ANALYSIS · REVENUE MODEL

Three Revenue Streams

Wholesale arbitrage (46%) + Capacity market (40%) + System services (13%) = EUR 149k/MW/yrEUR 7.5M/yr

EUR 149k/MW/yr (CENTRAL)EUR 7.5M/yr (CENTRAL)
Arbitrage
69k EUR/MW3.45M EUR/yr
CRM
60k EUR/MW3.0M EUR/yr
DASSA
20k EUR/MW1.0M EUR/yr
Total (Central)
149k EUR/MW7.5M EUR/yr
How was this page built?
1
Three Revenue Streams
Modeled separately then stacked: (a) Wholesale arbitrage from backtest results, (b) CRM capacity payments from T-4 auction data, (c) DASSA ancillary services from DS3 programme estimates.
2
Arbitrage Calibration
Backtest gives a range: EUR 56k/MW (Fixed Schedule, 54% capture) to EUR 93k/MW (Perfect Foresight, 100% capture). Central model uses 65% capture rate and 0.85 cycles/day, giving EUR 69k/MW. This was cross-checked against the Monte Carlo model (EUR 66k mean).
3
CRM Estimation
Used published T-4 auction clearing prices (2024-2028 delivery years) and BESS de-rating factors. 4-hour duration qualifies for higher de-rating than 2-hour. Applied a trend regression to recent auctions.
4
ML Trading Upside
Estimated that ML-optimised trading could achieve 75-85% of Perfect Foresight capture, pushing arbitrage to EUR 79-96k/MW. Based on academic literature on electricity price forecasting and conversations with trading firms.
Limitations: CRM is the biggest uncertainty — auction prices have been volatile and the mechanism may change. DASSA revenue is speculative (C4 confidence) as the programme is still evolving. The ML trading estimate is aspirational — no Irish BESS has demonstrated these capture rates yet.
Nerd level:
Show figures as
01 The Revenue Stack Year 1 (2028) · EUR/MW/yr by scenarioEUR thousands by scenario
Stacked Revenue per MW Three scenarios, Year 1 (2028)
Arbitrage C3 CRM C2-C3 DASSA C4-C5

Bear: EUR 96.2k/MW · Central: EUR 149.0k/MW · Bull: EUR 203.3k/MW.Bear: EUR 4.8M · Central: EUR 7.5M · Bull: EUR 10.2M. DASSA is the weakest number -- no auction has been held.
Source: REVENUE-BUILD-UP.md Section 5.2; SPREAD-PROJECTION.md; c004-crm-auction-results.md

02 Wholesale Arbitrage Build-Up The chain from spread to revenue
Revenue Derivation Chain Central scenario, Year 1
Gross Spread
86 EUR/MWh
4hr block spread (2028)
×
Capture Rate
65%
GB benchmark C3
=
Captured Spread
56 EUR/MWh
Net of RTE in formula
×
Annual Throughput
62,050 MWh/yr
50MW × 4h × 0.85cyc × 365d
=
Annual Revenue
3.45 EUR M
EUR 69k/MW/yrEUR 3.45M/yr
Gross spread C2 (historic data) Capture rate C3 (GB proxy)

Source: SPREAD-PROJECTION.md Section 3.2; KEY-FINDINGS.md Q1, Q2; BACKTEST-RESULTS.md

Central Arbitrage Revenue by Year 50 MW project, EUR thousands
Revenue declines with battery degradation (1.5%/yr compound). Spread projections remain relatively flat in the central scenario (EUR 83-86/MWh gross), but usable capacity erodes from 200 MWh to ~163 MWh by 2040. Net effect: arbitrage revenue falls ~3.5% over the project life.

Source: SPREAD-PROJECTION.md Section 3.2; ANNUAL-COSTS.md Section 4.1 (degradation)

Where Does Our Assumption Sit? Perfect foresight → ML-optimised → Fixed schedule

The model uses 65% capture and 0.85 cycles/day — representing an average professional operator. This sits between the backtest extremes. An ML-based trading system with good price forecasting (wind, demand, gas inputs) could close much of the gap to perfect foresight.

Trading Strategy Capture Rate Cycles/Day Revenue/MW/yrRevenue/yr vs Model Confidence
Fixed schedule (floor) 54% 0.94 EUR 56k/MWEUR 2.8M −19% C2 backtest
Current model (GB avg operator) 65% 0.85 EUR 69k/MWEUR 3.45M baseline C3
ML-optimised (conservative) 75% 0.95 EUR 79k/MWEUR 3.95M +14% C3 estimate
ML-optimised (optimistic) 82% 1.05 EUR 96k/MWEUR 4.8M +39% C4 estimate
Perfect foresight (ceiling) 100% 1.44 EUR 93k/MW*EUR 4.65M* +35% C1 backtest
Why ML could get close to PF: Day-ahead prices are partially predictable from wind forecasts, demand forecasts, gas prices, and interconnector flows — all publicly available inputs. The backtest’s fixed schedule is “zero intelligence” (same windows every day), while PF has perfect hindsight. A good ML model with adaptive dispatch sits naturally at 75–85% of PF. The gap between good and bad optimisers is widening as renewable penetration makes price patterns less regular — favouring ML over fixed rules.

IRR impact: ML conservative (+EUR 10k/MW = +EUR 500k/yr) swings IRR by ~+1.5pp. ML optimistic (+EUR 27k/MW = +EUR 1.35M/yr) swings by ~+3.5pp. However, even with ML-optimistic trading, the project base-case IRR remains negative (-3.5%) without D-TUoS reform. D-TUoS charges consume 41.1% of gross revenue; the spread must exceed EUR 90/MWh to clear an 8% hurdle without reform.
PF/FS from 7-year SEM backtest (C1–C2) ML estimates: no published Irish BESS ML trading data (C3–C4)

*PF at historic spreads (2019–2025 avg). Model uses projected 2028 spreads which are ~13% higher. Source: BACKTEST-RESULTS.md; KEY-FINDINGS.md Q1.

03 CRM (Capacity Market) Contracted availability payments · 40% of total
CRM Auction Clearing Price Trend EUR/MW de-rated/yr

Clearing prices have tripled since 2024/25, reflecting tightening capacity adequacy on the all-island system. T-4 2029/30 auction scheduled 26 March 2026 (not yet held).
Source: c004-crm-auction-results.md C1

CRM Revenue Calculation Central scenario
Installed
50 MW
×
De-Rating
40%
4hr duration C3
=
De-Rated
20 MW
×
Auction Price
149,960 EUR
T-4 2028/29 C1
=
CRM Revenue
3.0 EUR M/yr
EUR 60k/MW/yrEUR 3.0M/yr
Duration De-Rating De-Rated MW CRM/MW InstalledCRM Revenue Note
4hr 40% 20 MW EUR 59,984/MWEUR 3.0M Model assumption
2hr 25% 12.5 MW EUR 37,490/MWEUR 1.87M Standard 2hr rate
2hr (Derrymeen) 14% 7 MW EUR 20,997/MWEUR 1.05M Actual result -- below benchmark
Auction price C1 (published) De-rating factor C3 (estimated)

Source: c004-crm-auction-results.md; GB-BESS-PERFORMANCE.md Section 6.3

04 DASSA (System Services) Replacing DS3 · 13% of total · highest uncertainty
!
No DASSA Auction Has Been Held
This is the weakest number in the model. DASSA replaces the regulated DS3 programme (which expires Sep 2027) with competitive auctions. The first auction is expected May 2027. No clearing price data exists. The range below reflects genuine uncertainty, not analytical precision.
Confidence: C4-C5
DASSA Revenue Range EUR/MW/yr
Comparator Revenue Note
DS3 capped (2019 auction) EUR 58,000/MW/yrEUR 2.9M/yr Regulated tariff, now ending C2
GB ancillary services (2024) EUR 12-19k/MW/yrEUR 0.6-1.0M/yr Competitive market benchmark C2
DASSA conservative EUR 10,000/MW/yrEUR 0.5M/yr GB lower end; severe competition C4
DASSA central EUR 20,000/MW/yrEUR 1.0M/yr GB mid + Ireland premium C4
DASSA optimistic EUR 40,000/MW/yrEUR 2.0M/yr Near DS3 capped; regulatory support C4
DASSA expected May 2027, competitive auction format. DS3 currently provides ~88.5% of total Irish BESS revenue. The transition from regulated DS3 tariffs to competitive DASSA auctions is a cliff, not a slope. Industry consensus: DASSA will yield 30-60% less than DS3.

Source: KEY-FINDINGS.md Q3; GB-BESS-PERFORMANCE.md Section 6.1, 6.2

05 Revenue Projection (2028-2040) 50 MW project · three scenarios · stacked area
Total Revenue by Scenario 50 MW project, EUR thousands/yr
Scenario 2028 2030 2035 2040
Bear 4,810 3,790 3,503 3,273
Central 7,452 7,271 7,069 6,778
Bull 10,164 10,570 10,694 10,508
Central scenario declines ~9% over 12 years due to battery degradation (1.5%/yr compound). Bear collapses as BESS fleet growth compresses spreads. Bull rises as supply shortfalls widen spreads and limited BESS competition persists.
Overall confidence: C3

Source: REVENUE-BUILD-UP.md Section 5.2; SPREAD-PROJECTION.md Sections 3.2-3.4

06 Old Model vs New Model What changed and why
Model Comparison Previous assumptions vs audited figures
Metric Old Model New Model Change
Arbitrage/MW/yrArbitrage/yr EUR 80,000EUR 4.0M EUR 69,049EUR 3.45M -14%
Cycling 1.25/day 0.85/day -32%
Capture rate implied ~85% 65% -24%
CRM/MW/yrCRM/yr EUR 60,000EUR 3.0M EUR 59,984EUR 3.0M ~same
DASSA/MW/yrDASSA/yr EUR 30,000EUR 1.5M EUR 20,000EUR 1.0M -33%
Δ
Old model overstated arbitrage and understated D-TUoS impact
The previous model used 1.25 cycles/day and an implied ~85% capture rate. GB operational data from Modo Energy shows real-world 4hr batteries achieve 0.85 cycles/day and 65% capture. Correcting these two inputs alone reduces arbitrage revenue by 14%. However, the new model explicitly adds CRM and DASSA as separate streams, giving a more complete (and ultimately higher) total revenue figure of EUR 7.5M/yr.
Net effect: New model: EUR 149k/MW total revenueNew model: EUR 7.5M/yr total revenue -- but D-TUoS charges consume 41.1% of gross revenue (EUR 3,115k/yr), yielding a negative base-case IRR (-3.5%). The original IRR of 11.4% was inflated by ~5-6pp. NPV at 5% = EUR -22.7M. Project is ONLY viable with D-TUoS reform.
!
Financial Model Correction — DO NOT INVEST
Despite generating EUR 7.5M/yr gross revenue, D-TUoS charges consume 41.1% of gross revenue (EUR 3,115k/yr), making the project unviable without regulatory reform.
Corrected financial model outputs:
• Base-case IRR (no reform): NEGATIVE (-3.5%) — originally stated as 5-6%, which was inflated by ~5-6pp
• NPV at 5% discount rate: EUR -22.7M
• Simple payback: No payback
• With D-TUoS reform from Year 3: ~8.7% IRR (marginally clears 8% hurdle)
• With D-TUoS reform from Year 1: ~10.8% IRR
• Recommendation: DO NOT INVEST — WAIT for D-TUoS reform
See Operating Charges page for full D-TUoS impact analysis.
Sources: REVENUE-BUILD-UP.md, SPREAD-PROJECTION.md, BACKTEST-RESULTS.md, KEY-FINDINGS.md, GB-BESS-PERFORMANCE.md, c004-crm-auction-results.md, ANNUAL-COSTS.md. Every number is traceable to these source documents. For investment decisions, obtain independent bankable revenue forecasts from Cornwall Insight or LCP Delta.

Confidence scale: C1 = published/verified data · C2 = strong evidence · C3 = reasonable estimate · C4 = educated guess · C5 = placeholder/unknown.

Model date: 2026-02-22 (corrected) · 50 MW / 200 MWh (4-hour) BESS · Ireland all-island SEM