On 4 August 2020, a tropical storm knocked out power in many parts of New York City as well as neighboring counties and states. The electricity utility, Consolidated Edison, was able to fully restore service in Manhattan within a few hours. Meanwhile, in the surrounding boroughs of the Bronx, Brooklyn, Queens, and Staten Island, thousands of customers remained without electricity for days. There are technical reasons that contributed to faster repairs in Manhattan, but in general the neighborhoods that waited the longest to have their power restored tended to be poorer and less white.
For most people, a power outage is an inconvenience. But for some, it is an emergency that can quickly turn deadly, especially when the outage occurs during a heat wave or a winter freeze. Extended exposure to temperatures above 32° C can quickly cause health crises, especially in the elderly, children, and people with heart disease, poor blood circulation, and other pre-existing conditions. The recent record-breaking heat in Oregon and Washington state, for example, claimed more than 200 lives. Extreme cold can have similarly dire consequences, as we saw during February’s massive power outage in Texas.
Public health experts refer to those who are most at risk during power outages as “electricity vulnerable” or “electricity dependent.” In the United States, hundreds of thousands of people are in that category. A 2017 study estimated that about 685,000 Americans who live at home and have medical insurance are electricity dependent; of that group, roughly one fifth are vulnerable to even short power outages of 3 to 4 hours.
Normally during a heat wave, people have the option of escaping their homes and seeking cooler temperatures in public spaces like libraries, coffee shops, and stores. COVID-19 changed all that. The pandemic created a work-at-home paradigm that shifted electricity usage away from commercial buildings to residential neighborhoods, in ways that few expected and fewer planned for. It made finding relief from the heat logistically difficult. And it slowed urgent repair and maintenance of the power grid, with work crews having to practice social distancing due to the pandemic.
Step 1: Identify outages in real time
There’s a better way to do things. It requires that providers like New York City’s ConEd revise their priorities for repairs during outages. Instead of first serving areas with the greatest density of customers, as they do now, utilities would make repairs in those areas with a greater share of customers whose health is immediately endangered by the outage. This strategy would correct an endemic imbalance that puts greater stressors on less affluent neighborhoods and the electricity vulnerable. The existence of this imbalance isn’t just theoretical, as the storm last August demonstrated.
The NYU Power Outage Dashboard helps visualize power outages in New York City. In the historic data shown here, the red, orange, yellow indicate populations that are more vulnerable to outages, while blue and green indicate groups that are less vulnerable.
To help implement this strategy, my group at New York University has been developing a Power Outage Dashboard for New York City. The dashboard, created with funding from the National Science Foundation, collects data from ConEd about power outages in the city and integrates that data with open-source socio-demographic and environmental data to evaluate the severity of each outage for electricity-vulnerable groups.
Based on this evaluation, we compute a rank for each of New York City’s 300-plus zip codes that takes into account demographic information like household income, age, race, and gender, as well as public health data and the presence of low-income and senior housing; the Zip Code Rank also factors in dynamically changing environmental data, such as ambient temperature, humidity, and precipitation. From the Zip Code Rank, we can determine an Overall Severity Rank for the outages in each zip code, which can be used to prioritize repairs.
To aggregate this data, we designed a crawler that collects real-time outage data from Con Edison; we also have archives of historical data on hundreds of thousands of past outages. The addresses, zip codes, and demographic information come from NYC Open Data, a comprehensive set of public databases published by New York City agencies and their partners. A composite algorithm that we developed ranks the outages by the relative vulnerability of the customers in the zip code. This data is superimposed on a real-time outage map of New York City and color-coded by vulnerability—red for most vulnerable, blue for least. The dashboard is designed to allow users, including the public, to know which outages should have higher priority.
Even a cursory look at the dashboard shows that outages in Manhattan tend to be green or blue, while those in the outer boroughs tend to be yellow, orange, or red. For example, on 8 July 2021, there were 41 relatively large outages in New York City. Of these, 6 were in more affluent areas of Manhattan, and our algorithm coded most of them as blue. In Brooklyn, by contrast, there were 17 outages coded orange or red.
This chart shows a history of power outages by zip code in New York City. The Brooklyn zip code 11204, for example, had 607 outages. Poor neighborhoods are more likely to experience outages than wealthier areas are.
This wasn’t a one-off. When we look at the historical data, we can see that residents in the outer boroughs are more likely to lose power, with a clear correlation between the number and duration of power outages and the ethnic and class makeup of neighborhoods. A poor neighborhood with a larger minority population in the Bronx is much more likely to suffer an extended power outage than is a wealthier, whiter neighborhood in lower Manhattan.
There are a number of ways to explain this disparity. The outer boroughs have more overhead power lines compared to Manhattan, where the cables run underground, and overhead power lines are more prone to faults. Likewise, the residential buildings in the Bronx, Brooklyn, and Queens tend to be older or less well maintained, compared to the office buildings and luxury condos of lower Manhattan. However you explain it, though, there’s still an underlying problem of social injustice and societal inequality that is leaving vulnerable people in jeopardy and that must be corrected.
We hope to offer the dashboard as an open-source framework for use by utilities. In the future we will be designing functions to help route service vehicles to where they’re needed, based on the availability of repair teams.
Step 2: Prioritize repairs for the most vulnerable customers
Beyond just knowing where outages are and which groups of customers are being affected, a utility also needs to be able to forecast demand—predicting how much electricity it will need to supply to customers in the coming hours and days. This is of particular importance now, when many people are suffering from the lingering effects of COVID-19—so-called “long COVID” patients. Some of them are likely homebound and are now counted among the ranks of the electricity vulnerable.
This map shows real-time outages in New York City on 16 July 2021.
Demand forecasting tools rely on historic trends about electricity use. But in New York City, analyses showed that demand forecasting errors surged in the aftermath of the pandemic’s stay-at-home orders. That’s because the COVID-19 pandemic was a sui generis phenomenon for which there was no historic data. As consumption patterns shifted from commercial buildings to residential, the forecasting tools were rendered ineffective.
Any plan that could significantly alter demand forecasting must be considered with the power grid in mind. Last summer, for example, the mayor of New York City, Bill De Blasio, invested $55 million in a heatwave plan that included installing more than 74,000 air-conditioning units for low-income senior citizens. Although these units are providing necessary relief to a vulnerable population, they also are raising electricity demand in residential areas and causing additional stress on ConEd’s distribution system.
Now that many offices and businesses are reopening, it may be difficult or even impossible for utilities to predict exactly how electricity demand will change this summer and when, where, and what the actual demand peak will be. Just because a utility experiences reduced demand in one part of its system does not mean it will be able to accommodate increased demand in another part of the system. There are basic network limits on the ability to transfer electricity from one part of the system to another, such as voltage and power flow.
Grid operators must therefore proactively analyze the impacts of shifting demand and the reduced accuracy of demand forecasting tools on their systems. And they must factor their electricity-vulnerable customers into their planning. Electricity infrastructure is a complex engineering system, and its reliability cannot be 100-percent guaranteed, despite the best efforts of engineers, managers, and planners. Hence, it is important for a utility to consider every possible contingency and plan for mitigation and corrective actions. Such planning should be transparent and open for public comment and evaluation by experts from leading academic institutions, government labs, professional organizations, and so on.
Of the 685,000 people in the United States who are considered “electricity dependent,” about one fifth are vulnerable to power outages of just 3 to 4 hours.
Some readers may find it odd to link the power grid to social justice, but when you look at historic patterns, it’s hard to ignore that certain groups in our society have been marginalized and underserved. Going forward, we must do a better job of protecting vulnerable populations. Utilities can engage with the local community by surveying customers about their electricity needs. Companies will then be in a good position to assist their most vulnerable customers as soon as any power outage is reported.
Thankfully, New York City made it through last summer with relatively few heat crises. However, the pandemic didn’t end once the weather turned cool. Circumstances could be much worse this summer. The city needs a fundamental change and the tools to affect it, with repairs prioritized in such a way that the most vulnerable, not the most affluent, are serviced first. And ConEd and electricity providers like them need to begin planning now.
About the Author
Yury Dvorkin is an assistant professor of electrical and computer engineering at New York University’s Tandon School of Engineering and a faculty member of NYU’s Center for Urban Science and Progress.