Modern tokamaks are highly sophisticated devices consisting of a large number of state-of-the-art systems that must function in unison to obtain a successful plasma discharge. An unsuccessful discharge can result if one or more systems fail, and diagnosis in an efficient and timely manner can be difficult. The resulting reduction in tokamak availability and productivity can be expensive, justifying a significant effort for automated fault diagnosis.

For the DIII-D tokamak, a software system has been used for the past 5 years to automatically monitor and test the performance of hundreds of tokamak systems. The Fault Identification and Communication System (FICS) is automatically triggered to run immediately after each tokamak discharge and report its results via a simple color-coded graphical user interface. In addition to saving the operator time, the significant advantage of FICS is its ability to detect insipient faults that could lead to future machine failures. It has been estimated that FICS has saved an average of one to two shots per day, which equates to approximately 5% of all DIII-D pulses. The significant experience gained through the development and use of this post-discharge analysis tool also provides insight into future methods for on-line process monitoring of steady state devices