Summary of Meeting Paper

The 1996 Annual Meeting of the Society for Risk Analysis-Europe

Approach for Risk-Based Regulation and Risk Management of Nuclear Power Plants. H.P. Berg, Bundesamt für Strahlenschutz, Salzgitter, Federal Republic of Germany

INTRODUCTION

At international level, risk analysis and risk assessment techniques are increasingly used in assessing, accepting or validating the current safety state of a nuclear power plant built to earlier standards. This has led to a more stringent incorporation of the risk assessment into the regulatory framework in some countries.

In the Federal Republic of Germany, the nuclear licensing procedure is essentially based on deterministic safety analysis. In the context of periodic safety reviews of nuclear power plants in operation, also probabilistic considerations can be taken into account; this is a first smoothly step towards the international activities, but probabilistic safety assessment (PSA), at present, only supports but does not determine regulatory decision making.

However, when properly applied, the results of a PSA can be used to identify and prioritize the importance of hardware, human actions (operation and maintenance staff activities) and plant procedures to plant risk. The information contained in a PSA is also important in the development of a sound risk management programme that could be used for decision-making purposes.

An integral part of a comprehensive risk management programme at a nuclear power plant would be a living PSA that could be used as the basis for day-to day operational and maintenance activities and for the short and long-term assessment and prioritization of safety-related needs. Regarding risk management, the idea of such a living PSA has been supported in the framework of the German Nuclear Regulatory Research Programme.

RISK BASED REGULATION

Risk-based regulation is the utilization of the modem PSA tool in order to better distribute the resources of both the regulator and the nuclear industry. More specifically, resources would be distributed according to risk significance. Those items or events that have high risk significance would receive the most attention, while those with little risk content would command fewer resources.

Risk-based regulation has the potential of both improving nuclear power plant safety and reducing plant operating costs. This modem form of regulation could be applied to present operating plants and to advanced designs. In fact, it would help to quantify the safety improvements of advanced designs.

The application of PSA technology to the regulatory process can reduce public risks in several ways: by finding design weaknesses, by improving plant operations, and in developing severe accident management programmes. For example, a traditional product of a level I PSA is some estimate of the likelihood of a nuclear power plant having a core damage or core melt event. The overall core melt frequency is estimated by summing up many thousands of accident sequences, each providing some increment of core melt frequency. If a level 2 PSA is performed, then there will also be estimates of the containment failure frequency and the releases of radioactive material to the environment associated with each containment failure. A large sum, e.g., a high core melt frequency or high containment failure frequency, can be an indication of a poor design. Further, by examining the different contributors to the aggregate value, particular areas of design weakness can be pin-pointed. Therefore, PSA results can be used to evaluate the design of nuclear power plants. Numerous plant-specific design improvements have already been implemented in various nuclear plants based on PSA insights, thereby lowering nuclear risks.

More recently, PSA techniques have been applied in evaluating the operation of nuclear power plants focussing the interest on how plant risks vary with time. There are several mechanisms that can cause plant risks to change over time. The performance of individual components and whole systems may degrade due to aging or improve due to design modification or enhanced maintenance. Plant configurations also change from time-to-time as certain components are removed from (or restored to) service for tests and/or maintenance, while others may be removed through failure. Configurations also change when going from one plant operating mode to another, such as going from power operation to shutdown. Since the risk significance of a component or system is also a function of the plant's configuration, changing configurations yield different risk levels.

Just as earlier applications of integral PSA results were utilized to reduce the risks due to design weaknesses, present specific applications are increasingly dedicated to minimizing operational weaknesses, e.g., avoiding high risk plant configurations.

A major step towards risk-based regulation is the Final Policy Statement on the Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities" published by the US Nuclear Regulatory Commission in August 1995 which establishes an overall policy on the use of PSA methods in nuclear regulatory activities. It is stated that already existing probabilistic safety criteria and subsidiary numerical objectives are to be used with appropriate consideration of uncertainties in making regulatory judgments on the need for proposing and backfitting new generic requirements on nuclear power plant licensees.

Quantitative safety criteria and objectives -- correlated with the risk of each single individual in the vicinity of the plant and/or the societal risk of the population as a whole -- are used in the decision-making process, for example, in the United Kingdom and in the Netherlands. In both countries, this safety concept is not restricted to nuclear installations but it has been adopted as a more global safety policy regarding all potential hazardous industries and activities.

BENEFITS AND DISADVANTAGES OF RISK-BASED REGULATION

Although some countries, in particular USA, are strongly discussing a proposed transition process from deterministic to risk-based regulation, the pros and cons of risk-based regulation including challenges of such a transition process should be taken into account which are summarized in the following.

Main benefits of a risk-based regulation are:

On the other hand, there are a lot of disadvantages and difficulties which are posed by such a goal-setting approach:

Challenges which are associated with developing and implementing risk-based approaches to regulation are:

As desirable as it might be to write regulations in terms of the ultimate measure of probabilities, it is not even nearly possible to define the probability of a possible accident sequence with enough precision and enough replicability to use such probabilities as terms in the regulations' bottom line.

The fact that PSA results are not sufficiently robust is the main reason that PSA could not be the sole tool or basis for creating a new regulatory regime in the near future. Therefore, the determination of probabilistic safety goals is not supported in Germany from the legal point of view, neither as probabilistic limits nor as orientation values.

RISK MANAGEMENT TOOLS IN GERMANY

While many PSA projects were originally performed to address a specific need or decision, the living PSA concept is based on a dynamic, regularly updated model. Such a living PSA tool is, in particular, of most benefit to the utilities for raising the reliability of plants and increasing their availability. Nevertheless, it has been seen as a federal task to support the idea of such a living PSA.

Therefore, the tool Safety Analysis and Information System (SAIS) has been developed in the Federal Republic of Germany. SAIS consists of a plant specific level 1 + PSA including PSA models, data and computer tools for the modification and reevaluation of the event and fault trees. Moreover, system and component data and graphics are also part of SAIS so as to provide supporting information for plant engineers. This second option of SAIS enables to record plant specific data and experiences of the personnel which has been seen to be helpful in case of the assessment of the current safety status of nuclear power plants built to an earlier standard, e.g., in case of the Russian nuclear power plant in Kola. However, the use of PSA as a living PSA in German plants is -- at present -- still very limited.

The second example for a risk management tool is the computerized Reliability Adviser System (RELADS) which can be adopted and used by safety analysts as a living PSA and/or by the responsible plant staff as a risk monitor. This system is still under development. The main aim of RELADS is to provide the user on-line with the following informations

based on event and fault trees and cut sets of an already existing level 1 PSA.

The risk-level change is calculated and presented on three levels: loss of system function, occurrence of an event sequence and occurrence of a core damage (which could lead to a major accident). An optimization routine calculates the repair prioritization if more than one component is inoperable.

One example of an already running risk monitor on international level is the Essential System Status Monitor developed in the United Kingdom and implemented at Heysham 2 to provide power station operators with an on-line aid for planning plant unavailabilities required for maintenance in a coordinated manner.

CONCLUSIONS

For the time being, the deterministic and probabilistic methodologies are mainly kept separately. In a first step, all the deterministic rules have to be interpreted from a PSA and risk management point of view. Furthermore, the results have to be integrated into a consistent and realistic framework. This is certainly a major task (involving a large number of judgements) which may be worthwhile to be started in the future.

Considerations on possible probabilistic (i.e. quantitative) safety goals and criteria are, at present, not performed with high emphasis in Germany, but they may become more important with the progress of future reactor concepts like the European Pressurized Water Reactor, a common French-German project where the vendors have determined internal probabilistic safety goals to be reached by the design of such plant.

In particular, a risk-based procedure in the regulatory framework of the Federal Republic of Germany -- as highlighted by the US Nuclear Regulatory Commission -- is not part of current revisions of existing ordinances and safety criteria in the near future. On the other hand, first developments of risk management tools are underway and have to prove their benefits in practical applications as living PSA or risk monitor.