Cascading Failures and Critical Infrastructures in Future Renewable European Power Systems

Abstract

The world's power systems are undergoing a rapid transformation, shifting away from carbon-intensive power generation to renewable sources. As a result, electricity is being transported over ever longer distances, while the intrinsic system inertia provided by thermal power plants decreases. Together, these developments raise the probability of cascading line failures and reduce the stability of the system after a system split. In this article, we assess the risk of cascading failures and system splits in the European power grid for different carbon reduction scenarios. We analyze the most likely and most dangerous splits, and identify critical transmission infrastructures and we discuss potential countermeasures that can address the problem of cascades. Our results show that while the risks of splits causing power failures rises with decarbonization, it can be mitigated cost efficiently.

Figures (11)

• Figure 1: Schematic of the analysis.a, The power system model PyPSA provides scenario data for different CO$_2$ targets and an $(N-1)$-secure dispatch at each time step. For every scenario and time step, we simulate all double transmission failures. These may trigger cascades and eventual system splits, which are analysed statistically. b, Schematic of a cascading failure in a power grid. An initial outage of two transmission lines (lightning bolts) redistributes real power flows $f_{m\rightarrow n}$ and may overload additional lines. The system can split into components with generation–load imbalance $\Delta P_c$. Splits are most critical when $\Delta P_c$ is large and the effective inertia $H_c$ is low.
• Figure 2: Scenarios for the decarbonisation of the European power system.a,b, Spatial distribution of cumulative generation in July and December at CO$_2$ levels of $58\%$ and $0\%$ relative to 1990. Disk size indicates monthly electricity generation per node. The $58\%$ scenario approximates the 2022 European system and serves as a reference. At $0\%$, generation is dominated by wind, solar, nuclear and hydropower, with pronounced seasonal and geographic variation. c, Annual generation by technology as a function of CO$_2$ level. Fossil generation is progressively replaced by wind and solar. d, Spatial distribution of battery and H$_2$ storage output capacity in the $0\%$ scenario. Batteries cluster near solar generation, while H$_2$ storage concentrates near wind-dominated regions. e, Total installed storage output capacity (logarithmic scale). H$_2$ storage increases sharply at $0\%$ as gas plants are no longer available to balance prolonged wind deficits, while battery capacity grows strongly beyond 20% decarbonisation.
• Figure 3: Pre-outage risk factors.a, Total power flow distance across emission levels. Higher values indicate longer average transfer distances and increased grid loading. The increase is constrained by fixed transmission capacities, which are not optimised. b, Total rotational energy of the power system. As decarbonisation progresses, inertia declines sharply, with very low values prevailing in the fully decarbonised scenario.
• Figure 4: Statistical assessment of system splits triggered by cascading failures.a, Distribution of power imbalance $\Delta P_c$ across split components. The overall shape remains similar across emission levels, but larger imbalances, particularly negative ones, become more frequent. b, Distribution of rotational energy $E_{{\rm kin},c}$ across split components. Very low-inertia splits become substantially more likely in the fully decarbonised system. Components carrying less than 10% of total load are excluded for clarity, as they exhibit low inertia by construction. c, Number of split events as a function of CO$_2$ level, grouped by share of load not served. The secondary axis indicates the probability conditional on an $N-2$ failure. While all outage sizes increase as emissions decline, the rise is most pronounced for the largest events corresponding to global severe splits.
• Figure 5: Characteristic geographic patterns of system splits. Split events with more than 10% load not served are clustered by blackout pattern (see Appendix). For each cluster, $R$ denotes its contribution to total load not served and $\beta$ its share of all considered blackouts. The 12 most frequent clusters (highest $\beta$) are shown. For each cluster, the maps displays the probability that a node experiences a blackout due to positive or negative RoCoF violation or remains stable (pie charts), and the probability of transmission corridor failure (colour scale). Cut-set edges are omitted, as they connect nodes with different outcomes by construction. Histograms below each map show the distribution of total lost-load share within the cluster, normalised by the number of trigger events, for each CO$_2$ level.