The oldest reactor units in French nuclear power plants have reached an important milestone in their lifetime – their 3rd ten-yearly outage – and EDF has announced its intention to continue operating them up to the age of 40 years at least.
Within this context, the safety authority wishes for these third ten-yearly outages to include an analysis of equipment aging and obsolescence.
During this analysis, particular attention should be paid to assessing the adequacy and relevance of the operator's maintenance program for electrical equipment (instrumentation, actuators, I&C, electrical power supply) to detect and take into consideration any aging phenomena.
This is because the operator assesses aging mainly on the basis of reported failure rates, as it is difficult to predict the lifetime of such equipment based only on theoretical studies of its behavior.
In order to learn more about the various methods used for monitoring and processing failure rates and determine the effectiveness of these methods, IRSN has worked with NOVATOR to analyze failures reported on certain types of equipment (circuit breakers and sensors).
The method used involves calculating a statistical criterion based on the number of times the failure rate during each defined time interval increases in relation to the others.
This yields statistical criteria that indicate aging trends with a certain confidence level. These trends can be used for analyzing safety studies relating to aging.
The method is intrinsically designed to:
distinguish between failure rates according to equipment age, take into consideration earlier equipment replacements, truncate data where the degree of uncertainty is considered too high.
The results have led to a number of conclusions as to how suitable the method is for early detection of aging in the equipment studied, and certain implementation problems have been highlighted. Rising failure rate trends have also been revealed in certain groups of equipment studied.
Although a specific feature of the French nuclear power plant fleet is that it was built in series, commissioning of reactor units in any one series was spread over a period of ten years or so.
That is why it is so essential for the method used to be capable of distinguishing between failure rates according to equipment age.
A method that monitors the failure rate, but ignores equipment age, would only yield average, smoothed results (because it would mix equipment of varying age).
For this reason, the method discussed here probably offers an accurate assessment of the onset of aging phenomena.
It does, however, call for detailed knowledge of equipment life history (replacements/engineering changes), and gathering data on equipment replacement is not an easy task. One reason for this is that replacement equipment can come from other reactor units or stocks and its age is unknown. In addition, some equipment is only replaced in part.
By default, and if no additional data is available, equipment age has been based on the industrial commissioning date of the reactor unit (and the age can be reset in the event of equipment replacement).
An initial parametric assessment was made of the effect of replacement dates on results in order to assess the impact of any inaccuracies on data. It was found that taking replacements into consideration had little impact when a large population of equipment was studied.
Lastly, the confidence level of computed failure rates was found to increase with the amount of data obtained for each age group.
For this reason, data must be truncated for trend analysis (edge effects on the first and last observation periods, and more generally, all data where failure rates are affected by a high degree of uncertainty).
In view of this study, IRSN considers that this method should be used to analyze sets of equipment data in light of the various remarks made above. The method should then be compared with a technological and qualitative analysis of aging, based on close inspection of equipment to confirm or invalidate these purely statistical results.