<< Main report | Annex II >>
Based on the comments received during the previous Nuclear Safety Convention Review Meeting in 1999 STUK launched a two part study on the general coverage, consistency and prescriptivity of the Finnish nuclear regulatory guide system. The first part was focusing on coverage and internal consistency and it was carried out as a self-assessment within STUK. The second part was an independent study made by the Technical Research Centre of Finland, which was mainly concentrating on surveying the opinions of the licensee representatives about the prescriptivity of the regulations.
The internal study was conducted in such a way that nominated STUK senior experts compared the contents of the Finnish nuclear related regulations (five governmental decisions) and about 70 regulatory guides with the existing IAEA Safety Fundamentals and Safety Requirements (some earlier Safety Codes were included) documents. The task was to identify gaps in the coverage of the Finnish regulations or, gaps in the IAEA guidance. The consistency of Finnish regulations was assessed i.a. in sense of consistent terminology and structure as well as in sense of consistency between the various levels of regulatory documents' hierarchy.
The main findings of the internal self-assessment were as follows:
The findings of the internal self-assessment will be taken into account when developing the revised strategy for rulemaking, which is scheduled to be established before the end of 2001.
The independent assessment of the possible prescriptivity of the Finnish regulations and guides was carried out by VTT. In the following the key recommendations and conclusions of the study are presented as an excerpt. The whole report of VTT can be accessed on STUK's Internet site, at http://www.stuk.fi/english/publications/.
The findings of the independent study will be taken into account when developing the revised strategy for rulemaking, which is scheduled to be established before the end of 2001.
Based on the earlier chapters of the report this chapter develops some recommendations for how to further develop the YVL-guides. There are some obvious improvements to be implemented in the YVL-guides in a short-term perspective, but STUK should also initiate a discussion of a more long-term strategy for the development of the YVL-guides. There is also a need to discuss how the YVL-guides are written and used with the aim of making interpretations less dependent on the inspectors. Safety requirements build on a combination between deterministic and probabilistic considerations, but it sometimes appears to be difficult to find a proper balance between the two principles. This has to do with an interpretation of the residual risk and an agreement when it is small enough. The classification of functions, structures and equipment is an issue, which is under discussion in various standardisation committees and here some rethinking seems to be necessary. Finally, in a long-term perspective, there are obvious needs for harmonised regulatory approaches. STUK may take an active role in an international discussion of future approaches to regulatory oversight.
One would have expected that the development of the YVL-guides is guided by a clearly expressed strategy. A tacit strategy has evidently been governing their development over the years, but it is would be recommendable to make that strategy more overt. Further development of the strategy would most certainly benefit of general discussion within the nuclear community in Finland on ends and means of regulatory oversight.
There are also a number of larger and smaller needs for improving present guides. The far most problematic is YVL 5.5 which in the present form may even lead to non-optimal practices. Closely connected is YVL 2.1 of two reasons, firstly connected to efforts to reclassify present nuclear power plants and secondly connected to the application of the requirements of YVL 5.5. The requirements of YVL 1.0 that are connected to severe accidents seem to require modifications to balance the probabilistic and deterministic requirements in a logical way.
The possibility that a new nuclear power plant should be licensed in Finland places a large challenge on the whole regulatory system. The YVL-guides themselves give an eminent platform for the licensing process, but it may be necessary to make an assessment of the practical arrangements to ensure that undue requirements on documentation to be supplied will not introduce unreasonable delays in the design and construction process. Ideally a licensing process should be adapted to assess the solutions of a design and construction process in the order they are produced.
There seems to be some problems in maintaining a consistent interpretation of the YVL-guides. This is something, which has to be expected with requirements written in a natural language, and it may even be counterproductive to try to formulate the requirements in a way to minimise the room for interpretations. According to the interviews the problems are accentuated when new inspectors are taking charge of some area. The underlying problem seems to be partly connected to the maintaining a pool of knowledge at STUK and partly connected to the decision-making processes applied. If younger inspectors can get advice and support for their own judgement this problem should be possible to combat.
Another problem is that some YVL-guides are interpreted more stringently and others more freely. To some extent this is natural, because different issues have a different weight on nuclear safety, but the same general principles should still apply. This problem may be pre-empted with an increased internal dialogue within STUK to transfer a kind of a meta-interpretation between different areas of regulatory oversight.
Finally, in the light of the interviews, there seems also to be a need for a harmonisation of interpretations between different regulatory systems. A growing globalisation and a diminishing number of vendors makes it increasingly important to reach a better international harmonisation of safety requirements. Such a harmonisation can be achieved only in a dialogue between STUK and other regulators in the world.
The interactions between deterministic and probabilistic safety requirements are one of the keys to the high safety level reached in the nuclear industry. These concepts seem difficult to integrate in practice and people seem often to be tuned to one or the other. Finding a correct balance between deterministic and probabilistic safety thinking has to do with the fundamental question of what is safe enough. Deterministic requirements are needed for essential nuclear safety related components and phenomena. On the other hand, there should be a cut-off probability, below which deterministic safety requirements are not more asked for.
The relationship between deterministic and probabilistic criteria can be illustrated by a simple example. Consider a certain safety function at a nuclear power plant, which is designed according to applicable deterministic principles and with a certain reliability target in mind. Can the deterministic criteria be relaxed if the reliability of the system can be shown to be very high? Safety functions are typically implemented with systems, which for their function rely on various auxiliary systems. What principles should then be applied for setting deterministic and probabilistic requirements on such auxiliary systems? Deterministic design principles can sometimes be used to eliminate certain failure mechanisms and then it would be fair to credit for that property in a probabilistic analysis. Similarly if some sequence can be shown to be very unlikely in a probabilistic sense, then a corresponding accident analysis should be possible to restrict to a few representative cases.
This issue has also an application on the requirements set for the so-called process initiating events. What kinds of reliability requirements are prudent and reasonable for functions and systems, which are needed to cope with certain disturbances, incidents and accidents? When should the single failure criterion be applied also for sequences that can be considered very unlikely and what kind of credits can be given for diversity in functions. A resolution of these questions has to reflect a view on residual risks. It would be beneficial if STUK could enter a discussion on the relationships between such deterministic and probabilistic criteria and their interpretation in a few selected illustrative cases.
The classification system, by which functions, structures and equipment are graded with respect to their importance for safety, is a key to many other issues. If the classification system and the requirements in different classes cannot be agreed upon, there is little prospect that present disagreements and confusion will disappear. In this connection it is important to note that the concern is not only the classification system itself, but also how it is interpreted and how different functions, structures and equipment actually are allocated to different classes.
A second issue connected to the safety classification has to do with changes introduced in YVL 2.1. The guide has been changed two times and the nuclear power plants made serious attempts to comply with the new requirements. It became soon evident that a total plant reclassification would be counterproductive and it was agreed that only plant modifications would be classified according to the new system. This principle is practical, but it also has a potential of introducing confusion. It was a widespread opinion among the persons interviewed that the work connected to a reclassification actually has created considerable costs without any significant influence on safety.
Experience from incidents as well as PSA results demonstrate that the conventional systems can have an important influence on safety. Classification practices of today do not typically recognise this fact. If the principles for the classification of functions, structures and equipment are reconsidered there might be an opportunity to reconsider also principles for handling presently non-classified systems such as for instance the ultimate heat sink.
YVL 5.5 would also warrant some rethinking on how to carry out the classification. There have been international standardisation efforts aimed at a better approach for classification of instrumentation and control functions and systems important to safety. It is too early to predict in which directions these efforts will lead, but it is clear that STUK should involve itself in the discussions.
Any system of requirements should be updated to reflect the technical development. A reasonable requirement is also that nuclear power plants built today should be better than those built twenty years ago. The way such a general feeling for the need of improved solutions should be converted into safety regulation is a different matter. In this connection there may be a need for opening up a societal discussion on the risks of nuclear power as compared to risks of other sources for primary energy. STUK should probably not be too much involved in such a discussion, but an emerging debate may place a need on STUK to explain the content and the assumptions of nuclear regulation in a language, which can be understood by educated laymen.
Another question is how regulatory oversight will change in the future. STUK has had an outspoken policy to move away from inspecting technical details to inspect and review work processes of the licensees. According to the interviews such a development would be welcome, but signs of such a change have been small so far. Instead many of the interviewed expressed the opinion that the regulatory oversight had been moving further into details.
STUK has selected not to be directly involved in matters concerned with economics. Experience from nuclear power plants in the world demonstrate however, that safety can be achieved only when the economic situation of a nuclear power plant is sound. This observation may have some influence on future regulatory approaches.
One issue to be considered is how modern information technology can be utilised efficiently. It is evident that the technology has many potentials, which will require considerable investments before they are fully realised. Experience from information technology projects call however for realism in the expectations. STUK has in this connection taken a small, but welcome step forward in making the regulatory system easier to access in a new computer based system.
A final question towards the future has to do with how the technical development will continue. The conventional industry has already in many fields surpassed the nuclear industry in its quality requirements. Will this influence the nuclear regulatory systems and if so in what way? The nuclear industry itself is considered too small by the large international vendors to motivate the development of specialised nuclear products. This may lead to a situation where special nuclear grade equipment can not be bought at any price, or the equipment available is inferior as compared with normal industrial grade. It seems however likely that a requirement that the technology used should be proven will stay, but such a requirement should not be allowed to stop a search for better solutions. Again it is important that STUK takes an active role in a discussion of future directions in the development of regulatory oversight.
This chapter gives the general conclusions of the study. The main conclusion must be that the YVL-guides have to be considered as a large asset for the Finnish nuclear community. The guides in general can not be considered too prescriptive. There are however a number of improvements which could be made in the structure of the YVL-guides, in the content of specific YVL-guides, in the way they are interpreted and in the processes for keeping them up to date.
In a general evaluation of the YVL-guides one can conclude that they fulfil earlier described criteria reasonably well. They have a structure, which is logical and covering. They are relatively well balanced with a reasonable level of detail. The requirements put forward in the YVL-guides are reasonable and they are reflecting international practice. The guides are understandable and fairly straightforward to interpret. STUK is putting in a considerable effort to keep the guides up to date.
The main question asked in the assignment was whether or not the YVL-guides could be considered too prescriptive and binding for the nuclear utilities in Finland. Based on the interviews and a general assessment, this question has to be answered with a definitive no. All persons interviewed had a clear positive view of the YVL-guides and they were seen as giving structure to the safety activities at the plant. This positive view has however to be qualified with respect to a few problematic YVL-guides. These guides have been treated more in detail in the earlier chapters of the report.
The YVL-guides can be considered as an asset of the Finnish nuclear regulatory system. It is clear that STUK should continue the work they do in keeping the YVL-guides up to date. The strategy for further development of the guides should however be reconsidered, discussed and documented on a continuing basis. The possibility that a new nuclear power plant will be built in Finland gives a number of new challenges to STUK which have to be reflected.
In a long-term perspective, the position of the YVL-guides as a component of the regulatory oversight in Finland may change. The extent to which there will be an international harmonisation of regulatory guidelines remains to be seen. It is however evident that the nuclear community all over the world would benefit from more harmonised approaches to safety. With the present experience and skills STUK could most certainly play an important role in this development.
The open and frank discussions during the interviews are gratefully acknowledged. All persons interviewed showed a deep commitment to nuclear safety. The comments given were by the interviewers seen as sincere attempts by the interviewed persons to improve the present Finnish regulatory system. There were no indications that the views expressed could be interpreted as an unfounded groan on something the utilities have to live with. In a way this can be seen as a reflection of a good safety culture at the organisations visited.
<< Main report | Annex II >>