Från dödnätsstart till ödrift Uppsala Thesis (2025)

Uppsala University master’s thesis, 2025 — Från dödnätsstart till ödrift: Simulering av stabilitet, elkvalitet och pålastningsstrategier (From Black-Start to Island Operation: Simulating Stability, Power Quality and Loading Strategies). Author: Märta Strømme. Supervisor: Robert Eriksson; Subject reviewer: Göran Ericsson; Examiner: Elísabet Andrésdóttir. Teknisk-naturvetenskapliga fakulteten, Uppsala University.

Simulation study of a real (anonymized) municipality in central Sweden, modelling black-start and subsequent island operation from a hydropower plant and a CHP plant.

Summary

The thesis investigates how a waterpower plant and a CHP plant handle the transition from dead-grid start (dödnätsstart) to stable island operation supplying 15 prioritized buildings in the municipality. Four loading strategies (pålastningsstrategier) are evaluated using Simulink simulation models. Results show that both plant types maintain acceptable stability, but the CHP plant is more robust due to its higher capacity relative to the load. The choice of loading strategy significantly affects frequency and rotor angle transients.

Methodology

• Tool: MATLAB/Simulink
• Plant models: water turbine, steam turbine (CHP), excitation system, synchronous generator
• Grid model: transformers, cables, prioritized buildings as loads
• Data provided by the local (anonymized) energy company
• 15 prioritized buildings modelled as composite loads
• Simulations cover the initial load connection phase only — not long-term operation or reconnection to the main grid

CHP plant rated capacity: 30.6 MVA

Loading strategies evaluated

Four strategies for connecting priority loads after island startup:

Key results

Frequency and rotor angle (CHP plant)

• Maximum frequency dip: 49.87 Hz in the “Alla samtidigt” maximum load scenario — within the permissible range but near the lower limit
• Frequency, rotor angle, and voltage stabilize approximately 20 seconds after the last load connection, across all scenarios for both plants (per the thesis’s own conclusions; absolute settling time from simulation start is closer to 80–90 seconds, since the last load in most scenarios connects around 60–75 seconds in)
• Strategies with staged loading (time gaps between connections) produce significantly smoother frequency and rotor angle transient responses
• Kluster: best compromise — better frequency performance than all-at-once, better rotor angle performance than pure sequential strategies
• The one exception across all scenarios: the hydropower plant’s “Alla samtidigt” maximum-load scenario did not stay above 49 Hz — the thesis explicitly flags this as the sole case where frequency regulation could not handle the load step

Voltage

• All scenarios: stabilized voltage remains within ±5% p.u. (0.95–1.05 pu) across all loading strategies and scenarios
• Voltage transients at each load-connection step recover quickly but the recovery time differs by plant: ~0.6 ms for the hydropower plant, ~1 ms for the CHP plant (the CHP plant’s transients are smaller in magnitude but take slightly longer to settle)
• One scenario produced a notable transient overvoltage spike to 1.3 pu (CHP, “Storleksordning (minst)”, smallest load first) before all loads were connected — the thesis attributes this to the model’s pre-set reference-power ramp-up rather than a real limitation, and argues a correctly dimensioned excitation system would handle it in practice

Hydropower plant

• Also maintains acceptable stability in general, but with larger relative perturbations than the CHP plant — the CHP plant has more available capacity relative to the 15 priority buildings’ demand, making it more tolerant of load steps (see the “Alla samtidigt” exception above)

Power quality

• Voltage transients occur at each load connection step but recover quickly (see Voltage above); the thesis does not report a separate finding on sustained oscillatory instability in the results section

Discussion

Model limitations (as stated by the thesis itself): the plants ramp up toward a pre-set reference power from 5 seconds regardless of actual demand, which likely overstates real-world stability; loads are modelled as static rather than dynamic; the steam turbine is modelled as a single rotating mass; only one of the hydropower plant’s two generating units is modelled; the CHP plant’s simulation covers only its electrical output, not its district-heat off-take; and the simulations use phasor (fundamental-frequency) rather than instantaneous-value modelling, which filters out millisecond-scale transient detail. Results are indicative, not certifiable for production use.

Implication for loading strategy design: gradual, time-separated loading reduces stress on frequency and rotor angle stability. The Kluster strategy offers a practical middle ground between operational simplicity and stability performance.

Grid-forming capability (wiki context, not discussed in the thesis): both plants simulated are synchronous machines with inherent grid-forming characteristics — they establish voltage and frequency references without external support. This contrasts with inverter-based sources (solar, wind, BESS with grid-following control) which cannot independently sustain island operation without grid-forming control.

Preparedness gap: MSB (2022), as cited by the thesis, noted that Swedish municipalities are insufficiently prepared for large-scale outages and that there are too few reserve/backup power plants (“reservkraftverk”) to handle larger grid disturbances — the thesis does not specify that these must be black-start-capable. This thesis provides quantitative simulation support for the selection and operational design of island networks.

Relevance to other wiki pages

• Island Operation — loading strategies, simulation methodology, frequency/voltage results for island startup
• Energy Storage — grid-forming vs grid-following; role of synchronous generation as frequency anchor
• Electric Power Distribution — protection relay settings in weak (low kortslutningseffekt) island networks; transformer and cable modelling