Maintaining maintenance

Maintenance-related events represented almost one third of all events reported to the international agencies’ joint Incident Reporting System (IRS) system from January 2006 to September 2008.

Interrogation of the IAEA/OECD/NEA IRS database isolated 60 reports with 85 separate maintenance-related events from the reports. The study focused on the analysis of causes, safety significance, lessons learned and corrective actions of these 60 reports.

Based on the current study, the key areas where improvements are needed are:

• Quality of written procedures
• Comprehensiveness of testing and surveillance programmes
• Training of staff
• Supervision
• Management
• Work practices and communication.

The authors recommend continuous efforts for further improving and harmonizing the quality of event reports, as well as lowering the threshold on reporting of events.

Root causes

In most cases two or three different root causes were reported. The most frequent root cause was deficient written procedures, identified in about 62% of the 85 events. Other frequent root causes were deficiencies in testing and surveillance programmes (33%), lack of training or knowledge (28%), inadequate supervision (27%), and deficiencies in management (24%).

Figure 1: The reasons behind 85 events were attributed to 240 root causes.

Deficiencies in written procedures (53 events) included the following shortcomings:

• Missing procedure or documentation
• Outdated procedure or documentation
• Wrong procedure
• Inadequate procedure or documentation
• Breach of a procedure
• Written procedure was not followed.

Generally, the events with deficiencies in written procedures as a root cause had a combination of other root causes. The two most common accompanying root causes were a lack of supervision (17 events) and deficiencies in training (16 events).

Deficiencies in testing and surveillance programme were the root cause of 28 incidents. An adequate surveillance or testing programme together with taking corrective actions in a timely manner could probably have prevented all the six incidents having this sole root cause.

Deficiencies in training or lack of knowledge were one of the most frequent root causes, occurring in 24 events. These cases include:

• Inadequate training programme
• Lack of training
• Unqualified staff involved in carrying out an action
• Task was carried out for the first time without training or pre-job briefing.

Subcontractors were involved in ten events, almost 42% of the cases in this category.

Missing or inadequate supervision or monitoring of the work was reported as the fundamental cause of 23 incidents. Inadequate supervision was often accompanied by several other root causes. Subcontractors were involved in nine of total 23 incidents (39%).

Deficiencies in management occurred in a third of the events. These include shortcomings in:

• Resource allocation
• Coordination and organisation of work
• Decision making/decision process
• Management of safety-related measures and improvements in time.

Events and safety

Most of the reported maintenance-related events had a limited effect on safety. However potential consequences of maintenance-related events might lead to an initiating event (such as total loss of feedwater or loss of essential service water), degradation of defence in-depth, common cause failure, unavailability of safety significant equipment or safety function, and serious consequences for labour safety.

Some of the maintenance events were potential precursors of an accident. Two impacted on nuclear safety and were assessed at level 2 of the International Nuclear Event Scale (INES) system. One event led to a criticality accident and the other to fuel damage within the technical specifications. Fire, electrical shocks, exposure and hot water resulted in six deaths and 11 injuries. Almost 30% of the maintenance-related events had actual or potential common cause features.

Most of the common cause failure cases had potential safety significant consequences. Safety-significant consequences might have been encountered in seven cases because of the unavailability of safety-significant equipment, or because a safety function had been impeded by maintenance faults. Probabilistic risk assessment methods, seldom utilised with maintenance-related events, could provide a useful tool for analysing the potential safety significance of the event.

Recurrence of maintenance-related events was apparent. The existing operational experience feedback from the other units of a plant or other plants was either unavailable due to the poor dissemination or the event’s significance was not understood. All the recurrent events might have been prevented if operational feedback was taken into account together with the timely implementation of preventive measures.

Ulla Vuorio, Marc Noël and Manuel Martin-Ramos, European Clearinghouse on OEF for NPPs, Institute for Energy - Joint Research Centre - European Commission, P.O. Box 2, 1755 ZG, Petten, The Netherlands.