Quantifying cascading power outages during climate extremes considering renewable energy integration

Abstract

Climate extremes, such as hurricanes, combined with large-scale integration of environment-sensitive renewables, could exacerbate the risk of widespread power outages. We introduce a coupled climate-energy model for cascading power outages, which comprehensively captures the impacts of evolving climate extremes on renewable generation, and transmission and distribution networks. The model is validated by the 2022 Puerto Rico catastrophic blackout during Hurricane Fiona, the first-ever system-wide blackout event with complete weather-induced outage records. The model presents a novel resilience pattern that was not captured by the present state-of-the-art models and reveals that early failure of certain critical components surprisingly enhances overall system resilience. Sensitivity analysis of various behind-the-meter solar integration scenarios demonstrates that lower integration levels (below 45%, including the current level) exhibit minimal impact on system resilience in this event. However, surpassing this critical level without additional flexibility resources can exacerbate the failure probability due to substantially enlarged energy imbalances.

Figures (5)

• Figure 1: Schematic diagram of the CRESCENT model. The proposed CRESCENT model enables the high-resolution spatiotemporal analysis of the climate extreme effect on comprehensive energy systems and captures the cascading outage dynamics.
• Figure 2: The 2022 catastrophic blackout in Puerto Rico during Hurricane Fiona and contemporary Puerto Rico renewable power system.a, Spatiotemporal power outage map of the Puerto Rico power grid from 10:00 UTC to 18:00 UTC on September 18, 2022, along with the track (black line) and center location (yellow dot) of Hurricane Fiona. b, The track of Hurricane Fiona, highlighted by its maximum sustained wind speeds. Hurricane Fiona was a Category 1 hurricane during its landfall in Puerto Rico. c-d, Maps of (c) the penetration level and (d) the capacity of distributed renewable generation (rooftop solar systems; with data recorded at the distribution feeder level). e, Transmission network with generation capacity. f, Distribution network with the spatial distribution of demand. The power grid exhibits a supply-demand structure with major generation in the south and the primary load centers in the north (San Juan). The structure heightens the risk of power imbalance during disruptions of the transmission network. Despite certain distribution feeders reaching a renewable penetration level of 55%, the overall penetration level with a total installed distributed renewable generation capacity of 296 MW remains below 20% of the peak demand (2751 MW), approximately at 16%.
• Figure 3: Realizations of cascading power outages. a, Comparison of the percentage of total customers with electricity between the simulated and observed cases for the Puerto Rico power grid during Hurricane Fiona on September 18th, 2022. The 1000 simulated cases (blue) are generated from the proposed CRESCENT model with the contemporary grid configuration. The observed peak outage representing the degradation of customers with electricity (black) is obtained from the US power outage datasetsPoweroutageus recorded by local power utilities. The right-side subfigure shows the distribution of final system statuses (percentage of customers with electricity) across all simulations. The top subfigure shows the distribution of the times when catastrophic blackouts (100% failure) occurred across all catastrophic blackout cases. b, Distribution of the largest failures. The largest failure in each power outage realization is identified by the largest drop in the percentage of customers with electricity between two successive simulation time steps. A darker color in a hexagon indicates a higher density (frequency) of data points within that area. The top and right histograms show the distributions for times and system statuses, respectively, where the largest cascading failures occurred. Black stars in the top and right distributions mark the position of the observed case.
• Figure 4: Resilience and vulnerability patterns. a, Resilience pattern. The kernel density estimation of the largest failure (the most significant drop in system performance) in each resilient case (where the grid survives without a complete blackout) among the 1000 realizations. The darker color (blue) represents areas where the largest failures occur in the majority of resilient cases. b, Vulnerability pattern. Similar to (a) but for vulnerable cases (where the grid suffers a catastrophic blackout) among the 1000 realizations. c, Identified critical transmission lines in the transmission network. The critical index for a transmission line is defined as the proportion of instances where its failures directly contribute to catastrophic blackouts (a 100% complete outage occurs as soon as the line fails) across all realizations. Only those lines with a critical index higher than 5% are highlighted (blue), while the remaining lines are colored in grey. The names of substations associated with these critical lines are labeled. d, Resilience and vulnerability patterns of the top four critical transmission lines. The resilience pattern boxplots (blue) show the distributions of line failure times in instances where the system ultimately remains functional despite the failure of the corresponding lines. The vulnerability pattern boxplots (red) show similar distributions in instances where a catastrophic blackout occurs immediately following the failure of the specific lines. Tiny grey dots represent individual data points. The horizontal black line within each box indicates the median, box edges show the interquartile (50%) range, and whiskers extend to the 5th and 95th percentiles.
• Figure 5: Sensitivity analysis of renewable energy integration on catastrophic blackouts under the same hurricane event. a, Probability of catastrophic blackout occurrence at varying levels of renewable energy integration. The current grid (contemporary during Hurricane Fiona) with an average renewable integration level of 16.1% is marked by the red dot. textbfb, Violin plots for time distributions of catastrophic blackouts at varying levels of renewable energy integration. textbfc, Distributions of catastrophic blackout occurrences at different renewable integration levels. Each data point in the plots records the time of a catastrophic blackout occurrence and the corresponding system status (percentage of customers with electricity). A darker color in a hexagon indicates a higher density (frequency) of data points within that area. The probability for each blue point in (a) and the violin plots in (b) were calculated based on 1000 realizations generated by the proposed CRESCENT model at a specific renewable integration level under Hurricane Fiona. In the violin plot, each violin represents the distribution of blackout occurrence times at a specific level of renewable energy integration. The width of each violin indicates the frequency (probability) of blackouts at different times. The black bar within each violin shows the interquartile range, and the inside white dot represents the median time of catastrophic blackout occurrence. A larger violin size in (b) indicates more catastrophic blackouts.