Flex›Source - Energiforsk 2026-1157 Nationell Metod Effekt och Kapacitetsprognoser (2026)

Source - Energiforsk 2026-1157 Nationell Metod Effekt och Kapacitetsprognoser (2026)

Full citation: Andersson, M. & Olssén, H. (Sigholm AB). Nationell metod för effekt- och kapacitetsprognoser. Energiforsk Report 2026:1157. Stockholm: Energiforsk AB, January 2026. Program: Elnätens hållbara teknikutveckling och digitalisering. Project led by Erik Lejerskog, Energiföretagen. Working group: AG Kapacitetsprognoser (Energiföretagen), with Ellevio, Vattenfall Eldistribution, E.ON, Göteborg Energi, Öresundskraft, Svenska kraftnät.

Open access: Published by Energiforsk. This is Phase 1 (Etapp 1) of a two-phase project. Phase 2 (Etapp 2) will develop a Minimum Viable Product (MVP) of the data platform and establish a governance model.

Summary

This report develops a national top-down method for power and capacity forecasting for Swedish DSOs, with the aim of harmonizing DNDP (nätutvecklingsplan) forecasting practices across the ~170 Swedish DSOs. The project identifies a large gap between best practice and actual practice: 32% of DSOs lack documented forecasting methodology, 40% lack documented power templates, and 36% do not validate systematically. Forecasts have typically overestimated actual power needs — a systematic bias with direct implications for grid over-investment decisions.

The report’s deliverables are: (1) a top-down national forecasting methodology based exclusively on official, recurring national data sources; (2) worked examples for rooftop solar PV and home EV charging; (3) a conceptual design for a shared data platform (MVP to be built in Phase 2). The work draws on international experience from the Netherlands and Norway and domestic input from the large DSOs and Ei.

Key claims

The problem: fragmented and inconsistent forecasting

Survey of 50 DSOs + interviews with six major actors (E.ON, Vattenfall, Ellevio, Göteborg Energi, Öresundskraft, Svenska kraftnät):

  • Methods, definitions, and data sources vary enormously across companies
  • This variation makes DNDPs incomparable between companies and prevents national aggregation
  • Ei’s own experience collecting DNDPs for PM2025:03 confirms the same problem: manual input, inconsistent terminology, varying calculation methods, and inconsistent load diversity assumptions (Source - Ei PM2025-03 DNDP Sammanställning (2025))
  • Survey results:
    • 68% have documented methodology (50% of those use Energiforsk’s existing lathund model; 50% have own)
    • 32% lack documented methodology
    • 40% lack documented power templates
    • 36% do not validate systematically (only 19% do continuous validation against actuals)
    • Several DSOs note forecasts have overestimated actual power needs compared to outcomes
  • Top forecasting challenges: battery storage behavior (depends on frequency market participation, not predictable from network perspective); industry electricity intensity (driven by processes, not employment headcount); EV adoption rate uncertainty; customer behavior under new tariffs

The top-down methodology

Core principle: national prognosis → geographic disaggregation, based exclusively on official, recurring, nationally published data sources. This ensures:

  • Same national baseline assumptions for all DSOs
  • Traceable, documentable, regularly updatable
  • No local assumptions distorting aggregate totals

Three-step process per load category:

  1. Identify drivers and formulate assumptions — what economic, technological, and policy forces drive this load category’s growth?
  2. Match assumptions to official data sources — each assumption must have a qualifying data source (official, national, long time horizon, transparent methodology, regularly updated). If no qualifying source exists, the assumption is reformulated.
  3. Establish the national model — combine assumptions into a national forecast formula (low / expected / high scenarios); apply geographic disaggregation factors from national statistics

Geographic disaggregation: uses national proxies (proportional to existing installed capacity, traffic-work data, population by county, etc.). Does not capture local point loads, connection queues, or municipal development plans — these remain as bottom-up supplements by individual DSOs.

Worked examples

Rooftop solar PV:

  • National model: Energimyndigheten short-term energy forecast + long-term scenarios (ER 2025:13); solar energy → power via 1,000 full-load-hour assumption; national installed capacity baseline from EN0123 statistics
  • Geographic disaggregation: proportional to current installed solar capacity per municipality/county (Energimyndigheten data)
  • Formula: P_sol_nationellt = E_sol_nationellt / t; disaggregation: P_sol_i = P_sol_nationellt × (P_installerad_i / P_installerad_nationellt)

Home EV charging:

  • National model: Energimyndigheten transport energy scenarios → number of EVs → peak charging load using 1.1 kW/EV aggregated maxeffekt (from simultaneous charging factor × charger power); national load profile for home charging
  • Geographic disaggregation: proportional to traffic-work per county (Trafikverket data); reflects regional differences in vehicle use rather than just fleet size
  • Aggregated max power: 1.1 kW/EV (derived from typical home charger rates 3.7–11 kW × simultaneity factors 0.1–0.3)

International lessons

Netherlands (Netbeheer Nederland / TenneT capacity map):

  • Joint TSO/DSO capacity map updated monthly; color-coded (red/yellow/green) with next planned reinforcement
  • Started with Excel prototype, evolved to JSON-format automated platform
  • Dedicated “Team Semantics” for unified definitions; quarterly releases; mandatory shared governance
  • Key lesson: deliver early and iterate; uniform definitions are critical
  • Link: https://www.tennet.eu/nl-en/grid-capacity-map

Norway (Lede / Elbits / Elvia — wattapp.no):

  • First version delivered in 8 weeks (2022); shows capacity for connections >50 kW, both current and forward-looking per DNDP
  • Shared data structure; each DSO runs own calculations but on a common format
  • Link: https://www.wattapp.no/
  • Key lesson: start simple and build; shared platform and common language

Ei (Swedish experience):

  • Ei’s DNDP map tool (ArcGIS StoryMaps) confirms: manual input, inconsistent terminology, varying calculation methods, inconsistent load diversity assumptions
  • Ei plans a data portal for structured DNDP data input to improve standardization → directly feeds this project’s Phase 2 scope

Effektkommissionen (Region Skåne / effektprognoser.se):

  • Regional platform using 1 km² grid cells — found to be too fine-grained (misleading precision); municipality or grid area recommended instead
  • Industry sector is hardest to forecast: employment is a weak proxy for electricity intensity (process-driven, not headcount-driven)
  • Key lesson: a shared common reference is more valuable than false precision; stakeholder ownership requires co-development

Data platform concept

The report designs a four-component data platform for Phase 2:

ComponentFunction
Data ingestionAccepts open data from national sources; validates format and quality; standardizes to common schema
StorageStores both raw data and processed forecasts; full metadata (FAIR principles: Findable, Accessible, Interoperable, Reusable); version history
CalculationApplies the national top-down model; produces low/expected/high scenarios; performs geographic disaggregation
DisseminationUser interface (dashboards, maps, time-series visualization) + open API for programmatic access

MVP (Phase 2) priorities: searchable data catalog with FAIR metadata; downloadable datasets; basic time-series visualization; open API for system integration; version tracking. Advanced features (interactive maps, scenario comparison, governance-controlled access tiers) deferred to later phases.

Long-term vision: once Phase 2 platform is established and method validated, it could serve as the foundation for a national capacity map (kapacitetskarta) showing available grid capacity by area — following the Netherlands and Norway models.

Load categories prioritized and deferred

Covered in this report (Phase 1):

  • Rooftop solar PV
  • Home EV charging (personbilar)

Identified for Phase 2 expansion:

  • Public EV charging infrastructure
  • Solar parks / utility-scale solar
  • Battery storage / energy storage
  • Heavy transport electrification (tunga fordon)
  • Civic/commercial load (borgerlig last)
  • Industrial load (noted as hardest: process-driven, not amenable to national proxy)

Relevance to the wiki

This source directly informs or strengthens:

  • Distribution Network Development Plan — the comparability and standardization problem this report addresses is the same problem Ei PM2025:03 documents; the top-down method provides a systematic complement to DNDP bottom-up processes; Phase 2 data platform directly relevant to DNDP map tools and structured reporting
  • Flexibility Need Assessment — national forecasting standardization is a prerequisite for FNA quality; this report provides the infrastructure layer for consistent load scenario inputs to DNDP/FNA reporting chain; directly relevant to the FNA 2028 planning horizon
  • Distribution System Operator — the 32%/40%/36% forecasting gaps confirm broader digital maturity concerns; methodology support for smaller DSOs who cannot develop tools independently
  • Ei — the report’s scope directly complements Ei’s DNDP regulation (EIFS 2024:1), Ei’s map tool, and the proposed bemyndigande for structured DNDP data reporting; the report references EIFS 2024:1 as the regulatory requirement driving the standardization need

Internal links to other Energiforsk 2024–2026 program sources:

Data gaps

  • Phase 2 (MVP platform) not yet delivered; all references to the data platform are design concepts, not implemented systems
  • Battery storage excluded from the methodology — the report acknowledges this is a major gap (noted by Vattenfall and others) but does not propose a solution
  • Industry sector excluded from worked examples; the report correctly notes that employment-based proxies are inadequate but offers no alternative method
  • The worked examples (solar, EV charging) are stated as not validated — the methodology is conceptually demonstrated but backtesting against observed data is deferred
  • The relationship between this national method and the existing Energiforsk lathund för lokalnätsbolag (which 50% of DSOs already use) is mentioned but not fully specified — co-existence and integration need to be defined in Phase 2