A prescriptive model to assess the socio-demographics impacts of resilience improvements on power networks

https://doi.org/10.1016/j.ijdrr.2020.101777Get rights and content

Abstract

This paper provides a prescriptive resilience modeling framework for power grids that can account for the socio-demographic impacts of system improvements in the case of hurricanes. The power infrastructure failure rate and recovery duration models are developed based on Hurricane Hermine power outage data obtained from the City of Tallahassee, FL. For the component failures, physical factors such as component type and age, and building age in the surrounding area were used. For the component restoration, factors such as component age, critical facilities, and land use characteristics are considered. Monte Carlo simulation is utilized to estimate the potential impacts of two resilience policy/investment decisions: 1) investment to renew infrastructure components, and 2) reducing the component restoration time for faster recovery. For each scenario, the time evolution of affected populations (i.e., percentage of population with power at any time) is broken into socio-economic categories such as income, age, and ethnicity. Due to significant impact of infrastructure and neighborhood age, the scenario simulation results indicated that lower income populations were affected more (i.e., higher percentage of residents lost power) due to the Hurricane Hermine. Hence, for social equity considerations, it can be recommended that policy makers should prioritize infrastructure investments over improving recovery operations within the available budget constraints. The scenario analysis results also indicate that infrastructure investments which spatially target lower income areas can provide reasonable resilience improvements across the board while significantly closing the recovery gap between lower and higher income populations.

Introduction

The concept of resilience relates to the time evolution of a steady-state system after a disruption. In some domains (e.g., predator-prey populations), the system can transform to a new equilibrium state in response to a shock, and cannot necessarily return to its pre-shock state [[1], [2], [3], [4], [5], [6]]. However, for disaster resilience in urban settings, the emphasis is generally on the system's ability to return to its original operational state after a shock [5,7]. In order to identify whether a system returns to original or a new equilibrium, a quantitative system performance function formulation, also dubbed as figure of merit [3], is needed.

Depending on the infrastructure domain and the governing agency (e.g., department of transportation, power utility company), the system performance and functionality relate to various infrastructure parameters. The performance of transportation networks is related to network connectivity and travel times, whereas the performance of power networks is related to the number of customers with electricity. However, any urban settlement is a socio-technical environment where the infrastructure and society interact. Particularly for urban areas, the resilience refers to a bounce back to “normal” daily life, including all economic and leisure activities. Thus, the restoration of infrastructure does not tell the whole story of resilience. For instance, the status of schools is a critical resilience indicator [8,9] as for almost all parents, reopening of schools after a disaster means that daily life returning to “normal” so that they can get back to their daily routine [10]. Similarly, the functionality of public and private institutions (e.g., re-opening of public offices and businesses) also contributes to the return to “normalcy.”

Platt [6] also points out the ambiguity in “what is normal?” and argue that the recovery speed is not the only performance measure to consider unless “building back better” is achieved. In that respect, focusing purely on infrastructure recovery can overlook the societal impacts and conceal the true return to normality. In addition, the bounce-back to the normality is not necessarily experienced simultaneously by all segments of the population. The concept of “social vulnerability” addresses this concern by emphasizing different patterns of recovery within different population groups that live at the same hazard region [11]. Tapsell et al. [12] argue that the social impacts of hazard exposure often fall disproportionately on the most vulnerable populations such as the poor, minorities, children, the elderly and disabled. Vale [13] argues that resilience cannot remain useful as a concept unless it is associated with improvement of the disadvantaged groups’ life prospects. Vale [13] further discusses that this social dimension is often lost in conceptualizations of resilience in engineering and ecology domains.

The literature provides numerous conceptual frameworks which identify the factors to incorporate the technical and social considerations [[14], [15], [16], [17]] as well as economic factors [18] into disaster resilience and component importance assessment. Both for technical and social parameters, these studies commonly cite data availability to be an important challenge in order to implement and validate the proposed frameworks. The rare nature of the disasters exacerbates this problem since the disaster impact and response operations data are difficult to obtain and are mostly unique to each disaster.

Due to aforementioned data constraints, resilience improvements are performed within costly lessons-learned cycles after each disaster. At every cycle, the observed and some of the anticipated problems are fixed in an ad-hoc manner, just to wait for the next disaster to understand whether the solution or policy was effective, and/or was equitable across different population groups. In other words, the investments and policy decisions for resilience improvements are generally performed based on the after-the-fact descriptive analysis and do not necessarily include a prior assessment of the anticipated technical and societal impacts. On the other hand, rehearsing critical decisions before an event and applying science to inform decision making for building-back-better are emphasized as key issues for governments and decision makers [6]. In order to break the costly lessons-learned cycle and support improvement decisions, there is a need for prescriptive models that account for social vulnerability. Accordingly, the disaster preparedness plans can help prioritize resilience improvements based on the characteristics of the affected population simultaneously with infrastructure-related technical data.

This paper develops a prescriptive resilience model (to the extent possible with data availability) for one of the critical infrastructure networks, namely the power grid. Within this model, power infrastructure failure rate and recovery duration models are developed based on Hurricane Hermine power outage data obtained from the City of Tallahassee, FL. For the component failures, physical factors such as component type and age, and building age in the surrounding area are used. For the component restoration, the factors such as component age, critical facilities, and land use characteristics are considered. By using Monte Carlo simulation, multiple spatio-temporal infrastructure failure and recovery scenarios are created to represent different hurricane impact cases. The time evolution of recovery (i.e., percentage of population with power at any time) is broken into socio-economic categories such as income, age, and ethnicity.

The analyses provide the base case recovery findings with a focus on the affected socio-demographics for future hurricanes similar to Hermine. The infrastructure failure and recovery models help test the impacts of improvement alternatives. Two resilience policy/investment decisions are selected for further analysis: 1) investment to renew infrastructure components, and 2) increasing the response crew size for faster recovery. In addition, spatially specific improvement scenarios based on population income are also conducted. For each improvement scenario, the infrastructure failure and recovery rate models are used to simulate the improved recovery performance and recalculate the socio-demographic impacts. Overall, it is shown that the developed framework can calculate the varying impacts of improvement decisions on different demographics, and can help policy makers and emergency managers to prioritize resource investment plans based on affected demographics and maintain equity.

The paper is structured as follows: First, a literature review is provided and the lack of research on the impacts of infrastructure recovery actions on different populations segments is emphasized. Second, the Hurricane Hermine power outage data for the City of Tallahassee, FL are described, and the overall prescriptive modeling approach is explained. Third, the power network component failure and recovery models are described and estimated. These models help create power network resilience scenarios under hurricane threat, and help circumvent the problem of scarce data availability for assessment of resilience improvements. Then, the hypothetical scenarios are utilized to assess the socio-demographics impacts of alternative resilience improvement actions. Last, the conclusion and future research directions are summarized.

Section snippets

Literature review

With the ongoing climate change, extreme weather events such as hurricanes have become more prevalent with more devastating outcomes. Recently, three significant hurricanes, namely Hermine (2016), Irma (2017), and Michael (2018) hit the Florida Gulf Coast. The City of Tallahassee, capital of Florida, was affected severely by these hurricanes; particularly Hermine caused widespread power outages and roadway closures for a long duration. The impact of Hermine showed that the infrastructure of

Study area

In this study, Hurricane Hermine's impact on the power network of the City of Tallahassee, FL was analyzed. Tallahassee is the capital of Florida with a total population of 190,894 and is a home to two major universities, making it an important urban region in the Northwestern Florida. Hurricane Hermine hit Tallahassee on September 2nd, 2016. It was the first hurricane to make landfall in Florida since Hurricane Wilma in 2005, and was the first hurricane to directly hit Apalachee Bay since

Data

The power infrastructure data include municipality power lines (feeders) as well as components such as circuit breakers, reclosers, sectionalizers, switches, fuses, and transformers that serve 126,737 electricity customers in the city. These components are connected to each other within the power infrastructure, and all customers are connected to the power infrastructure to access power.

The power infrastructure is divided into feeder subnetworks and components connected to each feeder

The approach

The overall framework proposed in this study is presented in Fig. 3. The objective is to provide a modeling framework that can help decision makers assess resilience policy alternatives not only with respect to the technical aspects, but also with respect to the socio-demographic impacts. For this purpose, first, the available power network data is utilized to develop the component failure and power restoration models. By using these models, hypothetical hurricane scenarios are created through

Scenario generation for hurricane impacts on the power network

In order to generate scenarios for hurricane impacts on the power network, two models are estimated: 1) component failure and 2) component restoration. In essence, the failure model provides the spatial distribution of the power outages, and the restoration model provides the temporal recovery timelines that is necessary to calculate the resilience curves.

Assessment of alternative actions for resilience improvement

The resilience of urban systems is a multifaceted issue that is affected by external/exogenous factors such as the magnitude of an event as well as internal/endogenous features such as the condition of the infrastructure or preparedness of governments/communities. In general, the exogenous factors are outside the control of decision-makers and planner; however, the decision makers can act on altering the endogenous factors. Platt [6] studies earthquakes from multiple countries and the findings

Assessment of targeted policies for social equity

Overall, the developed framework provides a quantitative evaluation for strategies that are expected to yield improvements. For instance, the findings indicate that renewal of components more than 5-years old results in an average of 7 h earlier 90% recovery compared to reduction of recovery times by 10%. Meanwhile, the reduction of recovery times by 20% leads to 2 h earlier 90% recovery compared to the renewal of more than 5-years-old components. Such results establish the comparison of

Conclusion

This study develops a prescriptive modeling framework to assess the socio-demographics impacts of resilience improvement actions with a focus on power networks. The prescriptive nature of the approach help break the lessons-learned cycle in disaster management and help decision-makers have before-the-fact anticipation of the potential resilience improvement actions on socio-demographic groups. The framework utilizes the power outage and recovery data from the technical side, and demographics

Future work

It is mainly a matter of data availability to increase the accuracy of the model, to test more detailed scenarios, and to upgrade the framework itself towards a decision-making tool. First candidate for improvement is component failure model. By utilizing spatio-temporal wind speed and direction data (that is in enough detail to project on individual infrastructure components) can be used to formulate a component failure model that can be applied for hurricane scenario analysis. Nevertheless,

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

The authors would like to thank the City of Tallahassee, especially Mr. Michael Ohlsen, for providing data and valuable insight. The contents of this paper and discussion represent the authors' opinion and do not reflect the official view of the City of Tallahassee. This material is based upon work supported by the National Science Foundation under Grant No. 1737483 and Grant No. 1640587.

References (74)

  • J.J. Dong et al.

    Logistic regression model for predicting the failure probability of a landslide dam

    Eng. Geol.

    (2011)
  • K. Barker et al.
    (2014)
  • J.L. Renne

    International Journal of Disaster Risk Reduction Emergency evacuation planning policy for carless and vulnerable populations in the United States and United Kingdom

  • S. Davoudi

    Resilience: a bridging concept or a dead end?

    Plann. Theor. Pract.

    (2012)
  • C.S. Holling

    Resilience and stability of ecological systems

    Annu. Rev. Ecol. Systemat.

    (1973)
  • E de Oliveira Teixeira et al.
    (2013)
  • S. Platt

    Factors affecting the speed and quality of post-disaster recovery and resilience

  • U. Hassler et al.

    Resilience in the built environment

    Build. Res. Inf.

    (2014)
  • S. Matsuura et al.

    Concepts and approaches of school centered disaster resilient communities

  • K. Ronan et al.

    Promoting Community Resilience in Disasters

    (2005)
  • H. Howat et al.

    Lessons from crisis recovery in schools: how hurricanes impacted schools, families and the community

    Sch. Leader. Manag.

    (2012)
  • S.L. Cutter et al.

    Social vulnerability to environmental hazards

    Soc. Sci. Q.

    (2003)
  • S. Tapsell et al.

    Social Vulnerability to Natural Hazards. CapHaz-Net WP4. Flood Hazard Research Centre – FHRC

    (2010)
  • L.J. Vale

    The Politics of Resilient Cities: whose resilience and whose city?

    Build. Res. Inf.

    (2014)
  • H. Hill et al.

    From fatalism to resilience: reducing disaster impacts through systematic investments

    Disasters

    (2012)
  • A. Ostadtaghizadeh et al.

    Community disaster resilience: a systematic review on assessment models and tools

    PLOS Curr. Disaster

    (2015)
  • A. Garcia Tapia et al.

    Evaluating and visualizing the economic impact of commercial districts due to an electric power network disruption

    Risk Anal.

    (2019)
  • R. Berg

    National Hurricane Center Tropical Cyclone Report - Hurricane Hermine

    (2016)
  • C. Folke et al.

    Resilience thinking: integrating resilience, adaptability and transformability

    Ecol. Soc.

    (2010)
  • J.R. Stevenson et al.

    Resilience benchmarking & monitoring review

  • M. Bruneau et al.

    A framework to quantitatively assess and enhance the seismic resilience of communities

    Earthq. Spectra

    (2003)
  • A.-L. Barabasi et al.

    Emergence of Scaling in Random Networks. Science (New York, N.Y.) 286(5439)

    (1999)
  • S.V. Buldyrev et al.

    Catastrophic cascade of failures in interdependent networks

    Nature

    (2010)
  • A. Sumalee et al.

    Network capacity reliability analysis considering traffic regulation after a major disaster

  • S. Sridhar et al.

    Cyber–physical system security for the electric power grid

    Proc. IEEE

    (2012)
  • R. Albert et al.

    Statistical mechanics of complex networks

    Rev. Mod. Phys.

    (2002)
  • A.-L. Barabási et al.

    Mean-field theory for scale-free random networks

  • Cited by (5)

    1

    Current Affiliation: Department of Civil Engineering, University of Twente, Enschede, The Netherlands

    View full text