Read Development of a Reliability Data Handbook for Piping Components in Nordic Nuclear Power Plants text version

Proceedings of the 8th International Conference on Probabilistic Safety Assessment and Management May 14-18, 2006, New Orleans, Louisiana, USA



Anders Olsson/RELCON AB Michael Knochenhauer/RELCON AB Bengt Lydell/Erin Eng. Inc

SUMMARY/ABSTRACT Even though pipe breaks may not be a significant risk contributor in many PSAs it is none the less very important that the underlying data reflect current knowledge about piping components and their susceptibility to different degradation mechanisms. Within the Nordic PSA Group (NPSAG) a R&D project has been initiated that will look at the possibilities of using the international OPDE pipe failure database and statistical methods developed for RI-ISI in order to create a handbook (The "R-book") containing frequencies for different kind of piping component failures. One of the main purposes with this R&D project is to make sure that no matter what the application is, the data source for piping component failures shall be validated and common for all applications. This paper describes the R&D project with its different subtasks.


Copyright © 2006 by ASME

INTRODUCTION Piping reliability data is necessary for many different purposes within the safety regulatory work for Nuclear Power Plants. It is used in Probabilistic Safety Assessment (PSA), Risk Informed In-service Inspection (RI-ISI) of piping components, follow-up of occurred events and ageing mechanisms for piping components etc., just to mention a few examples. As for all kind of reliability data it is a delicate task to estimate the failure probability or frequency that is to be used in the desired application (PSA, RI-ISI etc.). Different applications may have different needs of degree of detail in the data; in PSA for instance it might be sufficient to determine the failure data for a part of a piping system while RI-ISI requires detailed information for each piping component (welds, T-joints, valves, bends etc.). Furthermore, it is not certain that the same people are involved in work with different applications and therefore they are not totally aware of the work that is being done in similar fields. In the Nordic countries a reliability handbook called the "T-book" has been developed for a number of safety important components that are found in Nordic nuclear power plants (NPP), but without including any piping components [1]. The purpose with the T-book is to provide input for PSA for the Nordic NPP's and also to make sure that the reliability data used in different PSA's is produced in a controlled manner. For piping components the situation is different, since no common database or method of producing the reliability data needed exist that can be easily used by the PSA engineer or within other fields of expertise. This lack of common methods means that it is difficult to compare different PSA's with respect to risk caused by piping failures. Furthermore the data used is often outdated, originating from the WASH-1400 study for example. These differences between the PSA's are of course undesirable. Also, the fact that it is difficult to update the data has the effect that when the need for data arises, for RI-ISI as an example, new data has to be produced, which is often both time consuming and quite costly. THE OPDE DATABASE Within the Nuclear Energy Agency of the OECD, a project is ongoing hat goes under the name OPDE ­ OECD Pipe Failure Data Exchange Project [2]. The objectives of the OPDE project include: · · · · Provide a validated database on pipe failures in commercial nuclear power plants from 1970 to present. Collect and analyze data to promote a better understanding of underlying causes. Generate qualitative root cause analysis insights. Support quantitative reliability assessments, including risk-informed applications.

The database was finalized in 2005 (version 1.0) and thereby an international pipe failure database has been established that is available for practical applications. When the OPDE project started in 1994 the main aims of the project were: · To review and update the "WASH-1400 perspective" on piping reliability (including LOCA frequency estimates) on the basis of existing operational experience. · To establish a comprehensive, up-to-date information resource for piping operating experience · To maintain the database and to make it available for applications ("T-Book philosophy", see [1]) The current aims for the project are: · Monitoring of aging trends (if any), evaluation of the effectiveness of ISI/RI-ISI and mitigation strategies, identification and tracking of new forms of degradation · Enhancement of quantitative methods and exploration of a "blended" approach to quantitative piping reliability analysis · Encouragement of database applications and of the exchange of operating experience


Copyright © 2006 by ASME

RELIABILITY HANDBOOK FOR PIPING COMPONENTS ("R-BOOK") Within the Nordic PSA Group (NPSAG) a R&D project has been initiated that will look at the possibilities of using the OPDE database and statistical methods developed for RI-ISI in order to create a handbook (the "R-book") containing frequencies for different kinds of piping component failures. The frequencies listed in the handbook can be used in order to obtain LOCA frequencies for PSA and as input for other applications as well, for instance PFM (Probabilistic Fracture Mechanics). Even though pipe breaks may not be a significant risk contributor in many PSA's it is none the less very important that the underlying data reflect current knowledge about piping components and their susceptibility to different degradation mechanisms. This is often dealt with in RI-ISI programs. When setting up an RI-ISI program one of the main tasks is to define the piping component failure data. One of the main purposes with this R&D project is to make sure that no matter what the application, the data source for piping component failures shall be validated and common for all applications. The R&D started in January 2006 and therefore no conclusions or findings of the project will be presented in this paper. Instead, the focus will be on describing the tasks that will be carried out within the R&D project. Scope of the R&D project The project has been divided into two phases, Phase 1 and Phase 2. Phase 1 is a pilot study that will act as a basis for the actual development of the R-Book, which will be done in Phase 2. As of today, only Phase 1 has been decided on and the decision about Phase 2 will not be taken until the pilot study has been completed. Project participants Participants in the R&D project are the following organizations: · · · · · · Oskarshamn Kraftgrupp, OKG ­ Swedish NPP Forsmark Kraftgrupp, FKA ­ Swedish NPP Ringhals, RAB ­ Swedish NPP SKI ­ Swedish Nuclear Power Inspectorate RELCON AB ­ Risk Management Consultant ERIN Engineering Inc. ­ Risk Management Consultant

Why use the OPDE database? Why the R&D project has chosen to use the OPDE database for its purpose is a justified question, as several other databases exist that could have been used. In order to answer this question a categorization of different database types can be used. Three main categories can be defined for piping reliability databases: 1. Databases developed to fulfill a one-time need ­ after the one-time objective has been fulfilled, no further database maintenance has been made. Furthermore, no specific QA requirements have been developed and no second-party reviews have been made. Doubts about the completeness of the database therefore remain and must be solved before the database is used. Databases developed as a part of long-term R&D, allowing multiple users and applications. Access rules and continuous database management and quality requirements are formulated in a quality program. Proprietary databases, i.e. a utility-/plant-specific databases, supporting augmented inspection programs for specific damage mechanisms (FAC, IGSCC, MIC, thermal fatigue). Access is often restricted to utility personnel.



When different databases are examined, the following observations can be made: · · · Many "Type 1" databases exist There is also a large number of databases of "Type 3" Beyond the OPDE database, few (if any) known "Type 2" databases on piping degradation and failure exist


Copyright © 2006 by ASME

The choice of the OPDE database was therefore an easy choice that came natural to the project participants. R&D PROJECT TASK DEFINITION ­ PHASE 1 The objective of Phase 1 is to address the viability of developing a Nordic Piping Reliability Data Handbook against the progress that has been made in piping reliability over the past years. A key output of Phase 1 will be the definition of piping reliability parameters to be considered in the handbook. The Phase 1 deliverable will also include brief summaries of current piping reliability developments, including event databases, analysis techniques and tools. Additionally, Phase 1 will include descriptions of the capability of existing event databases, techniques and tools that would be considered in the development of a Nordic Piping Reliability Data Handbook. Subtask 1 ­ Existing Pipe Failure Databases This task will identify and review the efforts to develop pipe failure databases. On the basis of a discussion of data quality and validity issues, this task will summarize the insights gained from independent reviews and applications of these and other databases. Subtask 2 ­ Database Users The anticipated users of the R-Book will be identified in Task 2. The user requirements will be established through interviews with utility personnel; PSA engineers, maintenance staff, inspectors (ISI), and material experts. This task will also include interviews with those using physical models (PFM) since they also are a potential user category of the R-book. Subtask 3 ­ Identifying the Data Needs Statistical parameter estimation should be done against a clear definition of piping reliability. The question then becomes whether an event database is of sufficient detail to support the requirements that are imposed on it by a piping reliability model. Ultimately it is the piping reliability applications requirements that provide directions for how to approach piping reliability analysis. This question has been addressed extensively in the work leading up to the OPDE Project. Equally important is the subsequent work to address the validity of different piping reliability analysis concepts (e.g., physical models versus parametric models). In identifying the data needs, the proposed work will address the requirements on absolute and relative piping reliability estimates. In risk-informed applications such as risk-informed in-service inspection (RI-ISI) it is sufficient to work with relative reliability estimates since the RI-ISI selections are made on the basis of CDF (Core Damage Frequency) and LERF (Large Early Release Frequency). Other types of applications place more emphasis on absolute piping reliability estimates. Specifically, this task will address analytical requirements (including uncertainties) of: · · · · LOCA frequency estimation, Internal flooding initiating event frequency estimation, Reliability assessment of alternative piping system designs, Risk informed applications (e.g., RI-ISI).

The consideration of piping reliability `attributes' and `influence factors' is a key to the parametric models. The methodology of SKI Report 98:30 [3] (i.e., the parametric attribute/influence methodology) has been applied in an U.S. Nuclear Regulatory Commission (NRC) R&D project to develop LOCA frequency distributions. The methodology is applied to five base cases (two involve BWR ASME XI Class 1 systems, and three involve PWR Class 1 systems). An important aspect of the NRC R&D project is the validity of parametric and physical models. Here the PRAISE, PRODIGAL codes and the parametric attribute/influence method are applied to the five base cases to establish a basis for comparison and to identify strengths and weaknesses. Task 3 will summarize the insights from this comparison: · · Impact of different assumptions on results, Correlation between leak rate and pipe failure rate.


Copyright © 2006 by ASME

In the parametric attribute/influence method, an attribute relates to the design and application of codes and standards in view of the service requirements. An attribute factor cannot be modified without changing the design of a piping system; e.g., replacing high carbon content austenitic stainless steel with nuclear grade stainless steel, or replacing carbon steel with super-austenitic stainless steel. An influence factor relates to the environmental and operational conditions that affect the susceptibility of piping to certain damage or degradation mechanisms. In data analysis, the attribute and influence factors are used to correlate the event data with specific piping systems as defined by material, diameter and susceptibility to degradation/damage mechanisms; this is the data binning process. An estimate of pipe failure frequency is developed from the following expression: fFAILURE = f(Event Data; Exposure Data) Where: "Event Data" is defined by the number of observed pipe failures in a given plant system (this is failure count) and "Exposure Data" defines the total number of piping components exposed to a certain damage or degradation mechanism (this is the success count).

An essential aspect of data analysis is to ensure that there is correspondence between the event data and exposure data. As an example, if we were to derive piping reliability parameters for BWR piping believed to be susceptible to IGSCC (Intergranular Stress Corrosion Cracking), the exposure term must account for the total number of piping components (i.e., welds and weld heat affected zones) exposed to this degradation mechanism. As another example, for piping susceptible to flow accelerated corrosion (FAC) the exposure term is defined by the total length of FAC susceptible piping. In summary, to perform piping reliability analysis we need two general types of input data: 1. 2. Event data, and Population data.

This task will include examples from recent applications to demonstrate the use of the attribute/influence concept in performing piping system design evaluations. Subtask 4 ­ Piping Population Data Phase 1 will explain the role of piping population data and its different (or application-specific) forms. There are some misconceptions about the appropriate definition of an exposure term for use in parameter estimation; i.e., the use of `pipe segments' versus `weld counts' or `length of piping'. Therefore, this task will demonstrate the importance of working with appropriate exposure terms. In addition, the task will address existing collections of piping design information, including piping component population data. CONCLUSIONS There is no doubt that the OPDE database can be used for producing failure data for piping components, that is one of the main purposes for the database. The main question however, is to define how the database can be used in order to produce a handbook that is suitable for different disciplines for this kind of data. It is therefore important to identify the requirements from those disciplines that will be the potential users and also to determine the tools that are to be used. This is the main objective for the R&D project initiated by the Nordic PSA Group (NPSAG). Development of a handbook, similar to the T-Book [1], will improve the quality of PSA and related applications, e.g. RI-ISI, and furthermore "unnecessary" differences between PSA studies, with respect to piping failures, will be minimized. The project as such will also be an important test case for the applicability of the OPDE database and therefore the project as such should be of interest for other members of the OPDE project as well.

REFERENCES [1] The TUD Office and Pörn Consulting, 2005, "T-book, Reliability Data of Components in Nordic Nuclear Power Plants ­ 6th edition," Sweden.


Copyright © 2006 by ASME

[2] OECD Nuclear Energy Agency, 2005, "OECD Pipe Failure Data Exchange Project (OPDE) 1st Term Status Report, OPDE(2005)1, Issy-les-Moulineaux (France). [3] Swedish Nuclear Power Inspectorate, 1999, "Failure Rates in Barsebäck-1 Reactor Coolant Pressure Boundary Piping. An Application of a Piping Failure Database, SKI Report 98:30, Stockholm (Sweden). [4] U.S. Nuclear Regulatory Commission, 2005, "Estimating Loss-of-Coolant Accident Frequencies Through the Elicitation Process," NUREG-1829, Washington (DC).


Copyright © 2006 by ASME


Development of a Reliability Data Handbook for Piping Components in Nordic Nuclear Power Plants

6 pages

Find more like this

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate