During the supervisory work of the Swedish Radiation Safety Authority (SSM), a need was identified to develop the methods used when evaluating control room work in the central control rooms of nuclear power plants. Benchmarking is commonly used today, with reference values from earlier Integrated System Validation (ISV), when ISV is available. Often, ISV performs well but has some weaknesses. Some of the elements of knowledge that are currently missing include how to establish strict clarity concerning the aspects that have individual importance and which aspects are important collectively, as well as how to match different measurable aspects.
Improved knowledge in this area, in addition to an advanced method, can give a credible outcome and provide guidance when formulating specifications of requirements for requisite skills, provide input for education and training programmes which may need sharper focus, and achieve a higher level of knowledge in-house at SSM in relation to supervision in the field.
The assignment to investigate the methods used when evaluating control room work in the central control rooms was given to GEISTT, which as part of a research project, placed a focus on how the methods of evaluation might be improved. This was done by means of an in-depth study of how data can be analysed and presented using a static method for modelling called Structural Equation Modelling (SEM). Examples of useful output from/benefits of SEM include the possibility to integrate the analysis of several different data collection methods and scales, as well as the possibility to present the outcomes in a way defining the factors of greatest significance, e.g. in order to illustrate acceptance values for different criteria. SEM can also be used for plant modifications both large and small.
The method has not yet been applied by the Swedish nuclear power industry. On the other hand, forms of cooperation for development of evaluation methods have been established not only with other agencies that regulate the nuclear power industry, but also with IFE/Halden, which have shown great interest. Based on the stringency and the outcomes produced within the project, SSM expects that further research will be carried out using quantities of data designed to better suit SEM, to which other sources of funding will also contribute.
The results indicate that SEM is a statistical modelling method that can meet needs and increase the level of knowledge to possibly benefit individual facilities, educational institutions and the Authority. The results also indicate that when conducting evaluations, it is essential from the outset, prior to the evaluations, to conscientiously look into how measures and variables are formulated and to set the parameters for the quantity of data while considering how the outcomes should be collected and analysed.
Need for further research
A current evaluation method such as ISV is designed to make detailed information available regarding Human Error Discrepancies, which is specific to and appropriate for the nuclear power plant in question, and limited to benchmarking only at this facility. However, since there is a need to have capability to perform comparisons on a more general level in order to develop this area of competence, SSM has established that there is a need for further research. One need that has been identified is investigating whether it is valuable to study the outcomes from previous evaluations of integrated systems and to raise them to a higher level of abstraction for the purpose of achieving comparability and reinforcing the reference values.
Contact person at SSM: Yvonne Johansson