- New algorithm named Squirrel detects weaknesses in the power grid.
- It can be applied to all types of hazards, including earthquake, malicious attack, and obstruction caused by squirrels.
What conditions and failures in the infrastructure of electric grids could lead to a dramatic blackout? And how private companies and government organizations could invest in to help prevent a disastrous shutdown?
Over the last couple of years, news has emerged that North Korean hackers penetrated an American energy utility, and Russian hackers breached a nuclear power plant. Such hacker-induced blackouts are now happening more frequently than ever. Natural disasters like earthquake also play a major role in the disruption of power grids. Even squirrels (that often chew into electrical wires) pose a serious threat to power grids.
To address these issues researchers at Lawrence Livermore National Laboratory have developed a new algorithm named ‘Squirrel’ that allows companies to detect weaknesses in the power grid. It’s a part of a 3-year long project (Quantitative Intelligent Adversary Risk Assessment) focused on identifying all possible threats to the grid.
Since the algorithm is ’cause agnostic’, it can be applied to all types of hazards, including earthquake, malicious attack, and even obstruction caused by squirrels. It solves the inverse problem (deduce cause from effect) and helps you decide where to put attention and how to allocate resources.
What Exactly Does It Do?
The most difficult part in identifying risks to the grid is the cascade effect. One faulty substation could affect the whole grid infrastructure. Researchers implemented their new algorithm with open-source power grid simulator (simulates transmission power flow) to study what sequence of events would have to occur to cause a loss of 1/2 billion watts of load on a small grid model.
The simulation detected more than 700 critical failures of consequence, and nearly half of these failures involved one specific relay. This type of information could be very useful, especially when resources are limited.
Simply put, if you are concerned about a faulty substation that will take 10 gigawatts of load offline and looking for the ways that could make it happen, the algorithm can help you determine what failures would actually lead to that outcome.
Reference: Lawrence Livermore National Laboratory
This will help agencies narrow the number of possible conditions and prioritize locations that could lead to dramatic failures. The algorithm is capable of changing several parameters and finding an efficient solution to avoid a massive outage.
Although power grids developed in recent years are much more automated and smarter, they are more vulnerable to hackers and adversaries (because they are connected to the internet). At present, using manual techniques to secure grids is not possible: no one can dream up all the events that would lead to a nasty outcome. So, it’s kind of necessary to use Squirrel to narrow down the list of events that cause failure.
What’s Next?
Before applying it to large-scale systems, the algorithm needs to be improved. Researchers plan to enhance its modeling capabilities so it can take multiple, complex consequences into account, such as electricity and gas, communication and power flow, and more intricate distribution models.
So far, they have worked with a basic model that could handle 46 transmission lines. These lines can either be switched off or on, thus there would be 2⁴⁶ possible arrangements – an extremely large number to process via brute force methods. For practical and smart analysis, their target is to model a 10,000 line system.
Read: A Linear Algorithm That Can Locate Smartphones Without GPS
In the future, researchers will also collaborate with agencies to detect vulnerabilities and carry out risk assessment using real power grids to produce more precise models.
