Energicentrum Gotland Tariff 2.0 Slutrapport (2026)

Final report of Tariff 2.0, a pilot on eastern Gotland of a dynamic, location-specific network tariff, by Energicentrum Gotland with Gotlands Elnät, Plexigrid (AI forecasting) and Ngenic (customer app). Two versions exist: Slutrapport 2026:04 (23 April 2026, 50 pages) and the fuller 2026:01 (23 June 2026, 142 pages, with annexes); this page follows 2026:01, which adds the finding that the forecast price signals missed the most critical load hours. The earlier slide deck is on Source - Energicentrum Gotland Dynamic Pricing and Tariff 2.0 (2025).

The pilot

• A network area under one distribution station with 1,700 subscribers and about 150 secondary substations; ten test pilots in five stations recruited by random invitation (a farm, a grocery store, a smokehouse, a restaurant, a sewage works and private persons), seven active; the rest are reference. Seven stations with 211 subscribers entered the analysis. A 12-month test year (1 April 2025 to 31 March 2026) was compared with a reference year.
• Flexible resources among the pilots: heat pumps (4, automatic via Ngenic), EV charging (6, manual), batteries (2, with no direct control from the price signal, tested through Home Assistant), defrost control for large freezers (2), direct electric heating (1), solar (7). Generic parts such as programmable Shelly relays and Home Assistant were needed where no commercial control existed, and the report calls the readiness of small-scale flexibility products limited.
• Because the tariff uses location signals, which Swedish rules had not allowed in local and regional grids, the pilot needed a regulatory sandbox. The report notes the government’s March 2026 decision to repeal the EIFS 2022:1 effektavgift requirement, with Ei to report a new proposal by 12 April 2027, and that the lagrådsremiss Nya lagar om elsystemet proposes removing the ban on geographically differentiated network charges from 1 January 2027.

Tariff design

• Dynamic energy price per secondary substation, rising with its loading degree (net load over capacity), symmetric between consumption and production (a mirror price for injection). Load-increasing flows pay a nätbelastningsavgift; load-relieving flows get a nätavhjälpningsersättning. A forecast engine publishes it day-ahead, shown per quarter-hour in the Ngenic app with a daily SMS; the maximum price is 20 SEK/kWh.
• Composite price: the local network price is added to the SE3 spot price into one signal, so spot dominates when the station is not loaded and the local price when it is. Only the consumption price was shown to pilots. Average local network prices were only about 2-10 öre/kWh against composite averages of 100-120 öre/kWh.
• Dynamic fuse charge or subscribed power with flexkompensation, an anti-location signal that returns the collected load charges per kW of connection in proportion to the station’s rolling annual average price, budget-neutral for the grid company, so that strong local energy prices do not create systematic unfairness. An earlier version used during the test changed the pilots’ invoices by less than the margin of error (-2.6 %).
• A customer-specific charge, and planned extensions: prices from overlying grids, voltage-based limits and probability-based pricing.

Results

• Users: the signal was easy to follow, the composite price was preferred, and loads moved in time (EV charging, heating, appliances). All respondents preferred a continuous, need-based price signal to peak-based effekttariffer, which some saw as misleading, unfair or punitive. Most saw potential in investing in flexibility such as battery storage if the signal stays, but only with more automation and support. Manual control was hard to sustain; automation is the key factor.
• Grid: effects on grid utilisation were limited at aggregate level, largely because the grid is lightly loaded most of the time. Day-ahead forecast signals did not capture the grid’s actual state in the most critical load situations, secondary substations are hard to forecast precisely, and pilots could use spot-linked flexibility more than the local grid price. In the most exposed station (36.6 % of connected capacity following the tariff) import loading fell and export loading rose between years, but this could reflect more own production. Station metering was missing 65-80 % of the time in January and February 2026 and was patched from subscriber data. The report concludes that the flexible tariff cannot on its own be the sought “vaccine” against grid problems.
• Summer export flexibility was largely absent.

Conclusions and recommendations

• Principles work, signal precision does not yet: add intraday updates (tighten limits during the day, never loosen), then real-time control for customers with automated resources, which Plexigrid rates the most promising long-term but whose metering is 3-7 years away for most of the grid; a later split into separate local and overlying price components.
• Separate steering signals from cost allocation; offer customers a choice of flexibility level (from zero flex with a higher fixed charge to full flex); standardised automation interfaces; allow location signals in regulation; test in more heavily loaded networks.

Relevance

The only Swedish field test of a dynamic location-specific tariff found so far, and a counterpoint to peak-based effektavgifter. See Power Tariffs and Home Batteries and Swedish DSO Tariff Reform — Three Parallel Tracks (2025–2027). It uses the same capacity-based method as Energiforsk’s UniFlex study (Source - Energiforsk 2026-1193 Unison Prissignal Timeffektavgift (2026)).

Limits of this summary

Read from 2026:01: the summary, chapters 1-3, 4.1, 5 and 6 and the start of the annexes. Chapter 4.2 (station-by-station quantitative analysis and its figures) and annexes 3-6 (flexkompensation, overlying prices, voltage methods) were not read in detail. 2026:04 was identified from its front page and summary only. Table 1 and the figures were not in the extracted text.