Prestorm Estimation of Hurricane Damage to Electric Power Distribution Systems

Hurricanes frequently cause damage to electric power systems in the United States, leading to widespread and prolonged loss of electric service. Restoring service quickly requires the use of repair crews and materials that must be requested, at considerable cost, prior to the storm. U.S. utilities have struggled to strike a good balance between over‐ and underpreparation largely because of a lack of methods for rigorously estimating the impacts of an approaching hurricane on their systems. Previous work developed methods for estimating the risk of power outages and customer loss of power, with an outage defined as nontransitory activation of a protective device. In this article, we move beyond these previous approaches to directly estimate damage to the electric power system. Our approach is based on damage data from past storms together with regression and data mining techniques to estimate the number of utility poles that will need to be replaced. Because restoration times and resource needs are more closely tied to the number of poles and transformers that need to be replaced than to the number of outages, this pole‐based assessment provides a much stronger basis for prestorm planning by utilities. Our results show that damage to poles during hurricanes can be assessed accurately, provided that adequate past damage data are available. However, the availability of data can, and currently often is, the limiting factor in developing these types of models in practice. Opportunities for further enhancing the damage data recorded during hurricanes are also discussed.     

Hurricane Isaac: A Longitudinal Analysis of Storm Characteristics and Power Outage Risk

In August 2012, Hurricane Isaac, a Category 1 hurricane at landfall, caused extensive power outages in Louisiana. The storm brought high winds, storm surge, and flooding to Louisiana, and power outages were widespread and prolonged. Hourly power outage data for the state of Louisiana were collected during the storm and analyzed. This analysis included correlation of hourly power outage figures by zip code with storm conditions including wind, rainfall, and storm surge using a nonparametric ensemble data mining approach. Results were analyzed to understand how correlation of power outages with storm conditions differed geographically within the state. This analysis provided insight on how rainfall and storm surge, along with wind, contribute to power outages in hurricanes. By conducting a longitudinal study of outages at the zip code level, we were able to gain insight into the causal drivers of power outages during hurricanes. Our analysis showed that the statistical importance of storm characteristic covariates to power outages varies geographically. For Hurricane Isaac, wind speed, precipitation, and previous outages generally had high importance, whereas storm surge had lower importance, even in zip codes that experienced significant surge. The results of this analysis can inform the development of power outage forecasting models, which often focus strictly on wind‐related covariates. Our study of Hurricane Isaac indicates that inclusion of other covariates, particularly precipitation, may improve model accuracy and robustness across a range of storm conditions and geography.     

Explainable deep learning powered building risk assessment model for proactive hurricane response

Climate change and rapid urban development have intensified the impact of hurricanes, especially on the Southeastern Coasts of the United States. Localized and timely risk assessments can facilitate coastal communities’ preparedness and response to imminent hurricanes. Existing assessment methods focused on hurricane risks at large spatial scales, which were not specific or could not provide actionable knowledge for residents or property owners. Fragility functions and other widely utilized assessment methods cannot model the complex relationships between building features and hurricane risk levels effectively. Therefore, we develop and test a building-level hurricane risk assessment with deep feedforward neural network (DFNN) models. The input features of DFNN models cover the meta building characteristics, fine-grained meteorological, and hydrological environmental parameters. The assessment outcomes, that is, risk levels, include the probability and intensity of building/property damages induced by wind and surge hazards. We interpret the DFNN models with local interpretable model-agnostic explanations (LIME). We apply the DFNN models to a case building in Cameron County, Louisiana in response to a hypothetical imminent hurricane to illustrate how the building's risk levels can be timely assessed with the updating weather forecast. This research shows the potential of deep-learning models in integrating multi-sourced features and accurately predicting buildings’ risks of weather extremes for property owners and households. The AI-powered risk assessment model can help coastal populations form appropriate and updating perceptions of imminent hurricanes and inform actionable knowledge for proactive risk mitigation and long-term climate adaptation.     

Power Outage Estimation for Tropical Cyclones: Improved Accuracy with Simpler Models

In this article, we discuss an outage‐forecasting model that we have developed. This model uses very few input variables to estimate hurricane‐induced outages prior to landfall with great predictive accuracy. We also show the results for a series of simpler models that use only publicly available data and can still estimate outages with reasonable accuracy. The intended users of these models are emergency response planners within power utilities and related government agencies. We developed our models based on the method of random forest, using data from a power distribution system serving two states in the Gulf Coast region of the United States. We also show that estimates of system reliability based on wind speed alone are not sufficient for adequately capturing the reliability of system components. We demonstrate that a multivariate approach can produce more accurate power outage predictions.     

Evaluating and Visualizing the Economic Impact of Commercial Districts Due to an Electric Power Network Disruption

Critical infrastructure networks enable social behavior, economic productivity, and the way of life of communities. Disruptions to these cyber–physical–social networks highlight their importance. Recent disruptions caused by natural phenomena, including Hurricanes Harvey and Irma in 2017, have particularly demonstrated the importance of functioning electric power networks. Assessing the economic impact (EI) of electricity outages after a service disruption is a challenging task, particularly when interruption costs vary by the type of electric power use (e.g., residential, commercial, industrial). In contrast with most of the literature, this work proposes an approach to spatially evaluate EIs of disruptions to particular components of the electric power network, thus enabling resilience‐based preparedness planning from economic and community perspectives. Our contribution is a mix‐method approach that combines EI evaluation, component importance analysis, and GIS visualization for decision making. We integrate geographic information systems and an economic evaluation of sporadic electric power outages to provide a tool to assist with prioritizing restoration of power in commercial areas that have the largest impact. By making use of public data describing commercial market value, gross domestic product, and electric area distribution, this article proposes a method to evaluate the EI experienced by commercial districts. A geospatial visualization is presented to observe and compare the areas that are more vulnerable in terms of EI based on the areas covered by each distribution substation. Additionally, a heat map is developed to observe the behavior of disrupted substations to determine the important component exhibiting the highest EI. The proposed resilience analytics approach is applied to analyze outages of substations in the boroughs of New York City.     

Integrating Models and Data to Estimate the Structural Reliability of Utility Poles During Hurricanes

Utility systems such as power and communication systems regularly experience significant damage and loss of service during hurricanes. A primary damage mode for these systems is failure of wooden utility poles that support conductors and communication lines. In this article, we present an approach for combining structural reliability models for utility poles with observed data on pole performance during past hurricanes. This approach, based on Bayesian updating, starts from an imperfect but informative prior and updates this prior with observed performance data. We consider flexural and foundation failure mechanisms in the prior, acknowledging that these are an incomplete, but still informative, subset of the possible failure mechanisms for utility poles during hurricanes. We show how a model‐based prior can be updated with observed failure data, using pole failure data from Hurricane Katrina as a case study. The results of this integration of model‐based estimates and observed performance data then offer a more informative starting point for power system performance estimation for hurricane conditions.     

Individual learning as a driver of changes in community vulnerability under repeated hurricanes and changing climate

The risks from singular natural hazards such as a hurricane have been extensively investigated in the literature. However, little is understood about how individual and collective responses to repeated hazards change communities and impact their preparation for future events. Individual mitigation actions may drive how a community's resilience evolves under repeated hazards. In this paper, we investigate the effect that learning by homeowners can have on household mitigation decisions and on how this influences a region's vulnerability to natural hazards over time, using hurricanes along the east coast of the United States as our case study. To do this, we build an agent-based model (ABM) to simulate homeowners’ adaptation to repeated hurricanes and how this affects the vulnerability of the regional housing stock. Through a case study, we explore how different initial beliefs about the hurricane hazard and how the memory of recent hurricanes could change a community's vulnerability both under current and potential future hurricane scenarios under climate change. In some future hurricane environments, different initial beliefs can result in large differences in the region's long-term vulnerability to hurricanes. We find that when some homeowners mitigate soon after a hurricane—when their memory of the event is the strongest—it can help to substantially decrease the vulnerability of a community.     

Nonparametric Tree‐Based Predictive Modeling of Storm Outages on an Electric Distribution Network

This article compares two nonparametric tree‐based models, quantile regression forests (QRF) and Bayesian additive regression trees (BART), for predicting storm outages on an electric distribution network in Connecticut, USA. We evaluated point estimates and prediction intervals of outage predictions for both models using high‐resolution weather, infrastructure, and land use data for 89 storm events (including hurricanes, blizzards, and thunderstorms). We found that spatially BART predicted more accurate point estimates than QRF. However, QRF produced better prediction intervals for high spatial resolutions (2‐km grid cells and towns), while BART predictions aggregated to coarser resolutions (divisions and service territory) more effectively. We also found that the predictive accuracy was dependent on the season (e.g., tree‐leaf condition, storm characteristics), and that the predictions were most accurate for winter storms. Given the merits of each individual model, we suggest that BART and QRF be implemented together to show the complete picture of a storm's potential impact on the electric distribution network, which would allow for a utility to make better decisions about allocating prestorm resources.     

Quantifying the Hurricane Catastrophe Risk to Offshore Wind Power

The U.S. Department of Energy has estimated that over 50 GW of offshore wind power will be required for the United States to generate 20% of its electricity from wind. Developers are actively planning offshore wind farms along the U.S. Atlantic and Gulf coasts and several leases have been signed for offshore sites. These planned projects are in areas that are sometimes struck by hurricanes. We present a method to estimate the catastrophe risk to offshore wind power using simulated hurricanes. Using this method, we estimate the fraction of offshore wind power simultaneously offline and the cumulative damage in a region. In Texas, the most vulnerable region we studied, 10% of offshore wind power could be offline simultaneously because of hurricane damage with a 100‐year return period and 6% could be destroyed in any 10‐year period. We also estimate the risks to single wind farms in four representative locations; we find the risks are significant but lower than those estimated in previously published results. Much of the hurricane risk to offshore wind turbines can be mitigated by designing turbines for higher maximum wind speeds, ensuring that turbine nacelles can turn quickly to track the wind direction even when grid power is lost, and building in areas with lower risk.     

Comparison and Validation of Statistical Methods for Predicting Power Outage Durations in the Event of Hurricanes

This article compares statistical methods for modeling power outage durations during hurricanes and examines the predictive accuracy of these methods. Being able to make accurate predictions of power outage durations is valuable because the information can be used by utility companies to plan their restoration efforts more efficiently. This information can also help inform customers and public agencies of the expected outage times, enabling better collective response planning, and coordination of restoration efforts for other critical infrastructures that depend on electricity. In the long run, outage duration estimates for future storm scenarios may help utilities and public agencies better allocate risk management resources to balance the disruption from hurricanes with the cost of hardening power systems. We compare the out‐of‐sample predictive accuracy of five distinct statistical models for estimating power outage duration times caused by Hurricane Ivan in 2004. The methods compared include both regression models (accelerated failure time (AFT) and Cox proportional hazard models (Cox PH)) and data mining techniques (regression trees, Bayesian additive regression trees (BART), and multivariate additive regression splines). We then validate our models against two other hurricanes. Our results indicate that BART yields the best prediction accuracy and that it is possible to predict outage durations with reasonable accuracy.     

Uncertainty Analysis for Computer Model Projections of Hurricane Losses

Projecting losses associated with hurricanes is a complex and difficult undertaking that is wrought with uncertainties. Hurricane Charley, which struck southwest Florida on August 13, 2004, illustrates the uncertainty of forecasting damages from these storms. Due to shifts in the track and the rapid intensification of the storm, real-time estimates grew from 2 to 3 billion dollars in losses late on August 12 to a peak of 50 billion dollars for a brief time as the storm appeared to be headed for the Tampa Bay area. The storm hit the resort areas of Charlotte Harbor near Punta Gorda and then went on to Orlando in the central part of the state, with early poststorm estimates converging on a damage estimate in the 28 to 31 billion dollars range. Comparable damage to central Florida had not been seen since Hurricane Donna in 1960. The Florida Commission on Hurricane Loss Projection Methodology (FCHLPM) has recognized the role of computer models in projecting losses from hurricanes. The FCHLPM established a professional team to perform onsite (confidential) audits of computer models developed by several different companies in the United States that seek to have their models approved for use in insurance rate filings in Florida. The team's members represent the fields of actuarial science, computer science, meteorology, statistics, and wind and structural engineering. An important part of the auditing process requires uncertainty and sensitivity analyses to be performed with the applicant's proprietary model. To influence future such analyses, an uncertainty and sensitivity analysis has been completed for loss projections arising from use of a Holland B parameter hurricane wind field model. Uncertainty analysis quantifies the expected percentage reduction in the uncertainty of wind speed and loss that is attributable to each of the input variables.     

Disasters as Learning Experiences or Disasters as Policy Opportunities? Examining Flood Insurance Purchases after Hurricanes

Flood insurance is a critical risk management strategy, contributing to greater resilience of individuals and communities. The occurrence of disasters has been observed to alter risk management choices, including the decision to insure. This has previously been explained by learning and behavioral biases. When it comes to flood insurance, however, federal disaster aid policy could also play a role since recipients of aid are required to maintain insurance. Using a database of flood insurance policies for all states on the Atlantic and Gulf coasts of the United States between 2001 and 2010, this article uses fixed effects models to examine how take‐up rates respond to the occurrence of hurricanes and tropical storms, as well as disaster declarations and aid requirements. Being hit by at least one hurricane in the previous year increases net flood insurance purchases by 7.2%. This effect dies out by three years after the storm. A presidential disaster declaration for floods increases take‐up rates by 6.7%. When disaster aid grants are made available to households, take‐up rates increase by 5%; this accounts for the majority of the increase in policies after occurrence of a hurricane. When the models are estimated taking into account which policies are required by disaster aid, hurricanes are estimated to lead to only a 1.5% increase in voluntary purchases. This overlooked federal policy that disaster aid recipients insure is responsible for a majority of insurance purchases postdisaster.     

Risk-Management and Risk-Analysis-Based Decision Tools for Attacks on Electric Power

Incident data about disruptions to the electric power grid provide useful information that can be used as inputs into risk management policies in the energy sector for disruptions from a variety of origins, including terrorist attacks. This article uses data from the Disturbance Analysis Working Group (DAWG) database, which is maintained by the North American Electric Reliability Council (NERC), to look at incidents over time in the United States and Canada for the period 1990-2004. Negative binomial regression, logistic regression, and weighted least squares regression are used to gain a better understanding of how these disturbances varied over time and by season during this period, and to analyze how characteristics such as number of customers lost and outage duration are related to different characteristics of the outages. The results of the models can be used as inputs to construct various scenarios to estimate potential outcomes of electric power outages, encompassing the risks, consequences, and costs of such outages.     

Statistical Modeling of Fire Occurrence Using Data from the Tōhoku,Japan Earthquake and Tsunami

In this article, we develop statistical models to predict the number and geographic distribution of fires caused by earthquake ground motion and tsunami inundation in Japan. Using new, uniquely large, and consistent data sets from the 2011 Tōhoku earthquake and tsunami, we fitted three types of models—generalized linear models (GLMs), generalized additive models (GAMs), and boosted regression trees (BRTs). This is the first time the latter two have been used in this application. A simple conceptual framework guided identification of candidate covariates. Models were then compared based on their out‐of‐sample predictive power, goodness of fit to the data, ease of implementation, and relative importance of the framework concepts. For the ground motion data set, we recommend a Poisson GAM; for the tsunami data set, a negative binomial (NB) GLM or NB GAM. The best models generate out‐of‐sample predictions of the total number of ignitions in the region within one or two. Prefecture‐level prediction errors average approximately three. All models demonstrate predictive power far superior to four from the literature that were also tested. A nonlinear relationship is apparent between ignitions and ground motion, so for GLMs, which assume a linear response‐covariate relationship, instrumental intensity was the preferred ground motion covariate because it captures part of that nonlinearity. Measures of commercial exposure were preferred over measures of residential exposure for both ground motion and tsunami ignition models. This may vary in other regions, but nevertheless highlights the value of testing alternative measures for each concept. Models with the best predictive power included two or three covariates.     

Hurricane Destructive Power Predictions Based on Historical Storm and Sea Surface Temperature Data

Forecasting destructive hurricane potential is complicated by substantial, unexplained intraannual variation in storm-specific power dissipation index (PDI, or integrated third power of wind speed), and interannual variation in annual accumulated PDI (APDI). A growing controversy concerns the recent hypothesis that the clearly positive trend in North Atlantic Ocean (NAO) sea surface temperature (SST) since 1970 explains increased hurricane intensities over this period, and so implies ominous PDI and APDI growth as global warming continues. To test this "SST hypothesis" and examine its quantitative implications, a combination of statistical and probabilistic methods were applied to National Hurricane Center HURDAT best-track data on NAO hurricanes during 1880-2002, and corresponding National Oceanographic and Atmospheric Administration Extended Reconstruction SST estimates. Notably, hurricane behavior was compared to corresponding hurricane-specific (i.e., spatiotemporally linked) SST; previous similar comparisons considered only SST averaged over large NAO regions. Contrary to the SST hypothesis, SST was found to vary in a monthly pattern inconsistent with that of corresponding PDI, and to be at best weakly associated with PDI or APDI despite strong correlation with corresponding mean latitude (R(2)= 0.55) or with combined mean location and a approximately 90-year periodic trend (R(2)= 0.70). Over the last century, the lower 75% of APDIs appear randomly sampled from a nearly uniform distribution, and the upper 25% of APDIs from a nearly lognormal distribution. From the latter distribution, a baseline (SST-independent) stochastic model was derived predicting that over the next half century, APDI will not likely exceed its maximum value over the last half century by more than a factor of 1.5. This factor increased to 2 using a baseline model modified to assume SST-dependence conditioned on an upper bound of the increasing NAO SST trend observed since 1970. An additional model was developed that predicts PDI statistics conditional on APDI. These PDI and APDI models can be used to estimate upper bounds on indices of hurricane power likely to be realized over the next century, under divergent assumptions regarding SST influence.     

Quantifying Community Resilience Using Hierarchical Bayesian Kernel Methods: A Case Study on Recovery from Power Outages

The ability to accurately measure recovery rate of infrastructure systems and communities impacted by disasters is vital to ensure effective response and resource allocation before, during, and after a disruption. However, a challenge in quantifying such measures resides in the lack of data as community recovery information is seldom recorded. To provide accurate community recovery measures, a hierarchical Bayesian kernel model (HBKM) is developed to predict the recovery rate of communities experiencing power outages during storms. The performance of the proposed method is evaluated using cross‐validation and compared with two models, the hierarchical Bayesian regression model and the Poisson generalized linear model. A case study focusing on the recovery of communities in Shelby County, Tennessee after severe storms between 2007 and 2017 is presented to illustrate the proposed approach. The predictive accuracy of the models is evaluated using the log‐likelihood and root mean squared error. The HBKM yields on average the highest out‐of‐sample predictive accuracy. This approach can help assess the recoverability of a community when data are scarce and inform decision making in the aftermath of a disaster. An illustrative example is presented demonstrating how accurate measures of community resilience can help reduce the cost of infrastructure restoration.     

Process Tracing Analysis of Hurricane Information Displays

To study people's processing of hurricane forecast advisories, we conducted a computer‐based experiment that examined 11 research questions about the information seeking patterns of students assuming the role of a county emergency manager in a sequence of six hurricane forecast advisories for each of four different hurricanes. The results show that participants considered a variety of different sources of information—textual, graphic, and numeric—when tracking hurricanes. Click counts and click durations generally gave the same results but there were some significant differences. Moreover, participants’ information search strategies became more efficient over forecast advisories and with increased experience tracking the four hurricanes. These changes in the search patterns from the first to the fourth hurricane suggest that the presentation of abstract principles in a training manual was not sufficient for them to learn how to track hurricanes efficiently but they were able to significantly improve their search efficiency with a modest amount (roughly an hour) of practice. Overall, these data indicate that information search patterns are complex and deserve greater attention in studies of dynamic decision tasks.     

An Integrated Scenario Ensemble-Based Framework for Hurricane Evacuation Modeling: Part 2—Hazard Modeling

Hurricane track and intensity can change rapidly in unexpected ways, thus making predictions of hurricanes and related hazards uncertain. This inherent uncertainty often translates into suboptimal decision-making outcomes, such as unnecessary evacuation. Representing this uncertainty is thus critical in evacuation planning and related activities. We describe a physics-based hazard modeling approach that (1) dynamically accounts for the physical interactions among hazard components and (2) captures hurricane evolution uncertainty using an ensemble method. This loosely coupled model system provides a framework for probabilistic water inundation and wind speed levels for a new, risk-based approach to evacuation modeling, described in a companion article in this issue. It combines the Weather Research and Forecasting (WRF) meteorological model, the Coupled Routing and Excess STorage (CREST) hydrologic model, and the ADvanced CIRCulation (ADCIRC) storm surge, tide, and wind-wave model to compute inundation levels and wind speeds for an ensemble of hurricane predictions. Perturbations to WRF's initial and boundary conditions and different model physics/parameterizations generate an ensemble of storm solutions, which are then used to drive the coupled hydrologic + hydrodynamic models. Hurricane Isabel (2003) is used as a case study to illustrate the ensemble-based approach. The inundation, river runoff, and wind hazard results are strongly dependent on the accuracy of the mesoscale meteorological simulations, which improves with decreasing lead time to hurricane landfall. The ensemble envelope brackets the observed behavior while providing “best-case” and “worst-case” scenarios for the subsequent risk-based evacuation model.     

Critical Assessment of the Foundations of Power Transmission and Distribution Reliability Metrics and Standards

The U.S. federal government regulates the reliability of bulk power systems, while the reliability of power distribution systems is regulated at a state level. In this article, we review the history of regulating electric service reliability and study the existing reliability metrics, indices, and standards for power transmission and distribution networks. We assess the foundations of the reliability standards and metrics, discuss how they are applied to outages caused by large exogenous disturbances such as natural disasters, and investigate whether the standards adequately internalize the impacts of these events. Our reflections shed light on how existing standards conceptualize reliability, question the basis for treating large‐scale hazard‐induced outages differently from normal daily outages, and discuss whether this conceptualization maps well onto customer expectations. We show that the risk indices for transmission systems used in regulating power system reliability do not adequately capture the risks that transmission systems are prone to, particularly when it comes to low‐probability high‐impact events. We also point out several shortcomings associated with the way in which regulators require utilities to calculate and report distribution system reliability indices. We offer several recommendations for improving the conceptualization of reliability metrics and standards. We conclude that while the approaches taken in reliability standards have made considerable advances in enhancing the reliability of power systems and may be logical from a utility perspective during normal operation, existing standards do not provide a sufficient incentive structure for the utilities to adequately ensure high levels of reliability for end‐users, particularly during large‐scale events.     

Low‐Probability Flood Risk Modeling for New York City

The devastating impact by Hurricane Sandy (2012) again showed New York City (NYC) is one of the most vulnerable cities to coastal flooding around the globe. The low‐lying areas in NYC can be flooded by nor'easter storms and North Atlantic hurricanes. The few studies that have estimated potential flood damage for NYC base their damage estimates on only a single, or a few, possible flood events. The objective of this study is to assess the full distribution of hurricane flood risk in NYC. This is done by calculating potential flood damage with a flood damage model that uses many possible storms and surge heights as input. These storms are representative for the low‐probability/high‐impact flood hazard faced by the city. Exceedance probability‐loss curves are constructed under different assumptions about the severity of flood damage. The estimated flood damage to buildings for NYC is between US$59 and 129 millions/year. The damage caused by a 1/100‐year storm surge is within a range of US$2 bn–5 bn, while this is between US$5 bn and 11 bn for a 1/500‐year storm surge. An analysis of flood risk in each of the five boroughs of NYC finds that Brooklyn and Queens are the most vulnerable to flooding. This study examines several uncertainties in the various steps of the risk analysis, which resulted in variations in flood damage estimations. These uncertainties include: the interpolation of flood depths; the use of different flood damage curves; and the influence of the spectra of characteristics of the simulated hurricanes.