On the record — Regulatory submission
CRU2026143: our response.
Joule's response to the Commission for Regulation of Utilities (CRU) consultation Electricity Network Tariff Structure Review (ENTSR), Phase 1: Objectives, Principles, Scope and Considerations on High-Level Tariff Design. Submitted through the CRU consultation portal on 8 November 2026 and reproduced here verbatim, as submitted.
ABOUTAbout the respondent
Joule is an independent software layer for the dispatch optimisation and settlement-grade evidence of large flexible electricity loads, with a focus on AI data centres. We are not affiliated with any generator, supplier, network operator, aggregator, or hardware vendor. In the interest of transparency: we have a commercial interest in demand-side flexibility, and we therefore limit our submissions strictly to positions supported by reproducible public evidence. Our methodology, reference model, and backtest results — including results that failed our own acceptance gates — are published in full at https://joule-os.com/methodology.html. The evidence cited below is computed exclusively from public data (SEMOpx / ENTSO-E Transparency Platform day-ahead prices; CRU and SEM Committee publications) and is reproducible.
We respond only to questions on which we hold direct, measured evidence.
Q1Objective of the ENTSR
We agree with the proposed single objective: to promote cost-efficient use of the electricity network in the context of electrification and decarbonisation.
We suggest one explicit addition to how the objective is operationalised: cost-efficient use requires that demand-side response be operable and observable. A tariff signal reduces network costs only if flexible load can (i) act on it within its operational constraints and (ii) demonstrate, at settlement grade, that it did. We therefore ask that the ability to measure and verify response be treated as instrumental to the objective throughout later phases.
Q2Assessment principles
We agree with the consolidated set of assessment principles, including cost reflectivity.
We propose one additional principle be made explicit: technology neutrality toward demand-side flexibility. Tariff structures should not implicitly assume that flexibility is provided only by generation or storage assets. Large demand users with schedulable load — data centres being the largest incoming class — can deliver equivalent network benefit, and tariff design should be neutral as to the technology providing the response.
Q3Weightings across principles
We agree that explicit weightings should not be fixed ex ante. We ask only that decisions under the ENTSR document, qualitatively, how each principle informed the outcome, so that stakeholders can understand the trade-offs made.
Q4 / Q4AScope (DUoS and D-TUoS)
We agree that DUoS and demand TUoS remain in scope, and that the analysis areas identified in the 2021 Call for Evidence remain in scope. No comment on Q5, Q6, Q8.
Q7System Services Charge costs in scope
We agree, with one interaction we ask the CRU to keep explicit in later phases: the costs recovered through the SSC relate to system services that flexible demand can itself provide under FASS/DASSA arrangements. Tariff design should avoid outcomes in which a flexible demand unit pays system-services-related charges on the same megawatts with which it provides system services. The treatment of this overlap will matter materially for the business case of demand-side flexibility.
Q9Exclusions from scope
We agree with the proposed exclusions.
Q10Interactions with out-of-scope topics
We flag one interaction with an out-of-scope topic that we believe the ENTSR cannot avoid: flexible (non-firm) connection agreements, including the arrangements contemplated under the CRU's Large Energy User connection policy and the sharing of MEC behind a single connection point (CRU202643). If network tariffs and connection agreements send inconsistent signals to the same large user — for example, if a non-firm connection is curtailable but its tariff charges are computed as if capacity were firm — the two regimes will conflict. We ask that the interaction between tariff design and non-firm capacity be addressed explicitly, even if connection charging itself remains out of scope.
Q11Marginal versus embedded cost approaches
We see merit in a hybrid approach: retain an embedded-cost base for revenue stability, while allowing marginal-cost-informed variation in time and location to carry the operational signal. Pure short-run marginal-cost network pricing would introduce volatility that is difficult for both network companies (revenue adequacy) and large users (budgeting and hedging) to manage, without commensurate benefit where locational differentiation is coarse.
Q12Short-run versus long-run cost signals
Both have distinct roles, and confusing them weakens both. Long-run signals inform siting and investment decisions — data centres make siting commitments on decade horizons and can respond to durable locational and time-of-use structure. Short-run signals inform operation, and they work only if published in advance: schedulable load plans at least day-ahead. To make this concrete with measured evidence: in a twelve-month backtest on SEM day-ahead prices (June 2025 – May 2026), a large flexible load using only day-ahead information captured 84.6% of the theoretical maximum arbitrage value, versus 29.9% for a fixed night-window schedule. Day-ahead visibility is what converts flexibility from a physical possibility into an operational resource.
Q13 / Q13A / Q14Locational signals
Yes for large users; no for domestic customers at this stage. For demand users above a threshold MIC (e.g., 1 MVA and above), locational signals should apply at the level at which constraints actually bind — regionally, and where data support it, at station level. Locational differentiation below that granularity risks complexity without measurable response. We would only add that locational signals are fairest — and most effective — when paired with the tools to respond to them (flexible connections, opt-in interruptible options); without those tools they risk reading as a pure penalty on exactly the locations where investment is otherwise most needed. No comment on Q13b, Q15–Q17 beyond noting that marginal-cost approaches and locational signals are natural complements for large users, and neither is a prerequisite for the other.
Q18Static time-of-use signals
Yes, as a baseline for all user classes where metering permits. Our analysis shows the existing 17:00–19:00 peak band is already directionally aligned with system value: across the same twelve months of SEM day-ahead prices, evening peak hours coincide with the highest system prices. Static ToU is proven to be understood by the market; its granularity should now catch up with settlement reality.
Q19Dynamic time-of-use signals
Yes for large users, under three conditions drawn from operational experience with schedulable load:
- Advance publication. Dynamic signals must be published at least day-ahead. Real-time-only signals exclude exactly the loads best placed to respond, because scheduled flexibility cannot re-plan inside the delivery hour.
- Settlement-granularity alignment. With the SEM now settled at 15-minute resolution since September 2025, tariff signals and their measurement should align to the same granularity; otherwise response is real but unverifiable.
- Methodological predictability. The methodology by which dynamic charges are computed must itself be stable and published, so that response can be planned and financed.
Q20Limiting time-of-use signals to particular customer groups
Yes. Dynamic time-of-use signals should be introduced first for large users, where interval metering, energy management capability, and materiality all exist. Static structures should remain available alongside, so that users who cannot respond are not penalised for a capability they do not have; the opt-in interruptible route (Q25) is the natural pairing for those who can. The domestic rollout can follow on its own evidence base; the two should not hold each other back.
Q21Dynamic vs static signals and short-run vs long-run costs
The two pairs map onto each other naturally. Static time-of-use signals are the right carrier for long-run cost differences: predictable, hedgeable, and sufficient for users whose flexibility is structural. Dynamic signals are the right carrier for short-run costs: scarcity driven by network and system conditions that only appears close to real time cannot be represented in a static band fixed years in advance.
Our measurements show why the distinction matters in practice (see also Q12). For a large flexible user on Irish day-ahead prices, a fixed night-window schedule — the behavioural equivalent of a static ToU band — captured 85–97% of the theoretical optimum in winter but only 20–35% in summer, when solar output shifts the cheapest hours to midday. A dynamic, day-ahead-published signal captured 84.6% year-round. Static bands alone would leave most of the summer value — and the associated network benefit — unrealised. The two signal types are complements: static for the long-run baseline, dynamic for the short-run residual.
Q24Whether dynamic signals require a prior marginal-cost, locational framework
No. Tying dynamic price signals to the prior adoption of a full marginal-cost, locationally-differentiated methodology would delay the lowest-regret element of reform by years. The evidence that flexible demand responds to dynamic signals exists today, under the current embedded-cost framework: day-ahead market prices are published around 13:00 the day before delivery, and measured response to them is material (Q12, Q21). A dynamic component can be introduced as an overlay — for example an opt-in adder or rebate on top of existing charges — without prejudging the marginal-versus-embedded question. That methodological debate should proceed on its own merits and timeline; it should not gate dynamic signals.
No comment on Q23, beyond the interaction already noted in Q13: locational and dynamic signals are complementary instruments, and each is fairest when users are given the tools to respond.
Q25Interruptible tariffs as an opt-in product
We strongly support this, and we believe it is the single highest-value design choice in this consultation for large flexible loads. An opt-in interruptible tariff converts connection-queue risk into a priced product: a large user that can accept a defined, verifiable interruption envelope receives a faster or cheaper connection and lower network charges, while the network receives firm, measurable relief at binding locations. This is the natural commercial complement to the non-firm connection arrangements the CRU is already developing. Two design features we consider essential: (i) interruption envelopes defined in advance (maximum frequency, duration, and notice), so the product is financeable; and (ii) compliance measured at settlement grade, so the network company can rely on it.
Q26Interruptible tariffs for energy storage units
Yes. Storage is the natural first candidate, and an enduring interruptible option would align ESU network treatment with the system value storage provides.
Q27Additional considerations
One: measurement and verification. Whichever combination of signals the ENTSR adopts, their value depends on the ability to verify response at settlement grade. We recommend that later phases include explicit treatment of baselining and M&V standards for demand-side response to tariff signals — including the option of independently maintained evidence records — so that compliance can be demonstrated without reliance on self-declaration alone. This protects the network companies, the CRU, and responding customers alike.