Sensitivity & Risk — BESS Feasibility Ireland

How was this page built?

Tornado Analysis

One-variable-at-a-time sensitivity on 7 inputs: D-TUoS regime, spread level, CRM price, capture rate, CAPEX, degradation, trading optimisation. Each variable moved to its plausible low and high bounds (sourced from upstream confidence intervals).

Per-Bar Sourcing

Unlike typical tornado charts, every bar is annotated with the source of its range. E.g., spread range comes from the Monte Carlo P10-P90; CRM range from the T-4 auction trend regression; CAPEX range from the hardware vendor survey.

Scenario Construction

Three scenarios (Bear / Base / Bull) constructed by combining correlated variables. Bear assumes all downside risks materialise simultaneously; Bull assumes favourable convergence.

Heatmap

2-variable sensitivity (spread level × D-TUoS regime) showing IRR across the full parameter space. Identifies the “viable zone” where IRR exceeds the 8% WACC hurdle.

Limitations: One-at-a-time sensitivity doesn’t capture interactions between variables (e.g., higher spreads AND higher CRM often move together). The scenario construction is subjective — we chose “reasonable worst/best” rather than formal probability weighting. The Bear scenario is plausible, not extreme.

Nerd level: classic sensitivity analysis, but the per-bar sourcing elevates it — every number has a receipt

Show figures as

01 Tornado Chart: IRR Sensitivity Each bar justified below

Unlevered IRR Sensitivity Central scenario, current D-TUoS regime, corrected base IRR -3.5%

1. D-TUoS Reform — Low: -3.5% (no reform) → High: ~10.8% (full reform Year 1); ~8.7% (reform Year 3). D-TUoS = EUR 3,115k/yr = 41.1% of gross revenue. Source: FEE-VERIFICATION.md §6.2, C1/C2. Reform = full exemption of System Services Charge on BESS imports (ESI/ECA recommendation, Mar 2025). CRU consultation closed Oct 2025; no decision paper published.

2. DASSA Revenue — Low: -5.6% (EUR 10k/MW/yrEUR 0.5M/yr) ↔ High: 0.4% (EUR 40k/MW/yrEUR 2.0M/yr). Swing: 6.0pp. All still negative without reform. Source: No DASSA auction held; first expected May 2027, C4-C5. Range anchored on GB ancillary EUR 12-19k/MW/yrEUR 0.6-1.0M/yr (Modo Energy) and DS3 historic cap EUR 58k/MW/yrEUR 2.9M/yr.

3. CRM Auction Price — Low: -6.1% (EUR 120k/MW de-ratedEUR 2.4M/yr) ↔ High: -1.1% (EUR 180k/MW de-ratedEUR 3.6M/yr). Swing: 5.0pp. Still negative at high test without reform. Source: SEM Committee T-4 2028/29 auction, EUR 149,960/MWEUR 3.0M/yr, C1-C2. Trend: rising sharply from EUR 48kEUR 0.96M (2024/25) to EUR 150kEUR 3.0M (2028/29).

4. Captured Spread — Low: -6.1% (EUR 45/MWh) ↔ High: -1.3% (EUR 65/MWh). Swing: 4.8pp. Source: SEM 7-year hourly data (C1) x 65% capture (GB proxy, C3). Base: EUR 55/MWh = EUR 85/MWh gross x 65% capture rate. To achieve 8% hurdle without reform, spread must exceed EUR 90/MWh.

5. Capex — Low: -1.6% (EUR 29.6M) ↔ High: -5.4% (EUR 39.9M). Swing: 3.9pp. Source: CAPEX-BUILD-UP.md, 17-component build-up from BNEF, Ember, EirGrid, C2-C3. Key driver: LFP cell pricing (EUR 6.8M at EUR 34/kWh FOB) and transformer (EUR 2.8M, 128-144 week lead).

6. Cycling Rate — Low: -4.1% (0.70/day) ↔ High: -2.8% (1.00/day). Swing: 1.3pp. Source: GB 2hr benchmarks (Modo Energy, 1.0/day, C2), C3-C4. Surprisingly small because D-TUoS System Services Charge (EUR 27/MWh) captures 57% of marginal cycling revenue. Under reform, this sensitivity is ~4x larger.

All sensitivities computed from corrected Central/Current base case (IRR -3.5%). Each variable tested independently; others held at base values. WACC threshold at 8%. No single variable can lift IRR above 8% without D-TUoS reform. Computed by scripts/compute_irr.py.

02 What Each Bar Means Expand for full justification

1 D-TUoS Reform +14.3pp (to 10.8%)

Base

No reform

Low Test

No reform

High Test

Full reform Yr 1

Source

EirGrid SoC 2025/26

Confidence

C1 rates / C2 timing

The D-TUoS charges total EUR 3,115k/year — 41.1% of Year 1 gross revenue (EUR 9,434k). This is the single largest cost item and the reason the base-case IRR is negative (-3.5%). Full reform (exempting ESS from the System Services charge, per the ESI/ECA recommendation of March 2025) lifts IRR to ~8.7% (reform from Year 3) or ~10.8% (reform from Year 1). The project is ONLY viable with reform.

Reform status: CRU consultation closed October 2025. No decision paper published. Estimated probability: 20-30% by October 2027. This is a binary policy variable — either the exemption is enacted or it is not. The tariff rates themselves are C1 (published), but the reform timing is genuinely uncertain. Recommendation: DO NOT INVEST until reform certainty exists.

2 DASSA Revenue 6.0pp swing

Base

EUR 20k/MW/yrEUR 1.0M/yr

Low Test

EUR 10k/MWEUR 0.5M

High Test

EUR 40k/MWEUR 2.0M

Source

GB proxy + DS3 cap

Confidence

C4C5

DASSA (Delivering a Secure Sustainable Electricity System) replaces DS3 as the ancillary services framework. No DASSA auction has been held. The first auction is expected May 2027. Our EUR 20k/MW/yrEUR 1.0M/yr central estimate is based on GB ancillary revenues of EUR 12-19k/MW/yrEUR 0.6-1.0M/yr (Modo Energy 2024, measured) plus an Ireland premium for higher RES-E penetration and smaller system size.

The range is wide because DS3 historic tariffs were EUR 58-300k/MW/yrEUR 2.9-15.0M/yr (capped/uncapped), but industry consensus is that competitive DASSA auctions will yield "significantly less" than regulated DS3 levels. The low test (EUR 10kEUR 0.5M) represents GB-level clearing. The high test (EUR 40kEUR 2.0M) reflects the possibility that Ireland's flexibility needs command a substantial premium.

3 CRM Auction Price 5.0pp swing

Base

EUR 150k/MWEUR 3.0M/yr

Low Test

EUR 120k/MWEUR 2.4M/yr

High Test

EUR 180k/MWEUR 3.6M/yr

Source

SEM T-4 2028/29

Confidence

C1 price / C3 de-rating

The CRM T-4 2028/29 auction cleared at EUR 149,960/MW de-rated/yearEUR 3.0M/yr — a published, verified number (C1). The trend is strongly upward: from EUR 48kEUR 0.96M (2024/25) to EUR 150kEUR 3.0M (2028/29), reflecting tightening capacity adequacy on the all-island system.

The main uncertainty is not the clearing price but the de-rating factor: our model assumes 40% for 4hr BESS, yielding 20 MW de-rated from 50 MW installed. The exact methodology requires the EirGrid IAIP document (unpublished). If de-rating falls to 35%, CRM revenue drops proportionally. The T-4 2029/30 auction (26 March 2026) is the next key data point.

4 Captured Spread 4.8pp swing

Base

EUR 55/MWh

Low Test

EUR 45/MWh

High Test

EUR 65/MWh

Source

SEM data + GB capture

Confidence

C2 spread / C3 capture

EUR 55/MWh = EUR 85/MWh gross spread (4hr battery mean daily, Jan 2024 - Jan 2026, from 64,670 hourly SEM prices, C1-C2) multiplied by 65% capture rate (GB 2hr operational benchmark from Modo Energy 2024, C3). The spread is well-established; the capture rate is the weaker link because it is a GB proxy applied to an Irish 4hr system.

Future spread depends on opposing forces: RES-E expansion widens spreads (+EUR 12/MWh per 10pp RES-E), while BESS fleet growth compresses them (-EUR 11/MWh per GW of BESS). See SPREAD-DRIVERS.md for detailed driver analysis. The net direction is approximately flat in the central case.

5 Capex 3.9pp swing

Base

EUR 34.0M

Low Test

EUR 29.6M

High Test

EUR 39.9M

Source

CAPEX-BUILD-UP.md

Confidence

C2C3

EUR 34.0M is a bottom-up 17-component build from BNEF (cells), Ember (core equipment), EirGrid (MEC bond), and Wood Mackenzie (transformer). Key items: LFP cells EUR 6.8M (EUR 34/kWh FOB China), grid connection EUR 4.5M (shallow), EPC EUR 5.1M, transformer EUR 2.8M (128-144 week lead time).

No Irish BESS installed cost benchmark exists. The range reflects cell price volatility (EUR 25-45/kWh is the global range), grid connection uncertainty (EUR 1.3-5.3M), and contingency sizing. The downside risk is transformer lead-time delays causing cost escalation.

6 Cycling Rate 1.3pp swing

Base

0.85/day

Low Test

0.70/day

High Test

1.00/day

Source

GB 2hr benchmarks

Confidence

C3C4

GB 2hr batteries average 1.0 cycles/day (Modo Energy 2024, measured, C2). Our 0.85 central applies a duration penalty: 4hr systems need 8 hours per full cycle vs 4 hours for 2hr, reducing the number of economically viable cycles per day. No published 4hr BESS cycling data exists anywhere globally (C5). Cushaling (Statkraft, Ireland's first 4hr BESS, Feb 2026) is the first potential datapoint but no data has been published.

Why this sensitivity is surprisingly small (1.3pp): Under the current D-TUoS regime, the System Services Charge (EUR 27/MWh on imports) captures 57% of marginal cycling revenue. Going from 0.85 to 1.00 cycles/day adds EUR 609k/yr gross but EUR 347k/yr in D-TUoS, netting only EUR 262k. Under reform, cycling sensitivity is approximately 4x larger because marginal import costs drop to EUR 4/MWh.

7 Trading Optimisation (ML vs Fixed Rules) +1.5 to +3.5pp upside

Base

65% capture, 0.85 c/d

ML Conservative

75% capture, 0.95 c/d

ML Optimistic

82% capture, 1.05 c/d

Source

PF/FS backtest + GB data

Confidence

C3C4

The base model uses 65% capture rate (GB average operator, Modo Energy) and 0.85 cycles/day. This represents competent but not exceptional dispatch. An ML-based trading system using publicly available inputs (wind forecasts, demand forecasts, gas prices, interconnector flows) could plausibly reach 75–85% of perfect foresight. The backtest shows PF averages EUR 93k/MW/yrEUR 4.65M/yr while the fixed schedule averages EUR 58k/MW/yrEUR 2.9M/yr — the gap is large enough that ML optimisation represents meaningful upside.

Revenue impact: ML conservative (EUR 79k/MWEUR 3.95M) adds EUR 500k/yr (+1.5pp IRR). ML optimistic (EUR 96k/MWEUR 4.8M) adds EUR 1.35M/yr (+3.5pp IRR). This is asymmetric upside only — no operator would do worse than the fixed schedule floor (EUR 56k/MWEUR 2.8M). The declining capture rate trend (73% in 2020 → 54% in 2025) means adaptive ML dispatch is increasingly important relative to fixed rules.

03 Reform Timing Matters IRR vs reform start year

IRR by Reform Start Year Central scenario — reform required to reach WACC threshold

Without reform, the project has an IRR of -3.5% and never breaks even (total FCF: EUR -21,356k). With reform from Year 3: ~8.7% IRR (marginally above WACC). With reform from Year 1: ~10.8% IRR (comfortably above). Every year of delay costs approximately EUR 1.5-2.0M in foregone D-TUoS savings. The project is ONLY viable if reform is enacted. Recommendation: DO NOT INVEST until reform is certain.

04 Two-Variable Heatmap Reform timing x DASSA revenue — the two biggest unknowns

IRR: Reform Timing vs DASSA Revenue Red < 8% WACC · Yellow 8-10% · Green > 10%

DASSA EUR 10k/MWDASSA EUR 0.5M

DASSA EUR 20k/MWDASSA EUR 1.0M

DASSA EUR 40k/MWDASSA EUR 2.0M

No reform

-5.6%

Negative IRR

-3.5%

Base case (corrected)

0.4%

Still below WACC

Reform Year 3

6.6%

Below WACC

8.7%

Just above

12.5%

Above WACC

Reform Year 1

8.7%

Marginal

10.8%

Above WACC

14.6%

Strong

Corrected: 4 of 9 combinations fail the 8% hurdle. The entire "No reform" row is negative IRR — the project is unviable under any DASSA assumption without D-TUoS reform. Even with high DASSA (EUR 40kEUR 2.0M), IRR is only 0.4% without reform. With reform from Year 3, the project reaches ~8.7% at base DASSA. The project is ONLY viable with D-TUoS reform. Probability-weighted IRR: ~1-2%.

Both variables tested simultaneously. All other assumptions at central values. Source: scripts/compute_irr.py, two-variable sensitivity.

05 Confidence Assessment 10 key assumptions rated C1-C5

Assumption Confidence Register From EVIDENCE-CHAIN.md

C1 Published official data

C2 Verified independent data

C3 Reasonable estimate

C4 Informed guess

C5 No data / pure estimate

#	Assumption	Value	Confidence	Critical Gap
1	Capex	EUR 34M	C2C3	No Irish BESS installed cost data
2	Annual fees (current)	EUR 3.54M/yr	C1	Published tariffs; low risk
3	Annual fees (reform)	EUR 2.16M/yr	C1 / C2	Reform timing uncertain
4	Captured spread	EUR 49/MWh	C2C3	Capture rate is a GB proxy
5	Cycling rate	0.85/day	C3C4	No 4hr data exists globally
6	CRM revenue	EUR 60k/MW/yrEUR 1.2M/yr	C2C3	De-rating factor approximate
7	DASSA revenue	EUR 20k/MW/yrEUR 1.0M/yr	C4C5	No auction held; pure estimate
8	Spread drivers	See tornado	C2C4	Cannibalisation slope unverified for Ireland
9	Degradation	1.5%/yr	C2	Standard LFP; low risk
10	4h vs 2h duration	4h wins on ROI	C2C3	Forward applicability of backtest

Overall model confidence: C3. The strongest inputs (fees, CRM price) are C1-C2. The model is dominated by two C4-C5 assumptions: DASSA revenue (no auction held) and cycling rate (no 4hr data exists globally). These two variables alone account for a combined 7.3pp of IRR uncertainty. However, even optimistic assumptions for all market variables cannot overcome the D-TUoS drag: base-case IRR is -3.5% and no combination of market variables achieves 8% without reform.

06 Model-Killing Uncertainties Four items that could flip the investment decision

C4-C5 DASSA Revenue

EUR 30k/MW swingEUR 1.5M swing

Range: EUR 10k - 40k/MW/yrEUR 0.5 - 2.0M/yr. No DASSA auction has been held; first expected May 2027. The central estimate (EUR 20kEUR 1.0M) is based on GB proxy data. If DASSA clears below EUR 15kEUR 0.75M, the central revenue case is impaired. If above EUR 35kEUR 1.75M, the project clears 8% WACC even without D-TUoS reform.

C3-C4 Cycling Rate

EUR 18k/MW/yr swingEUR 0.9M/yr swing

Range: 0.70 - 1.00 cycles/day. No published 4hr BESS operational data exists anywhere globally. GB 2hr averages 1.0/day; we apply a duration penalty to reach 0.85. Cushaling (Statkraft, Ireland, Feb 2026) may provide the first datapoint. Under reform, cycling sensitivity is ~4x larger than under the current D-TUoS regime.

C1/C2 D-TUoS Reform

+12.3pp IRR swing (from -3.5% to ~8.7%)

Binary variable: reform or no reform. D-TUoS charges = EUR 3,115k/yr = 41.1% of gross revenue. Without reform, IRR is -3.5% (negative) and the project never breaks even. The charge rates are published (C1), but the reform decision is political/regulatory. CRU consultation closed Oct 2025 with no decision paper. Probability: 20-30% by Oct 2027, rising to 50-60% within 5 years. This single variable determines whether the project is viable at all.

C3-C4 BESS Fleet Growth

EUR 33k/MW/yr spread differenceEUR 1.65M/yr spread difference

2 GW vs 5 GW of BESS by 2030 creates a EUR 33k/MW/yrEUR 1.65M/yr difference in arbitrage revenue through spread compression. Each GW of BESS deployed compresses the spreads that all BESS assets profit from. Currently 1.05 GW installed, 10 GW pipeline on paper, but only 83 MW under construction. The trajectory matters enormously.

Sources & methodology: All sensitivity analysis computed by scripts/compute_irr.py with full 20-year cash flows. Every assumption traces to the source documents referenced in brackets. Confidence ratings from EVIDENCE-CHAIN.md. The tornado chart sensitivities are one-at-a-time (OAT); the heatmap shows two-variable interaction. For investment decisions, obtain independent bankable revenue forecasts from Cornwall Insight or LCP Delta, and project-specific cost quotations from EPC contractors. This analysis is conducted for research purposes and does not constitute investment advice.

Date: 2026-02-20 · System: 50 MW / 200 MWh LFP BESS, Ireland · Project life: 20 years (2028-2047) · WACC: 8% unlevered

What Swings The Decision?