, hydrological limits, cross-border constraints or load growth can increase the need for fast flexibility even as BESS capacity rises.

Third: market design can open new revenue pools. Faster integration into European balancing platforms, new congestion products or contracted flexibility could offset erosion in existing services.

Fourth: optimizer sophistication can create value from intraday and multi-market rotation not visible in the simple PZU diagnostic.

These possibilities argue for preserving operational optionality. They do not justify using unproven future revenue to size today's debt.

What would change the conclusion. Increase merchant revenue credit only after operating-entry data show saturation is slower than assumed, realised comparable-project settlements outperform the stressed model, or durable contracted/new market products create a demonstrable additional pool. Reduce it further if operating stock accelerates, balancing prices compress or weekend spreads narrow materially.

One-page financing summary

What we would finance / what we would not finance.

This is a project-structure screen, not a securities recommendation. It states the evidence threshold a Romanian BESS should meet before the investment case is presented as financeable.

What we would finance

Rights first. Revenue second.

A 2h standalone BESS with final bidirectional grid rights, executable construction package and independent PT15 revenue model.
A partially contracted BESS where fixed availability/tolling income covers a meaningful share of debt requirements and merchant optimisation is upside.
A fully merchant 2h BESS with conservative leverage, institutional sponsor, independent market/technical diligence and a financing structure capable of surviving simultaneous revenue compression.
A 3–4h asset where the incremental third/fourth hour has separately demonstrated value after degradation and incremental capex.
A solar + BESS project where curtailment, captured-price loss or shared connection economics are measured and the common POI is modelled correctly.
A subsidy-supported project only after grant award and eligibility conditions are contractually integrated into construction and financing.
A project whose balancing case can be removed without invalidating the fundamental capital structure, unless a balancing revenue floor is contracted.
A system with guaranteed AC-to-AC efficiency, throughput, usable energy and availability mapped directly to the financial model.
A project with funded augmentation, lender-grade fire engineering, insurer acceptance, telemetry qualification and replaceable optimiser arrangements.
A project that still clears investment and credit hurdles after the combined strong-competition downside.

What we would not finance

A spread story with missing rights.

A project whose economics are presented as PZU Base minus daylight-price proxy.
A project that assumes today's balancing prices and activations remain unchanged while Romanian storage stock and flexible demand expand.
A 4h or longer battery justified only by “more renewables need longer storage”.
A co-located project that assumes both PV and BESS can export nameplate MW through a smaller shared connection.
A project with export rights but unproven charging/import rights.
A model using cell/DC efficiency where settlement occurs at the AC grid meter.
A model with cycles or throughput beyond warranty and no degradation price or augmentation reserve.
A project whose EPC, OEM, PCS and EMS contracts leave performance interfaces with the SPV.
A project with unresolved insurer/fire-engineering conditions at FID.
A capital structure that only works if refinancing occurs at a better rate, merchant revenue grows or terminal value rescues weak operating cash flow.

Final investment position

The Romanian BESS opportunity is financeable, but early-mover economics must not be capitalised indefinitely.

The market evidence is sufficiently strong to support storage investment and sufficiently mature to require conservative competition assumptions.

Romania's current power-market shape clearly rewards temporal flexibility: daylight prices were approximately half of PZU Base on average over the retained 90-day period, all observed negative PT15 intervals were concentrated on weekends, and multiple retained operating days shifted from midday export to substantial evening import while reported storage output rose into the evening ramp. Those observations establish the physical and market use-case. [S01]

They do not establish a perpetual merchant rent. Reported operating storage passed 1 GW and 2.3 GWh by September, new projects are being financed at scale, the Modernisation Fund is supporting at least another 2,174 MWh, large corporate portfolios hold connection rights, and controllable demand has entered balancing. Competition therefore belongs inside the base underwriting architecture, not in a remote downside appendix. [S04] [S08] [S12] [S19]

The strongest current configuration is consequently not “the BESS with the highest modelled IRR.” It is a system with enough duration to address its evidenced duty cycle, but no more; firm bidirectional grid rights; technically warranted cycling; a revenue stack that can rotate as markets change; conservative merchant credit; and a capital structure that survives substantial spread and ancillary-service compression.

Final decision rule. For a new Romanian project without a specific contracted long-duration use case, start underwriting at 2h. Put conservative PT15 energy arbitrage in the base case; add balancing only to the extent product-level evidence survives saturation; treat speculative grid services and uncontracted future products as upside; move to 3–4h only when incremental-hour economics are proven; and reject the project if grid rights, warranty, fire/insurance, integration or combined-revenue downside fail their hard gates.

Source register and limitations

Evidence remains separated by class and maturity.

Primary sources are used wherever directly available. Issuer disclosures support facts about their own projects. Secondary reporting is retained only where a directly indexed primary route was not identified and is labelled accordingly.

S01Solar Industry Romania — Data History. Retained Transelectrica operational observations and separately stored OPCOM PZU references; Europe/Bucharest hourly arithmetic averages; missing hours not interpolated.User-maintained public-source evidence
S02Solar Industry Romania — public 90-day API. Machine-readable retained market/system history used as the current analytical window.Public API / derived register
S03Transelectrica — 2025 Supervisory Board report. Storage installed power 494 MW and 913.68 MWh at 5 Dec 2025.Primary TSO
S04Solar Industry Romania — 2 Sep 2026 briefing. Reports 1,084.72 MW / 2,310 MWh operating BESS stock attributed to Transelectrica; first all-load mFRR activation; system-profile evidence.Original research / secondary operator attribution for stock
S05ANRE — project status at 1 Mar 2026. Connection contracts, building permits and projects reported with storage; project-export MW kept separate from BESS capacity.Primary regulator
S06ANRE — 21 May 2026 connection/licensing reform. Connection guarantee, capacity-allocation guarantee and establishment-authorisation financial guarantee; standalone-storage applicability.Primary regulator
S07ANRE — electricity authorisation/licensing and technical rules. Includes Order 56/2025 on network-tariff exemption for stored electricity withdrawn from the grid and licensing requirements for standalone storage.Primary regulator
S08European Commission — Romanian standalone BESS scheme. €150m direct-grant scheme targeting at least 2,174 MWh of new standalone storage.Primary EU authority
S09EBRD — Buciumi BESS. 88 MW / 176 MWh standalone fully merchant project; non-recourse senior financing; expected InvestEU first-loss cover.Primary lender
S10EBRD — Scornicești BESS. 127 MW / 254 MWh standalone project; non-recourse financing; partially merchant structure and InvestEU support.Primary lender
S11EBRD — Econergy Părău 2 PV & BESS. 342 MWp solar + 150 MW / 300 MWh BESS; signed senior secured financing and DSRA facility.Primary lender
S12IFC — Aukera BESS portfolio. Five Romanian standalone two-hour BESS projects up to 0.95 GW / 1.9 GWh; IFC and wider debt financing structure.Primary lender
S13E-INFRA / Nova Power & Gas — Florești. Company disclosure of 200 MW / 400 MWh commercial operation in Dec 2025.Issuer disclosure
S14ESS News — Aukera Gura Ialomiței phase 1. Reports commissioning of 150 MW / 300 MWh phase and planned second phase.Specialist secondary / company attribution
S15Monsson — Mireasa BESS. Company disclosure of commissioned 6 MW / 24 MWh four-hour hybrid storage and charging from wind, PV and grid.Issuer disclosure
S16ENGIE Romania — Călan acquisition. 54 MW / 216 MWh four-hour ready-to-build BESS, planned January 2029 operation.Issuer disclosure
S17OPCOM — Day-Ahead Market coupling results. Official PT15 and Base market results reference.Primary market operator
S18ANRE — 18 Feb 2026 authorisation release. 849.36 MW renewable generation and 596 MW storage authorisations approved in one committee session; authorisation is not operation.Primary regulator
S19Electrica — current reports. 30 Jun 2026 disclosure that ATRs had been obtained for 17 new BESS projects totalling 700 MWh.Issuer disclosure
S20Transelectrica — network-capacity allocation information. 2026 auction platform, capacity-allocation process and applicant registers.Primary TSO

Material limitations. The 90-day market window is summer-weighted; it is not a full seasonal history. Public asset-level realised BESS revenue was not identified. A complete long-run balancing settlement series was not reconstructed for this report. Several national stock and pipeline references use different dates and definitions. Those limitations are why the report recommends higher evidence thresholds instead of manufacturing long-run merchant returns.