Source - Lund Arholma Microgrid Fault Detection (2025)

Lund University MSc Thesis, June 2025 Limit in microgrid network size with respect to fault detection and clearing capability: A simulation study of the Arholma Microgrid Authors: Jonathan Molin & Axel Thunholm Division of Industrial Electrical Engineering and Automation (LTH), Lund University Supervisor: Professor Olof Samuelsson Company supervisors: Arne Berlin, Kristin Bobeck, Marcus Seisay (Vattenfall Eldistribution AB) File: raw/PDF extractions/Limit_in_microgrid_network_size_with_respect_to_fault_detection_and_clearing_capability_Web/

Document metadata

• Type: Academic MSc thesis, done in collaboration with Vattenfall Eldistribution AB
• Institution: Faculty of Engineering (LTH), Lund University
• Completed: June 2025
• Simulation tool: DIgSILENT PowerFactory (network model originally from Vattenfall R&D department)
• Scope: Fault detection and clearing in island (ö-drift) mode only; excludes transient/voltage/small-signal stability

Summary

The thesis investigates a fundamental engineering challenge for inverter-based microgrids in island operation: conventional protection devices are calibrated for fault currents from synchronous generators, but inverter-interfaced DERs (IIDERs) produce far lower fault currents. This mismatch can cause protection failure — faults go undetected, or wrong protection devices trip.

Using a PowerFactory model of the Arholma microgrid (provided by Vattenfall R&D), six simulation cases explore how the network can be expanded before protection fails. The thesis derives benchmark equations for future microgrid design.

The core problem: inverter fault current gap

Synchronous generators (SGs) produce fault currents of ~6 p.u. of rated current. IIDERs are limited to 1.1–2.0 p.u. depending on inverter type:

The Arholma BESS inverters are GFMIs — necessary for forming the island grid in the absence of synchronous machines. Once the inverter hits its current limit, it switches from voltage source behaviour (normal operation) to constant current source behaviour (fault condition). This means fault current no longer increases with decreasing fault impedance in the expected way — theoretical calculation methods break down, and conventional protection settings can fail.

Additional complication: at low Thévenin impedance (short, stiff networks), LL faults can produce smaller fault currents than SLG faults — the opposite of the conventional protection assumption. SG-based protection is typically dimensioned for SLG faults as the worst case; this assumption does not hold in inverter-dominated networks.

Arholma system parameters (from thesis)

Note on island duration discrepancy: This thesis states the island design target as “1 hour at 99%.” The operational Vattenfall communication (Source - Vattenfall Arholma Microgrid (2025)) describes the system as a “two-hour system.” The Energiforsk pre-commissioning study (Source - Energiforsk 2023-948 Reliability Analysis Microgrid (2023)) used 320 kW × 1 hour = 320 kWh as the design unit. Possible resolution: the total BESS capacity is 2 × 336 kWh = 672 kWh, which at 320 kW discharge gives ~2.1 hours; the formal reliability target (99th percentile including load variability) may be 1 hour. This discrepancy is a data gap — precise design intent requires clarification from Vattenfall.

Key findings by simulation case

MV network size (Case A)

Constraint: Zero-sequence capacitance accumulation in the resonance-grounded (Petersen coil) MV network. Each cable addition increases C₀; at approximately +1 µF above baseline, the ground overvoltage relay threshold is passed and the relay fails to detect SLG faults.

For future microgrids: This is manageable — the Petersen coil is tuned to total C₀ at construction time. A newly built microgrid starts with balanced resonance grounding. The true MV limit for future microgrids is reactive current injection from long cables, which trips the overcurrent relay by inflating apparent load current. Long overhead lines (OHLs) have far lower capacitance than underground cables and extend achievable MV network size substantially (>300 km OHL tested without C₀ constraint).

LV network size (Cases B, C, D)

Constraint: Rising Thévenin impedance at the fault location.

For normal household fuses (16–25 A): Swedish standards (SS 424 14 06 maximum ground loop impedance) remain applicable in island mode. Standard protection settings hold.

For large fuses (32 A, 63 A) or substations with LVCBs: Inverter current limitations become the binding constraint. The impedance at which LL fault current falls below SLG fault current is:

$$Z_{th} = \frac{V_{LL}}{2 \cdot (I_{lim} - I_{prefault})}$$

For current limitations of 1.2/1.5/2.0 p.u., this limiting impedance is 0.24/0.18/0.15 Ω respectively. Below this threshold, LL faults are the dimensioning case, not SLG faults — contrary to standard practice.

LVCB advantage: LVCBs are programmable and update their settings when the microgrid enters island mode (via cEMS communication). They can clear faults that fuses cannot in low-fault-current conditions. LVCB limitation: for substations with very large customer fuses (63 A), the LVCB’s 0.2 s clearing time requirement reduces the maximum allowable ground loop impedance compared to fuses (0.4 s), creating tighter network size constraints.

MV-only protection (Case E)

Finding: It is possible to eliminate LVCBs and rely solely on MV overcurrent relays for LV fault clearance, if MV radial load stays below a calculable threshold. The theoretical maximum Thévenin impedance for MV-only protection is:

$$Z_{0,1,2} \leq \frac{3\sqrt{3} \cdot V_{1\phi,LV}^2}{S_{3\phi}}$$

For a 150 kVA total load in an MV radial, this gives Z₀,₁,₂ ≤ 1.83 Ω — corresponding to a cable length of about 1.5 km in the Arholma network. Practical implication: cost reduction option for simple microgrid designs; suitable where selectivity is not required. Simulations show deviation from theoretical values at high load due to feeder power losses.

Inverter limitations (Case F)

Finding: All LV substations in the Arholma network have Thévenin impedances low enough for inverters to operate at current limit during 3ϕSC and LL faults. The constant-current-source behaviour that results causes voltage sags throughout the microgrid. For LV cable lengths shorter than ~2 km (>0.33 Ω), voltage drops below 90% of nominal — a power quality threshold under EIFS 2023:3.

Practical recommendation for operators: Two options to restore adequate margin between load current and fault current:

1. Oversize the inverter relative to maximum load current (expensive but maintains standard protection philosophy)
2. Increase the inverter current limiter setting (cheaper, but risks IGBT damage if fault duration is long; probability is low since island faults are rare)

Implications for future microgrid design

The thesis proposes the relationship between inverter size and maximum Thévenin impedance as a planning rule of thumb for new microgrids. Designing for the “one-fault-after-islanding” scenario is distinct from on-grid protection design and requires explicitly choosing between:

• Selective fault clearing (locate and isolate the fault while keeping rest of island energised) — requires LVCBs at each substation, strict settings, possible conflicts with large customer fuses
• Non-selective fault clearing (trip the entire island on any fault) — simpler, cheaper, achievable with MV-only relays if load stays within threshold

The thesis also recommends LVRT requirements for class A generators (< 1.5 MW, currently unregulated) as future work — in island mode, voltage sags from inverter current limiting could cause all class A DERs to disconnect simultaneously, causing cascaded blackout.

NextGen Arholma

The thesis mentions an ongoing Vattenfall R&D project called NextGen Arholma, investigating customer inclusion and flexibility to strengthen the microgrid. This project is a direct follow-on to the BESS-only phase, consistent with the DSR extension described in operational sources.

Companion reports

• Energiforsk 2023:957 — “Felbortkoppling i mikronät” — companion industry report on fault clearing in Swedish microgrids (not yet ingested into this wiki)
• Source - Energiforsk 2023-948 Reliability Analysis Microgrid (2023) — pre-commissioning reliability baseline for Arholma; provides the LOLP 1.5% figure that motivates the DSR extension
• Source - Vattenfall Arholma Microgrid (2025) — operational context: dashboard and news article describing the commissioned system

Relevance to existing wiki pages