Energiforsk 2023-948 Reliability Analysis Microgrid (2023)

Energiforsk Report 2023:948 — Reliability Analysis of Microgrid Author: Ying He (Vattenfall AB) Programme: Risk- och tillförlitlighetsanalys (Energiforsk)

Programme chair: Jenny Paulinder (Göteborg Energi Nät)

Funded by / stakeholders: Vattenfall Eldistribution, Ellevio, Svenska kraftnät, Göteborg Energi, Elinorr, Jämtkraft Elnät, Öresundskraft, Skellefteå Kraft, Umeå Energi, Kraftringen Nät, Jönköping Energi Nät, and ~20 additional DSOs; Ei and Elsäkerhetsverket as adjunct members

Purpose: Develop a practical, reusable reliability analysis methodology for microgrids; demonstrate it via case study on a real Swedish microgrid

Summary

The report proposes an analytical (non-Monte-Carlo) reliability evaluation method for microgrids that can handle the unique complexity of:

1. Intermittent/random output from wind and PV DGs
2. Flexible switching between grid-connected and island (ö-drift) operation modes
3. Battery energy storage system contribution in island mode

The method evaluates reliability from three angles: load point and customer reliability indices (SAIFI, SAIDI, AENS, ASAI), generating capacity adequacy indices (LOLE, LOEE, LOLP), and islanding operation indices (IOSR, MIOP).

The case study applies the method to the Arholma system using actual load data from Oct 2015–Jun 2019 supplied by Vattenfall Eldistribution AB. Note: the report was written while Arholma was still in the building phase — the BESS was commissioned in August 2023 after this report was completed.

Arholma case study — system description

At the time of the case study, Arholma had:

• 80 kWp PV system (measured output used directly)
• Two BESS units (ESS1 and ESS2): each 160 kW maximum active power; combined 320 kW discharge capacity
• Load data: 213 customers across 19 load points
• Pre-commissioning BESS assumed design target: 1 hour at 320 kW = 320 kWh per island cycle

Grid topology: 11 kV MV feeders and 0.4 kV LV feeders; mix of underground cables and overhead lines; overhead lines (L23–L25, L27, L28) have higher failure rates and longer repair times, dominating the worst-performing load points (16–19).

Key quantified results

Load point and customer reliability — three-mode comparison

Island mode represents the theoretical maximum improvement if the microgrid were always in island mode — not the operational reality. Hybrid mode is the practical result, combining grid-connected operation (most of the time) with island fallback during upstream faults.

Individual load point improvement in island vs grid-connected mode: unavailability reduced 27–53% across all 19 load points. Load points 16–19 (served by overhead lines) see proportionally the largest absolute improvement.

Generating capacity adequacy — island sufficiency

With the 2015–2019 load profile and the pre-commissioning generation spec (80 kWp PV + 320 kW BESS for 1 hour):

Interpretation: The local generation is insufficient to meet island demand for approximately 130 hours per year on average. This means island operation fails (load must be shed or mainland connection restored) 1.5% of the time. The report attributes this to three general conditions: absence of light at night, insufficient light during the day, and insufficient energy in the battery storage system — it does not single out “winter” specifically as the driver.

The report itself does not discuss any planned demand-response or heat-pump extension to Arholma — that content, if accurate, comes from a different, later source (e.g. the 2025 Vattenfall update or the Lund thesis) and should not be attributed to this 2023 report.

Component reliability data used

The PCC failure rate of 2.1 f/yr (mainland cable faults) dominates system SAIFI — it is the most frequent failure event and the primary trigger for island mode activation.

Methodology overview

The recommended method follows four steps:

1. Input data: power network data, customer/load data (hourly time series), DG specifications, component reliability data
2. Load point indices: calculated separately for grid-connected, island, and hybrid modes using analytical equations; system indices aggregated via standard formulas (SAIFI = Σλᵢ·Nᵢ / ΣNᵢ; SAIDI = ΣUᵢ·Nᵢ / ΣNᵢ)
3. Generating capacity adequacy: chronological hourly comparison of combined PV+BESS generation vs load; LOLE = hours/year where generation < load
4. Operation indices: IOSR and MIOP from field performance data (not calculable for Arholma at time of publication — microgrid not yet operating)

For DG production modeling when measured data are unavailable: the report surveys four PV output methods (linear irradiance/temperature, quadratic irradiance, Beta distribution) and two wind speed models (Weibull, Rayleigh); recommends measured data or Renewables.ninja simulation as inputs.

Relevance to existing wiki pages

Companion reports

• Energiforsk 2023:957 — “Felbortkoppling i mikronät” (fault clearing in microgrids) — companion report on the protection system challenge. Referenced in the Lund University 2025 thesis; not yet ingested into this wiki.
• Source - Lund Arholma Microgrid Fault Detection (2025) — MSc thesis doing protection simulations of the same Arholma network; provides the engineering-level detail on why fault detection fails in inverter-dominated island operation.