Cyber security: When is an attack, an attack?

By Control Engineering Staff June 6, 2007

This story is a little scary: On August 19, 2006, the Brown’s Ferry Nuclear Power Station had an incident where the variable frequency drives (VFDs) running the reactor recirculation pumps on Unit 3 tripped and stopped. Operators shut the reactor down manually. Everybody was OK and everything following the actual event happened as it was supposed to. The incident happened 10 months ago, but the question remains: why did it happen? Some suspect a cyber intrusion, but the NRC doesn’t want to investigate it on that basis. Congressional committees for Homeland Security want the NRC to think again.

On May 18, 2007, Bennie G. Thompson, chairman, Committee on Homeland Security and James R. Langevin, chairman, Subcommittee on Emerging Threats, Cyber Security, and Science and Technology, sent a letter to Dale E. Klein, chairman, U.S. Nuclear Regulatory Commission (NRC) with some very specific questions.

“After conducting a review of the event, the licensee [i.e., the plant operators, which is the Tennessee Valley Authority in this case] determined that the root cause of failure was the malfunction of the VFD controller due to‘excessive traffic’ on the plant’s computer network. The licensee notified the NRC of the incident and the corrective actions implemented, which included placing a firewall that limits connections and traffic to any devices on the plant’s integrated computer system network. In accord with current regulations, NRC staff decided against investigating the failure as a ‘cyber security incident’ because 1) the failing system was a ‘non-safety’ system rather than a ‘safety’ system, and 2) it was determined by the licensee that the incident did not involvean external cyber attack on the system. We have deep reservations about the NRC’s hesitation to conduct a special investigation into this incident.”

There’s more in the full text of the document , which makes for interesting reading. Here are some questions taken from the discussion that you might want to think about when evaluating your own systems:

Would you know an attack if you saw one? Here the NRC seems willing to say that nothing sinister happened based on the operator’s desire to close the issue. Do you have sufficient forensic capability to examine a real or suspected incident?

Do you have protection within your system to handle data storms? Some PLCs are sensitive to unexpected traffic levels and simply crash, taking all their I/O devices with them. Even a simple ping command can knock some off balance. Do you know how robust your control devices are?

Do you know all the paths to your control devices? While there is no specific answer yet, one possible scenario was that the data storm in this case was the result of another PLC malfunctioning, possibly because of a cyber attack, causing it to send out bad data that choked the VFD controller. Would you be able to follow the connection chain and tell if the event was internally or externally caused in your system?

This is a complicated incident but one that is worth following as a cautionary tale for your control systems. Read the full text of the congressional letter. It has seven questions that they want answered by the NRC. While some of them are unique to this situation, you might ask yourself if you could answer similar questions for your plant.

—Edited by Peter Welander, process industries editor, ,
Control Engineering Weekly News