Search

The resilience of electric power system is critical to economic prosperity, as well as public health and safety. In the Northeastern United States, where severe weather events occur throughout the year, it is important to quantify the weather-induced power outages and understand the impact of climate change on power system resilience in the future. This study focuses on assessing the future resilience of the power grid for a service territory in the Northeastern United States, by developing a holistic framework that employs high-resolution climate data along with the power outage prediction models (OPMs). The OPMs are a group of machine-learning based models designed to predict the amount of power outages for different types of weather events. The climate data used in this study are based on selected General Circulation Model (GCM) products that follow two Representative Concentration Pathways (namely RCP 4.5 and RCP 8.5), which are statistically bias-corrected over an 18-km grid. The climate data are, then, used to identify and classify the possible weather-related outage events and quantify associated number of power outages by OPMs. The results aim to, i) quantify the severity and frequency of the occurrence of weather-induced power outages in the next 40 years (2020-2060); and ii) provide a basis for a comparison of the difference of power outages under the two RCP scenarios and the current climate conditions. This information can be useful for making decisions on power grid strengthening plans that accommodate potential future climate change impacts on the power grid resilience.

Weather-related power outages affect millions of utility customers every year. Predicting storm outages with lead times of up to five days could help utilities to allocate crews and resources and devise cost-effective restoration plans that meet the strict time and efficiency requirements imposed by regulatory authorities. In this study, we construct a numerical experiment to evaluate how weather parameter uncertainty, based on weather forecasts with one to five days of lead time, propagates into outage prediction error. We apply a machine-learning-based outage prediction model on storm-caused outage events that occurred between 2016 and 2019 in the northeastern United States. The model predictions, fed by weather analysis and other environmental parameters including land cover, tree canopy, vegetation characteristics, and utility infrastructure variables exhibited a mean absolute percentage error of 38%, Nash–Sutcliffe efficiency of 0.54, and normalized centered root mean square error of 68%. Our numerical experiment demonstrated that uncertainties of precipitation and wind-gust variables play a significant role in the outage prediction uncertainty while sustained wind and temperature parameters play a less important role. We showed that, while the overall weather forecast uncertainty increases gradually with lead time, the corresponding outage prediction uncertainty exhibited a lower dependence on lead times up to 3 days and a stepwise increase in the four- and five-day lead times.

The increasing frequency and intensity of high impact storms, especially in Northeast United States, requires utilities and emergency managers to be increasingly prepared for lengthy power outage restorations. Historically, restoration has relied on emergency managers decennial experience with limited access to predictive models. This study highlights the development of a combined system composed of the UConn Outage Prediction Model (OPM) for predicting weather-related damage in the distribution system and an Agent-Based Model (ABM) for estimating the time to electric power restoration. The combined system is validated using Outage Management System (OMS) and crew deployment information for four historical extreme weather events that occurred in the State of Connecticut in the past decade. Through the ABM’s ability to test different restoration strategies, we study the impact that human knowledge and decisions have on the outage restoration curve. Furthermore, we use the model to test how the restoration could have been different if crews were allocated to area work centers based on the location of damage predictions from the UConn OPM and on increased crew counts, reflecting a more aggressive storm preparedness. This test highlights how an OPM-ABM system can benefit emergency preparedness and response managers in advance of storms impact.

One method of measuring precipitation and wind over the ocean is through analysis of the underwater ambient acoustics. In this study, the ambient ocean noises recorded by a passive aquatic listener (PAL) in the Mediterranean are used to compare the effectiveness of the machine learning techniques for measuring the wind speed and precipitation rate with the empirical methods from previous works. The data were collected over the timeframe of June 2011 to May 2012 and included two storms that caused severe coastal flooding in Genoa, Italy. A spar buoy at the surface above the PAL provided high-quality in situ measurements to act as the reference data for model training and validation. The results using the machine learning models show correlation coefficients of 0.95 between the acoustic data and wind speed and a reduction in unexplained variance by over a third from previous methods. For precipitation, CatBoost and random forest models are used to measure the total precipitation for 12 events using leave-one-event-out cross-validation, demonstrating mean errors of 28% and 34% and median errors of 18% and 17%, respectively. The ability to measure wind and precipitation by applying machine learning on data from underwater acoustic recorders shows potential to help improve in situ measurements over oceans globally.

In this communication, we present application of the automated near-real-time (NRT) system called RAdar-Produced Inundation Diary (RAPID) to European Space Agency Sentinel-1 synthetic aperture radar (SAR) images to produce flooding maps for Hurricane Dorian in the northern Bahamas. RAPID maps, released 2 d after the event, show that coastal flooding in the Bahamas reached areas located more than 10 km inland, covering more than 3000 km2 of continental area. RAPID flood estimates from subsequent SAR images show the recession of the flood across the islands and present high agreement scores when compared to Copernicus Emergency Management Service (Copernicus EMS) estimates.

Thunderstorms are complex weather phenomena that cause substantial power outages in a short period. This makes thunderstorm outage prediction challenging using eventwise outage prediction models (OPMs), which summarize the storm dynamics over the entire course of the storm into a limited number of parameters. We developed a new, temporally sensitive outage prediction framework designed for models to learn the hourly dynamics of thunderstorm-caused outages directly from weather forecasts. Validation of several models built on this hour-by-hour prediction framework and comparison with a baseline model show abilities to accurately report temporal and storm-wide outage characteristics, which are vital for planning utility responses to storm-caused power grid damage.

This paper develops a statistical framework to analyze the effectiveness of vegetation management at reducing power outages during storms of varying severity levels. The framework was applied on the Eversource Energy distribution grid in Connecticut, USA based on 173 rain and wind events from 2005–2020, including Hurricane Irene, Hurricane Sandy, and Tropical Storm Isaias. The data were binned by storm severity (high/low) and vegetation management levels, where a maximum applicable length of vegetation management for each circuit was determined, and the data were divided into four bins based on the actual length of vegetation management performed divided by the maximum applicable value (0–25%, 25–50%, 50–75%, and 75–100%). Then, weather and overhead line length normalized outage statistics were taken for each group. The statistics were used to determine the effectiveness of vegetation management and its dependence on storm severity. The results demonstrate a higher reduction in damages for lower-severity storms, with a reduction in normalized outages between 45.8% and 63.8%. For high-severity events, there is a large increase in effectiveness between the highest level of vegetation management and the two lower levels, with 75–100% vegetation management leading to a 37.3% reduction in trouble spots. Yet, when evaluating system reliability, it is important to look at all storms combined, and the results of this study provide useful information on total annual trouble spots and allow for analysis of how various vegetation management scenarios would impact trouble spots in the electric grid. This framework can also be used to better understand how more rigorous vegetation management standards (applying ETT) help reduce outages at an individual event level. In future work, a similar framework may be used to evaluate other resilience improvements.

During the night between 9 and 10 September 2017, multiple flash floods associated with a heavy-precipitation event affected the town of Livorno, located in Tuscany, Italy. Accumulated precipitation exceeding 200 mm in two hours was recorded. This rainfall intensity is associated with a return period of higher than 200 years. As a consequence, all the largest streams of the Livorno municipality flooded several areas of the town. We used the limited-area weather research and forecasting (WRF) model, in a convection-permitting setup, to reconstruct the extreme event leading to the flash floods. We evaluated possible forecasting improvements emerging from the assimilation of local ground stations and X- and S-band radar data into the WRF, using the configuration operational at the meteorological center of Tuscany region (LaMMA) at the time of the event. Simulations were verified against weather station observations, through an innovative method aimed at disentangling the positioning and intensity errors of precipitation forecasts. A more accurate description of the low-level flows and a better assessment of the atmospheric water vapor field showed how the assimilation of radar data can improve quantitative precipitation forecasts.

Every year millions of people in the US are affected by power outages, disrupting the economy and daily life. Many of these outages are caused by events such as strong winds, heavy rains, thunderst...

In this manuscript, we discuss the importance of using accurate weather forecasts based on remote sensing observations for improving the accuracy of electric distribution grid power outage prediction. Weather variable accuracy is critical to limit the significant power outage modeling errors arising from the strong non-linearities in the relationships between hydrometeorological variables and power outages. We consolidate recent findings related to the most important variables for outage prediction, and to the propagation of their uncertainties in the University of Connecticut Outage Prediction Model (UConn-OPM). UConn-OPM is an operational machine learning-based framework predicting power outage in electric distribution networks. Combining remote sensing datasets with weather analysis used for OPM training, and post-storm verification, would directly contribute to improving the OPM prediction accuracy.

A growing number of electricity utilities use machine learning-based outage prediction models (OPMs) to predict the impact of storms on their networks for sustainable management. The accuracy of OPM predictions is sensitive to sample size and event severity representativeness in the training dataset, the extent of which has not yet been quantified. This study devised a randomized and out-of-sample validation experiment to quantify an OPM&rsquo, s prediction uncertainty to different training sample sizes and event severity representativeness. The study showed random error decreasing by more than 100% for sample sizes ranging from 10 to 80 extratropical events, and by 32% for sample sizes from 10 to 40 thunderstorms.&nbsp, This study quantified the minimum number of sample size for the OPM attaining an acceptable prediction performance. The results demonstrated that conditioning the training of the OPM to a subset of events representative of the predicted event&rsquo, s severity reduced the underestimation bias exhibited in high-impact events and the overestimation bias in low-impact ones. We used cross entropy (CE) to quantify the relatedness of weather variable distribution between the training dataset and the forecasted event.

Hurricane Sandy (2012, referred to as Current Sandy) was among the most devastating storms to impact Connecticut’s overhead electric distribution network, resulting in over 15 000 outage locations that affected more than 500 000 customers. In this paper, the severity of tree-caused outages in Connecticut is estimated under future-climate Hurricane Sandy simulations, each exhibiting strengthened winds and heavier rain accumulation over the study area from large-scale thermodynamic changes in the atmosphere and track changes in the year ~2100 (referred to as Future Sandy). Three machine-learning models used five weather simulations and the ensemble mean of Current and Future Sandy, along with land-use and overhead utility infrastructure data, to predict the severity and spatial distribution of outages across the Eversource Energy service territory in Connecticut. To assess the influence of increased precipitation from Future Sandy, two approaches were compared: an outage model fit with a full set of variables accounting for both wind and precipitation, and a reduced set with only wind. Future Sandy displayed an outage increase of 42%–64% when using the ensemble of WRF simulations fit with three different outage prediction models. This study is a proof of concept for the assessment of increased outage risk resulting from potential changes in tropical cyclone intensity associated with late-century thermodynamic changes driven by the IPCC AR4 A2 emissions scenario.

Lack of real-time, in situ data on the extent of flooding in many parts of the world can hinder efficient disaster response. With the advent of satellite-based synthetic aperture radar (SAR) sensors, we can deploy techniques to identify flooded areas worldwide while storms are occurring. In this communication, we present an automated near-real-time (NRT) system called RAdar-Produced Inundation Diary (RAPID), applying it to European Space Agency Sentinel-1 SAR images to produce flooding maps for Hurricane Dorian in the northern Bahamas. Images from RAPID released two days after the event show coastal flooding in the Bahamas reached areas located more than 10 km inland, covering more than 3,000 km2 of continental area. RAPID flood estimates from subsequent SAR images show the recession of the flood across the islands.

This paper describes the development of two Outage Prediction Models (OPMs) for power outages caused by snow and ice storms on electric distribution networks, and their performance evaluation in the Northeastern United States. The first is a Machine Learning (ML) based model that is set up to predict power outages on a regular grid (4-km grid spacing). The second is a Generalized Linear Model (GLM) that is set up to predict total power outages at the town level. Both models are fed with Numerical Weather Prediction (NWP) outputs, satellite-derived leaf area index (LAI) and land cover data, and utility specified infrastructure and historical outage data. For both models the most important variables are the amount of assets on the territory, LAI, snow density, and – for the ice model – the amount of freezing rain. Results from cross-validation experiments based on 54 outage events spanning three orders of magnitude show median absolute percentage errors around 70% for both models. The GLM is able to predict extreme events (such as the devastating October 2011 nor’easter) better than ML models do, while ML models have better performance for lower impact events and present better characterization of the spatial distribution of power outages.

Enhanced Tree Trimming (ETT) is a vegetation management standard consisting in the trimming or removal of trees in proximity of overhead lines. We quantify the effects of this practice, designed to improve the resiliency of the overhead electrical power grid, using two independent methodologies. The first approach is a statistical study of the change of frequency of outage-free locations. The second approach uses an Outage Prediction Model (OPM) as a vulnerability assessment tool to evaluate the change in the number of outages before and after ETT. The OPM is a machine learning based framework that relates weather, soil, vegetation and electric grid characteristics to storm-related power outages. The two methods introduced in this work are compared against a straightforward assessment of the ETT impact on the number of outages. Applying both methods, the occurrence of power outages is studied for varying tree trimming amounts, for 144 storms occurred between 2005 and 2017 in the Northeastern United States. From the statistical approach we find a reduction of the outage-free grid cells during storms between 49% and 65% due to tree trimming. The OPM based analysis suggests that the number of outages during storms are reduced between 16% and 48% after performing ETT.

Two new variables have been introduced to better identify the potential vorticity (PV) anomalies due to the intrusion of dry stratospheric air from those induced by the diabatic latent heating. This new approach has been applied to the analysis of three Mediterranean tropical-like cyclones characterized by heavy precipitation patterns. Model simulations show that the interaction between an upper level PV streamer, located on the left exit of a jet stream and a middle-low level PV anomaly, induced by the convection development around the low level vortex, plays a key role in the intensification of cyclones in all cases. These anomalies, despite their strong mutual interaction, do not form a fully developed PV tower. In the mature stage, the shape of the upper level PV anomaly around the cyclone is different for each case and appears somehow dependent on the lifetime of the vortex. A first comparison with satellite-derived products seems to confirm the initial results from model simulations.