Pukhraj Singh is a former analyst at India’s National Technical Research Organization (NTRO). In a report, he connected a malware report published by VirusTotal to a cyberattack on the computers at the Kudankulam Nuclear Power Plant (KNPP). Singh says that a North Korean virus called Dtrack managed to achieve “domain-controller level access” at Kudankulam. His findings have been reported to the Indian government.
    It does not appear that the attack affected any of the controls for the reactors, but research and technical data may have been stolen. It seems that the target of the attack was a collection of technical information. The attack employed a Windows SMB network drive share with credentials hard-coded in the malware which collected targeted files. North Korea’s Lazarus threat group was connected to Dtrack based on the fact that similar code was found in DarkSeoul, an attack by malware that erased hard drives in South Korean banks and media groups in 2013.
     Singh mentioned the attack in a September 7th tweet. He wrote, “I just witnessed a casus belli in the Indian cyberspace and it sucks at every level.” He admitted that he learned of it attack from a “a third party.” Singh presented the information to India’s National Cyber Security Coordinator on September 4. The third-party Singh mentioned had shared indicators of compromise “over the preceding days.” The virus tracking organization Kaspersky made the identification of the virus with Dtrack after Singh reported it.
    Officials at Kudankulam have stated that the computers that actually run the reactor are isolated from the administrative computers and are immune from cyberattack. They have not said what valuable information may have been stolen from the administrative computers. Kudankulam issued a press release in which the training superintendent and information officer for the KNPP said that the plant “and other Indian Nuclear Power Plants Control Systems are standalone and not connected to outside cyber network and the Internet… Any Cyber attack on the Nuclear Power Plant Control System is not possible.” The officials at the plant claimed that the two reactors at the plant are currently up and running “without any operational or safety concerns.”
    KNPP is the biggest nuclear power plant in India. Controversy has bedeviled the plant since construction started in 2002. Local activists including fishermen managed to delay the activation of KNPP by a decade. There are plans for KNPP to collaborate with the Russian Atomstroyexport in the construction of and operations of a total of six nuclear power reactors but for the present there are only two operating reactors at KNPP. There have been many reports of safety violations at the plant. The plant has no offsite spent nuclear fuel storage facility. There was a court battle over whether or not the plant should be allowed to operate until such a facility had been constructed.
     The reactors at KNPP have been shut down seventy times since the reactors were turned on in 2013. On October 19 of this year, the second reactor at KNPP was shut down because there was a fault in the reactor’s steam generation. The officials claim that that shutdown was not related to the Dtrack virus.

Ambient office  = 68 nanosieverts per hour

Ambient outside = 126 nanosieverts per hour

Soil exposed to rain water = 122 nanosieverts per hour

Red bell pepper from Central Market = 65 nanosieverts per hour

Tap water = 112 nanosieverts per hour

Filtered water = 100 nanosieverts per hour

Ambient office  = 90 nanosieverts per hour

Ambient outside = 136 nanosieverts per hour

Soil exposed to rain water = 137 nanosieverts per hour

Red bell pepper from Central Market = 109 nanosieverts per hour

Tap water = 77 nanosieverts per hour

Filtered water =67 nanosieverts per hour

Ambient office  = 111 nanosieverts per hour

Ambient outside = 100 nanosieverts per hour

Soil exposed to rain water = 95 nanosieverts per hour

Red bell pepper from Central Market = 87 nanosieverts per hour

Tap water = 89 nanosieverts per hour

Filtered water = 82 nanosieverts per hour

Dover sole – Caught in USA = 108 nanosieverts per hour

Part 3 of 3 Parts (Please read Parts 1 and 2 first)
     The three major components of the Outage AI solution are bracketed by a data transfer process that sends candidate nuclear unit outage schedules into and out of the cloud-hosted environment directly from and to the on-premises servers and software systems at OPG. This provides a seamless integration with the existing technical stack. There is little or no interruption in the outage work schedulers’ computer aided scheduling duties.
    During the development of this Outage AI solution at OPG, the team has embedded in the solution three key elements that are considered critical to the long-term success of the project. First, the team has ensured that the solution actually improves human labor efficiency. Second, the team has incorporated in the solution the ability to be scaled up within OPG to assist in the planned maintenance of other power generating sites and assets. Third, a framework has been developed for OPG that can be expanded in the future to include scope changes and to also include the balance of resources to enhance the capabilities of the current solution.
     AI, deep learning, NLP and intelligent automation are not discrete single solutions to problems. Each of them is located on a spectrum based on the complexity of the problem under consideration and the predictive power required of the algorithm. They must be designed with the correct combination and industry expertise to provide useful results.
      Implementing these new technologies into one specific stage of the nuclear unit outage planning process enables the OPG staff to focus on more important activities instead of repetitive tasks related to creating the first version of the schedule. The Outage AI solution also facilitates improving and optimizing scheduling by rapidly processing big volumes of historical outage data that are becoming too big and time consuming for outage team members to completely analyze. This enables the reduction of human scheduling errors and oversights. It also allows better identification of missing tasks. This has the potential to reduce downtime for nuclear unit outages.
     During the Outage AI project, there have been other opportunities for applications of AI identified beyond the generation of the first schedule. The most attractive AI opportunity lies in scaling the Outage AI solution to create draft schedules that can be used for planned maintenance outages at similar sites.
      Another important potential application of AI for OPG would be improved balancing of workloads, resources and capacities. This would promote the effectiveness of processes and highlight operational efficiencies in planning outages. There may also be an opportunity to overlay AI based techniques onto the connected supply chain. This could drive performance and increase productivity by leveraging the data in a more prescriptive manner as opposed to being diagnostic or even predictive.
      This list is not intended to be exhaustive, but it does present a glimpse into potential benefits that result for the use of AI to facilitate data-driven decision making and effectively complement the work of the human teams involved in the nuclear industry.

Ambient office  = 73 nanosieverts per hour

Ambient outside = 140 nanosieverts per hour

Soil exposed to rain water = 148 nanosieverts per hour

Blueberry from Central Market = 101 nanosieverts per hour

Tap water = 158 nanosieverts per hour

Filtered water = 136 nanosieverts per hour