Predicting and Preventing Power Outages Across the National Electric GridFebruary 2014
Topics: Critical Infrastructure Protection, Emergency Preparedness and Response, Emergency Management
Ten years ago in Ohio, a high-voltage power line brushed against some trees, which shut down a power line, which knocked out a transformer, which cascaded through the northeast electric power grid until 50 million people from Ontario to New Jersey were without power. The Northeast Blackout resulted in 11 deaths and cost about $6 billion. It affected every segment of our society—from healthcare, transportation, and commerce to public safety and national security. In a modern world, electrical power is more than a convenience. It's a necessity.
Imagine if we had the capability to sense the conditions that make a large-scale power outage possible, detect the initial disturbance in real time, and drop a barrier that would insulate it from the rest of the grid. That's one of the goals of a MITRE research team—to sense a likely event before it happens, and then detect and localize the fault and wall it off.
While much has been done in the past decade to understand the vulnerabilities of the national grid, much remains to be done. In 2012, the average American experienced about 112 minutes without electricity—an increase of more than 20 percent over the last 10 years. Every day about 500,000 Americans experience a disruption in electrical power of one hour or more.
The vulnerabilities we're facing include:
- The age and complexity of the grid, which has a basic architecture dating back to the 1880s. The infrastructure is no longer able to keep up with demand at desired levels of reliability.
- Reduction of reserve margins. Regulations are forcing reduced reserve margins, demand management, and interconnects that place greater stress on the grid to adjust to generation-demand imbalances.
- Natural causes such as storms, earthquakes, severe geomagnetic storms, and wildfires.
- Physical attacks against the infrastructure, from vandalism to terrorism, as well as cyber attacks.
Who's Monitoring the Grid?
Today, industry, governmental regulatory entities (including NERC, the nonprofit North American Electric Reliability Corporation), and academia monitor the grid. They measure voltage frequencies to look for problems.
The grid is a network that couples large and small rotary generators with numerous end-use devices and appliances. The rotation of the generators changes slowly, and it's that electro-mechanical "transient" that they monitor today. Current instrumentation monitors the grid to detect and localize these relatively slow transients. However, disruptions can be very rapid—creating fast-moving transients that may not be recognized in time to take corrective actions.
MITRE researchers working in the area of protecting the nation's critical infrastructure recognized a particular challenge in the electric power grid. We conjectured that a deeper understanding of the dynamics of the power grid was limited by the instrumentation used currently to monitor it, theorizing that monitoring the grid faster would produce more valuable information.
To conduct such high-speed monitoring, our researchers leveraged a prior MITRE resource: the Sentinel network, which consists of nine sensors deployed at MITRE sites across the United States. These sensors monitor the stability of the grid at 50 kHz, which is 1,000 times faster than instrumentation used by the utility industry and universities.
Detecting Fast Transients a Crucial First Step
After just one year of research, we have seen that monitoring the grid at higher sampling rates enables the detection of rapid electro-magnetic transients such as lightning or fault-related over-voltages. Such events typically occur and disappear too quickly to be seen by instrumentation currently used to monitor the grid and may precede a major event like a generator trip.
These fast transients could hold the key to detecting conditions that enable a small fault, which could cascade out of control into a major outage. With our system, we have observed low-level, fast transients hours before a generator trip, up to three hours before they led to a bigger outage. This is a first step toward the creation of an automated system to detect and isolate small events so that they do not get bigger.
We also developed a technique to quantify the complex interaction among sensor number, topology, sample rate, the velocity with which transients propagate along the U.S. power grid, and the geo-location of transients and faults. Our objective is to identify the location of transients consistent with initial events—such as the tree falling and shorting out a power line—instead of detecting the terminal events such as a generator trip, which is the object of currently deployed monitoring networks.
Helping Make Decisions for Next-Generation Sensors
We have demonstrated that faster sampling rate results can be used to reduce the number of instruments monitoring the grid while achieving the same of geo-location precision. We believe these findings will help stakeholders make decisions about the next generation of sensors that will be needed to support the smart grid, as well as government oversight of the security of the power infrastructure.
As noted earlier, power disturbances are not discrete events. Rather, they are complex events that evolve over time and distance. In addition to sensing the progression of that evolution with fast monitoring of the alternating current (AC) waveform, MITRE has investigated the use of social media—specifically Twitter—to track the when-where evolution of power outages. These results not only provide a valuable source of "ground truth" to our other analyses but are valuable as a separate product. One of our sponsors recently asked us to write a white paper on monitoring small-scale power disturbances (those below the detection threshold of current instrumentation and systems) using Twitter.
MITRE's goal is to continue working on a solution in collaboration with other stakeholders. To that end, we've been sharing our research results with industry, the government, and academia.
This year we are focused on predicting conditions favorable for the development and propagation of large-scale disturbances. We continue to assess the complex but strategically important question of why certain disturbances are easily contained while others cascade into major disturbances and how government and industry can come together to enhance the security and reliability of the nation's electric power grid.
—by Beverly Wood