The criteria are used to measure the performance of the different failure detection approaches. The selection is very important, since the result will be different, if not all relevant criteria are included.

As most of the methods are still in development, a qualitative evaluation is applied. The criteria are as follows:

• Automatic Failure detection included?

• Accuracy/effectiveness of failure detection

• Automatic Failure identification included?

• Accuracy/effectiveness of failure identification

• Costs (operational and hardware)

3. Overview of Failure Detection Methods

The methods for failure detection will be described in the next sections. Table 3.1 lists some characteristics of the different methods, like what time scale and for what type of systems they are applied to.

key figures were studied. These are compared to the values determined in the planning phase, which are based on the irradiation and temperature profile of a typical reference year [2].

The optimisation phase was very effective regarding failure detection, however, it is time-consuming and therefore costly. The routine supervision phase is not that time-consuming, but does not deliver a quick feedback if the system is working properly and it does not locate failures.