- Date
- August 19, 2006
- Location
- United States
- Industry
- Energy
- Access vector
- Not applicable
- OT targeting
- None
- Primary peril
- None
- Impact duration
- Unknown
Browns Ferry Unit 3 manual scram from network traffic, 2006

The incident path, stage by stage
Each stage of the attack, from intent to financial loss, with this incident’s path highlighted. Solid cells are confirmed, dashed cells are reported by a source but not confirmed, dotted cells are implied by the facts, and Unknown marks what is not publicly known. Click a stage title or a highlighted cell to read that part of the analysis.
Scroll sideways to see every stage.
On this page
Highlights
What happened. On August 19, 2006, the variable frequency drive (VFD) controllers for both reactor recirculation pumps at Browns Ferry Unit 3, in Alabama, became unresponsive, both pumps tripped, and operators manually scrammed the reactor[1]. The licensee found the root cause to be excessive traffic on the plant’s own control network, according to the US Nuclear Regulatory Commission (NRC). There was no attacker, no malware and no intrusion[1][2].
Why it matters. A failure on a non-safety control network, with no adversary anywhere, forced a safety action at an operating nuclear unit[1]. The NRC published the event to the US nuclear fleet as a lesson in network design[1].
About this analysis. This analysis sets out what is known, stage by stage. It rests on the NRC’s information notice, which a later GAO report repeated without adding facts. This account reflects the best available information as of October 22, 2026; new information could confirm, change or discredit parts of it.
Highlights
- Both recirculation pumps, 3A and 3B, tripped when their drive controllers stopped responding, and plant procedures then required a manual scram[1].
- The licensee traced the root cause to excessive traffic on the plant integrated computer system network. It could not settle whether a failed demineralizer controller caused the drive failure, or whether traffic from another source caused both[1].
- Neither the NRC nor GAO describes a malicious actor or an attack[1][2].
- No source states how long Unit 3 was offline or what the event cost[1][2].
Attack flow
- Updated: 2026-10-22
- Intent: Accidental [confirmed]
- Threat actors: Not applicable [not in path | Accidental event, no attacker]
- Access vector: Not applicable [not in path | No intrusion]
- IT impact: Not in path [not in path | OT-native event, no IT stage]
- OT targeting: Not applicable [not in path | No attacker]
- OT sophistication: Not applicable [not in path | No attacker]
- OT impact: Loss / Denial of Control [confirmed | Drives unresponsive, per the NRC]
- Perils: No Perils [confirmed | Protective shutdown, no damage]
- Financial losses: Business Interruption [confirmed]; Cyber Incident Response [confirmed]; Assets & Equipment [implied]
What happened
Unit 3 was operating. When both recirculation pumps tripped, the NRC describes the result as a “high power, low flow condition”, and operators manually scrammed the reactor, as plant procedures required after the loss of recirculation flow[1]. The Unit 3 condensate demineralizer controller, a programmable logic controller (PLC), “had failed simultaneously with the Unit 3 VFD controllers”[1].
The licensee determined that “the root cause of the event was the malfunction of the VFD controller because of excessive traffic on the plant ICS network”, its integrated computer system network. It could not conclusively establish whether the PLC failure made the drive controllers unresponsive, or whether excessive traffic from a different source caused both to fail. The PLC vendor said the PLC failure was “a likely symptom of the excessive network traffic”[1].
Recovery of the drives was simple: operators “cycled the control power off and on, reset the controllers, and restarted the VFDs”. The licensee then disconnected the Unit 3 VFD controllers and the demineralizer controller from the plant network “before restart”, and also disconnected the Unit 2 VFD controllers[1]. On April 17, 2007, the NRC published the event to the US nuclear fleet as Information Notice 2007-15[1], and GAO summarized it in a September 2007 report on control system security[2]. Neither says how long Unit 3 was offline.
Intent: accidental
Neither the NRC notice nor GAO mentions a malicious actor or an attack; both trace the event to excessive network traffic[1][2]. The path therefore starts at Accidental. The event is carried in this library because the failure began inside a plant control network and ended in a nuclear safety action, with no adversary anywhere in the chain.
Threat actors: none
There was no adversary, no malware and no intrusion[1]. One scenario study, the 2015 Business Blackout report by the Cambridge Centre for Risk Studies and Lloyd’s, lists this event in an annex as a worm exploiting a database flaw at a reactor down for maintenance. The primary sources contradict that: there was no worm, and Unit 3 was operating[1]. The description matches the Slammer worm event at Davis-Besse in January 2003, a different plant three years earlier. No threat actor is highlighted on the map.
Access vector: not applicable
No intrusion of any kind took place. Nothing entered the plant network from outside: the NRC traces the event to traffic on the plant’s own integrated computer system network and names no external connection, device or person as its source[1]. Where the excess traffic came from remains unknown. This stage is marked not applicable on the map.
IT impact: not in the path
There is no IT stage. This was an event native to OT: a non-safety Ethernet control network degraded, and the process responded[1]. No server, workstation or corporate system is described as involved, and no data was affected. The stage is marked Not in path, so the flow on the map passes it by.
OT targeting: not applicable
No attacker means no targeting. The drive controllers and the PLC sat on the plant’s own control network, and they failed because of what was happening on that network, not because anything was aimed at them[1]. The stage is shown as not applicable.
OT sophistication: not applicable
There was no attacker whose skill could be judged. The mechanism was ordinary: devices on a shared network stopped responding under excessive traffic. Restoring them took no more than cycling the control power and resetting the controllers[1]. The stage is marked Not applicable on the map.
OT impact: drives that stopped responding
The drive controllers for both recirculation pumps became unresponsive, and both pumps tripped[1]. That is shown as Loss / Denial of Control, confirmed by the NRC and caused by a failure, not by an attack. The consequence reached the plant’s safety response: plant procedures required a manual scram after the loss of recirculation flow, and operators carried it out[1]. The protective action worked as designed. What failed was a non-safety network whose loss still forced a safety action.
Perils: no physical event
The manual scram took the unit to a safe state, and no source reports equipment damage, fire, a release or an injury[1][2]. A protective shutdown of this kind is a business interruption, not a peril.
Traffic on a shared plant network was enough to scram a reactor, with no attacker anywhere.
Financial losses: no figure disclosed
Neither the NRC nor GAO states how long Unit 3 was offline, its lost generation or any cost, and no Tennessee Valley Authority disclosure was located[1][2]. A scrammed unit produces no power until restart, so the loss is real but unstated, and no estimate is made here.
Loss evidence
- Confirmed: Lost revenue and productivity (Unit 3 offline for an unstated period); Network event response and recovery (root-cause investigation, licensee event report, disconnection before restart); Voluntary improvements (Unit 2 drives disconnected, firewall devices listed)
- Implied: Startup and return-to-service validation
- Not supported: Third-party physical and societal impact (no customer loss of supply described)
The NRC notice “requires no specific action or written response”, so the corrective work was voluntary. The notice does not say whether the firewall work was finished[1].
What this incident shows
There was no adversary here for network segmentation to keep out, and the loss happened anyway, because the recirculation drive controllers sat on a shared plant network[1]. The licensee’s fixes were architectural: it disconnected the controllers from that network and listed firewall devices for each unit’s drive and demineralizer controllers[1]. Separation in this case protected against a consequence, not an attacker.
The NRC drew the same lesson for the whole fleet: “Careful design and control of the network architecture can mitigate the risks to plant networks from malfunctioning devices, and improper network performance, and ultimately result in safer plant operations”[1]. For anyone assessing control-system risk, a malfunctioning device on a shared network can stop a process just as an attack could.
References
- U.S. Nuclear Regulatory Commission. “Effects of Ethernet-based, Non-safety Related Controls on the Safe and Continued Operation of Nuclear Power Stations.” Information Notice 2007-15, ADAMS ML071010303, 17 Apr. 2007. nrc.gov/docs/ML0710/ML071010303.pdf
- U.S. Government Accountability Office. “Critical Infrastructure Protection: Multiple Efforts to Secure Control Systems Are Under Way, but Challenges Remain.” GAO-07-1036, 10 Sept. 2007. govinfo.gov/content/pkg/GAOREPORTS-GAO-07-1036/html/GAOREPORTS-GAO-07-1036.htm


