Energiforsk 2025-1148 Syntesrapport Risk Och Tillförlitlighet (2025)

Energiforsk’s synthesis report for the Risk- och tillförlitlighetsanalys (Risk and Reliability Analysis) research programme’s 2021–2025 period — popular-science summaries of 22 individual sub-projects, one of which is the stochastic operational-security report covered separately.

• Title: Risk- och tillförlitlighetsanalys 2021–2025 — En syntesrapport
• Authors: Andrea Bille Pettersson, Daniel Hirsch, Mathilde Ingrosso Hildingsson (Energiforsk)
• Report number: 2025:1148
• ISBN: 978-91-89918-55-9
• Type: Programme synthesis report (22 sub-projects + 2 training courses)

Scope

Programme running since 2006 (this synthesis covers the third period, 2021–2025; a fourth period, 2026–2030, follows). Steering-group chair: Jenny Paulinder (Göteborg Energi Elnät); programme manager: Susanne Stjernfeldt (Energiforsk). Each of the 22 projects gets a standalone popular-science summary; the report does not attempt a cross-project synthesis beyond the foreword. Note: the report’s own front matter states “22 reports,” but the table of contents lists only 21 distinct project write-ups (plus 2 training-course entries) — a pre-existing inconsistency in the source document, not introduced here.

Findings by topic

Hosting capacity / acceptansgräns — two related projects produce a practical handbook standardising deterministic, stochastic, and time-series methods for estimating how much new generation/load a distribution network can accept without reinforcement. Deterministic methods risk understating overvoltage; stochastic methods require agreeing an acceptable planning risk (past practice assumed 10% without documented justification); time-series methods are most realistic but data/compute-heavy. EV charging data still too sparse for reliable limits.

HV component reliability data — first large Swedish dataset (Svenska kraftnät + 100+ DSOs, 2014–2021) covering failure frequency, outage duration, and availability across 10 high-voltage component categories (36–400 kV); intended as a generic industry reference for companies lacking their own statistics.

Microgrid reliability — new reliability-index method (Arholma case study) for assessing robustness under stochastic DER production and grid-connected/island switching.

MVDC feasibility — strongest case found in urban substations (redundancy, lower losses, controllability); rural benefit assessed as limited at the low-voltage/local-grid level, with international experience suggesting DC’s rural benefit is more likely at higher voltage levels (e.g. regionnät) — a separate, unrelated recommendation calls for new research on low-voltage DC (LVDC) applications specifically; proposes a national MVDC forum and a Stockholm/Gothenburg demonstration project.

Voltage dips: 10 of 11 large industrial electricity users report power-quality problems, dominated by short dips (50–400 ms); current Swedish regulation doesn’t require DSOs to fix short dips, so the cost burden (UPS, protection systems) falls on end customers. A companion report finds the underlying 1990s-era knowledge base outdated, flags that dips may become more frequent/deeper under low-short-circuit-power operating states, and identifies knowledge gaps in modern electronics’ dip immunity.

Stochastic computation methods — new statistical techniques cut required simulation counts drastically while preserving accuracy, enabling more scenarios to be studied without over-dimensioning; includes automatic overvoltage/resonance mitigation and a new Swedish benchmark test network.

Machine-learning grid control — deep reinforcement learning outperformed conventional control in all tested scenarios, restoring security margins faster with fewer control actions and less load shedding; preventive control often needs just one step, and emergency control activates flexible resources (storage/demand response) instead of load-shedding. Performance degrades under measurement error.

Touch voltage / fault clearing time: recommends raising the current Swedish touch-voltage limit (220 V, max 0.5 s clearing time) for shorter clearing times, per the general European practice that permissible voltage depends on disconnection time (the raw report does not name IEC specifically for this recommendation); estimated ~10% higher spark/arc energy content under the reform. Fire-incident risk is described as based on very little existing material (not as “no data found either way”), and — importantly — the report’s own risk analysis is explicitly scoped to material/property damage only, not risk to people or animals.

Transformer probabilistic loading — 22-month measurement + simulation shows transformers often tolerate higher loading than currently permitted, especially accounting for weather/cooling; recommends shifting long-term planning limits from current-based to temperature/ageing-based, explicitly stating this “can give greater flexibility in the grid, e.g. for connecting solar panels or EV charging,” and that day-ahead risk quantification “opens up more efficient use of flexibility in the grid.”

Battery storage — three projects:

• Fire/explosion risk: not eliminable by any single measure; pressure-relief/deflagration protection needed (blast zones up to 25 m if underdesigned; toxic smoke spread practically 10–20 m vs. “hundreds of metres” in worst-case theoretical models). No national Swedish guidelines yet exist for BESS siting/protection — an explicit gap.
• Voltage/connection rules: fear that batteries cause large voltage variations is overstated; existing rule-of-thumb connection limits (AMP 2020, originally designed for solar PV) understate battery hosting capacity — batteries can be connected faster and in greater volume without reinforcement. Full-power discharge can still cause voltage-angle shifts/oscillation risk if co-located with synchronous generators (e.g. CHP plants). Networks are generally over-dimensioned for N-1/peak, so distributing load via batteries could allow more connections.
• Power quality: batteries can reduce THD and stabilise voltage, but only modestly and not as a standalone solution; effect depends heavily on placement. Profitability comes mainly from frequency-regulation markets (which DSOs themselves cannot access), not power-quality services. Recommends multi-purpose design (ancillary services + backup + PQ), standardised measurement, and clarified DSO/aggregator/TSO responsibility.

Connection to the N-1/probabilistic-methods thread

The stochastic-operational-security project (2026:1165) is one of the 22 summarized here, under the heading “Riskbaserad drift kan minska stora elavbrott – men kräver mer forskning” (“risk-based operation can reduce large outages — but needs more research”). The synthesis situates it within a programme-wide methodological shift: the same deterministic-to-stochastic move recurs independently in the hosting-capacity/acceptansgräns work and the transformer probabilistic-loading work — this is not an isolated transmission-only story.

Explicit flexibility/DSO/congestion connections

• Chairman’s foreword names the current key question as: “how can we use the grid more efficiently — e.g. connecting more customers without expanding it — by using flexibility and energy storage?”
• Battery/voltage report: DSOs currently cannot own their own batteries to shave peaks and avoid unnecessary reinforcement, since batteries count as “production” under unbundling rules — Paulinder is quoted wanting this changed. (This report itself does not cite an article number; the specific “Art. 36” unbundling citation is external context already covered in Energy Storage › EU regulatory framework and step 8 of the site’s DSO guide, Site/Guides/DSO-resan.en.md.)

Recommendations (selected)

Shorten Swedish touch-voltage/fault-clearing times toward general European practice; develop national BESS siting/protection guidelines; standardise severity/consequence factors and run control-room pilots for risk-based transmission operation; revise AMP/ASP-based battery connection limits; shift transformer planning to temperature/ageing-based limits; standardise BESS power-quality measurement and clarify operational responsibility; establish a national MVDC forum with an urban demonstration project.

Explicit scope exclusions

No EU-level regulatory discussion (Clean Energy Package, Network Codes, ACER) — this is a Swedish technical/engineering research programme, not a policy document. No local flexibility market design or FSP/procurement discussion — flexibility appears only as a grid-engineering lever. The chairman’s foreword flags “the growing threat to our grids from malicious actors” as an emerging concern for the next (2026–2030) programme period — broader than cybersecurity specifically (could include physical sabotage) — with explicitly no projects yet started on it in the period this report covers.

Relevance to wiki

• N-1 Criterion — corroborates the programme-wide deterministic-to-probabilistic shift and situates the stochastic-driftsäkerhet report within it
• Energy Storage — BESS fire-safety guidance gap (no national standards), battery voltage/connection-limit findings, DSO storage-ownership barrier
• Grid Capacity Utilization — hosting-capacity/acceptansgräns methodology, transformer probabilistic loading enabling flexibility connections
• Congestion Management — machine-learning-based preventive/emergency grid control activating flexible resources instead of load-shedding