In a fusion experiment, the scientists are continuously monitoring parameters such as plasma temperature and density, the magnetic fields, the velocity of the plasma particles and many more. Because the plasma is 100 million degrees Celsius, one cannot directly insert sensors without them melting, or turning into plasma themselves. Consequently most diagnostic systems work indirectly.
As an example, laser beams or microwaves trigger certain plasma emissions or scatter on the plasma particles. These effects are correlated to plasma temperature or density. Other systems use detectors to measure the radiation or particles emitted by the plasma providing information about plasma temperature, density and impurities inside the plasma. Some techniques reach deeply into the plasma core; others provide good data from the plasma edge. Some provide high spatial resolution; others are fast enough for real time measurements.
JET, for instance, is surrounded by more than 100 different diagnostic systems and 60 of them are in use during an average experiment. During one experimental day, which comprises about 20 experiments, the diagnostics systems generate nearly a terabyte of data. The recordings from more than 80,000 shots performed on JET are archived and continuously transferred to up-to-date storage material. If a special occurrence during an experiment reminds the scientists of a previous pulse, however far in the past, the information is readily available.
Plasma diagnostics serves two main purposes: First, scientists use the data to learn about and analyse the behaviour of the plasma. They observe how the plasma responds to the heating systems, how instabilities form or how impurities from the vessel wall interact with the plasma. Secondly, the measurements are used to protect the machine. Instabilities, for instance, can cause the plasma to expand rapidly and harm the vessel wall if they are not recognized and controlled early enough. Some data feeds into the systems that automatically control the plasma. These systems, for instance, automatically adjust the magnetic field to keep the plasma away from the walls. They require large processing capabilities to calculate the corrections required before any damage is done.