|
|
| Fig. 1: When a transformers primary winding insulation degrades, the software can analyze the type, concentration and ratio of dissolved gases in the transformers insulating oil and compare against previous samples to identify if the fault is simply degradation or something more complex. (Source: Wikimedia Commons) |
Diagnosing problems at nuclear power plants is not only vital to the production of energy, but, more importantly, for the health of people working or living nearby. Finding and solving problems within power plants helps to avoid disasters such as those in Chernobyl and Three Mile Island. Lots of resources and time are generally required to find and diagnose these problems that are so vital. Fortunately, there is an abundance of data available around a nuclear plant that can help diagnose these problems. The average nuclear power plants contains around 10,000 sensors and detectors with about 20 neutron detectors, 60 RTDs, as many as 100 thermocouples, and 500 to 2,500 pressure transmitters. [1] The problem in the past was that analyzing this data was very intensive, but recently software suites have been released, such as EPRI's Fleet-Wide Prognostics and Health Management tool that can be used proactively to find problems before they have negative effects. [2]
The software is built on four basic modules. The Diagnostic Advisor compares current operating data with asset fault signatures to find impending failures. In other words, it finds measurements that are concerning and notifies workers of those problems. [2] The Asset Fault Signature database is a database that keeps track of measurements that signify faults. It collects data from power plants around the world and begins to identify which of the measurements and values result in failures. An example is the analysis of dissolved gas in a transformers insulating oil (See Fig. 1). The Remaining Life Advisor makes data-backed estimates on the remaining life of certain assets. It uses a collection of measurements to identify how much an asset has aged and it uses past data points to identify how much it will age. The Remaining Useful Life database keeps track of predictions around the world for asset lives to better understand the assets in the current power plant. In other words, it keeps track of lives around the world to better understand the lives of assets in the power plant involved. The four models work together to create an entire image of the current and future state of the nuclear power plant so that the workers can easily identify concerns and plan ahead. [2,3]
The importance of this software cannot be understated. Nuclear power plants have continuously been argued to be dangerous forms of energy and this software provides the ability to make it safer. The asset fault signatures are extremely valuable industry wise. Being able to interpret values relayed by sensors around the power plant is vital and the asset fault signatures can become a universal measurement of safety within a nuclear power plant. With 449 power plants worldwide, the data collected everyday is enormous. The automated aspects of this software take out a lot of human error which is generally perceived as beneficial. There is argument that nuclear reactor accidents have occurred due to human error and this software can remove some of the human error from these reactors. In conclusion, this software, which is being tested in several power plants now, has the ability to make nuclear energy safer and more reliable simply by utilizing data that has been collected for many years.
Although software does have the ability to analyze large amounts of sensor data, it does lack many human capabilities and the reliance on software to do control such powerful technology is frightening. It does appear useful to include software as another form of security for humans, but relaxing and allowing the software to fully control nuclear power plants is not ideal in many minds. Software should serve as a tool for better diagnostic, but should not be seen as a crutch for overall safety.
© Jensen Price. The author warrants that the work is the author's own and that Stanford University provided no input other than typesetting and referencing guidelines. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
[1] H. M. Hashemian, "Nuclear Power Plant Instrumentation and Control," in Nuclear Power- Control, Reliability and Human Factors, ed. by P. Tsvetkov (InTech, 2011).
[2] V. Agarwal,et al., "Development of Asset Fault Signatures for Prognostic and Health Management in the Nuclear Industry," IEEE 7036366, 22 Jun 14.
[3] V. Agarwal et al., "Online Monitoring of Plant Assets in the Nuclear Industry," Idaho National Laboratory, INL/CON-13-29233, October 2013.
