CSNI-WG-RISK - Workshop level 2 PSA and severe accident management – mars 2004
STATUS STATUS OFOF IRSN IRSN LEVEL 2 PSA
Abstract
Emmanuel Raimond, Cataldo Caroli, Bernard Chaumont, Michel Durin
IRSN, France
e
During the few last years, IRSN has developed a level 2 PSA for French 900 MWe PWR. A new step has been passed with an updated version of the study achieved in 2003. This version of the study concerns the power states of the reactor. Beyond the results obtained in terms of frequency of release categories, this version demonstrates the ability of the methods chosen and of the codes used. A second version of the study is still under development, with some major evolutions: the modelling of hydrogen recombiners (which are planned to be installed by French Utility) , the evaluation of core reflooding situation during late phase of core degradation; the treatment of cold shut down states.
The aim of the paper is to highlight some of the specificities of IRSN level 2 PSA :
- the preferential use of validated codes results against expert judgements; which remains still necessary for code results interpretation and code options or parameters values;
- the severe accident computer codes used;
- a detailed interface between level 1 and level 2 PSA;
- the construction of simplified physic models for consequence evaluations of severe accident phenomena in the accident progression event tree; the construction of these models allows uncertainties evaluation and has conducted to original studies (three examples are provided in a second IRSN paper) ;
- a specific method for containment leak evaluation by systems (b mode) ;
Abstract
The paper first presents general objectives, steps of development and elements of methodology of the 900 MWe PWR level 2 PSA achieved by the IRSN.
It then highlights some of the most important specificities of the IRSN study: a detailed level 1 to level 2 interface, use of so-called physical models to represent physical phenomena in the Accident Progression Event Tree (APET), a specific and detailed model to assess containment leakage through penetrations, an on line simplified model physical model to estimate the radioactive releases and a powerful dedicated software for the APET quantification.
1. Introduction
During the last few years, the IRSN has developed a level 2 PSA for French 900 MWe PWR. A new step has been passed with an updated version of the study achieved in 2003. This version of the study still concerns the power states of the reactor. Beyond the results obtained in terms of radioactive releases categories frequency, this version demonstrates the ability of the methods and of the codes used.
The paper gives general information on the IRSN level 2 PSA approach and illustrates some of the specificities of the study.
2. General objectives
A level 2 PSA of 900 MW PWRs is developed by the IRSN with the following objectives:
- to contribute to reactors safety level assessment,
- to estimate the benefits of accident management procedures and guides to reactor safety,
- to provide more quantitative judgment elements about the advantages of any modifications to reactor design or operation,
- to acquire quantitative knowledge for emergency management teams and tools,
- to help in the definition of research and development programs in the severe accidents field
Learning from the detailed studies achieved for 900 MWe PWRs level 2 PSA are also extended to other French plants (1300 or 1400 MWe PWRs).
The study progresses according to the following steps:
- a preliminary internal version (1.0), based on the IRSN level 1 PSA published in 1990 has been completed in 2000,
- a version 1.1, achieved in 2003, is a revision of the preliminary version and integrates several improvements (detailed physical calculations of containment failure, assessment of uncertainties of radioactive releases …),
- a version 2.0, based on the updated level 1 PSA and limited to reactor power states, is planned to be finished in 2004,
- a version 2.1, also based on the updated level 1 PSA with extension to reactor shutdown states is planned for 2005.
The development of the versions 2.0 and 2.1 is currently in-progress. These versions will notably take into account hydrogen recombiners, which are planned to be installed by the French Utility.
3. Content of the study
The general methodology has been initially based on those developed in the NUREG-1150 study, and includes four main steps: (1) binning of level sequences into Plant Damage States (PDS) according to interface variables, (2) representation of important severe accident events in a Accident Progression Event Tree (APET), (3) binning of level 2 sequences into Release Categories and (4) assessment of radioactive releases into the environment for each release category.
Detailed support studies have been performed in order to quantify the APET events as regards physical phenomena, systems behavior, human actions and releases assessment, using all updated knowledge available in IRSN on severe accidents, gained either from own R&D programs, or from international cooperation.
Some of the specific methods used in the framework of the study are illustrated in the following paragraphs.
4. A detailed interface between level 1 and level 2 PSA
The Plant Damage States (PDS) are defined the combination of values of 19 interface variables. These variables (Table 1) represent initiator events, systems and containment states but also residual power and activation of the plant emergency plan. These last variables values are then used respectively for physical calculations and for human reliability assessment. The high number of interface variables (Table 1) is accompanied by a high level of description of systems states. For example, Table 2 gives the values of the AS (CHRS availability) LH (high voltage electric board) variables might have.
PT – RCS break size / SF – Component cooling or essential service water systemsPL – RCS break localization / AP – Water makeup to RCS availability
RT – SGTR number / BA – Safety injection water tank
VL – V-LOCA / SE – Secondary system break
AS – CHRS availability / SO – Pressurizer safety valve availability
BP – Low pressure safety injection availability / IE – Containment isolation
HP – High pressure safety injection availability / CR – Core criticity
GV – SG availability / PR – Residual power
LC – Electrical board availability (low voltage) / PU – Emergency plan
LH – Electrical board availability (high voltage)
Table 1 – Interface variables
AS variables values / LH variables values1 = CHRS available and in service
2 = CHRS available and not in service
3 = CHRS not available, failure occurred at demand
4 = CHRS not available, failure occurred in function – not contaminated
5 = CHRS not available, failure occurred in function – contaminated / 1 = LHA et LHB electric board available
2 = only LHA available
3 = only LHB available
4 = LHA and LHB not available
Table 2 – Examples of AS and LH variables values
Construction of interface between level 1 and 2 PSA has then led to around 150 PDS for power states of reactor. PDS have been grouped in function of the initiator event. Table 2 gives the number of PDS obtained for some initiators.
The high number of PDS introduced in the APET is probably an originality of the IRSN approach. For each PDS, a system / thermo-hydraulics transient calculation is defined. For the version 2 of the study, around 100 transients have been defined (Table 3).
Numberof PDS / Number of
thermal-hydraulics transients
LOCA (large break) / 17 / 9
LOCA (medium break) / 24 / 14
LOCA (small break) / 8 / 8
LOCA (very small break) / 10 / 10
SGTR / 20 / 15
Secondary break / 13 / 13
Loss of heat sink / 13 / 10
Loss of steam generator water injection / 17 / 17
Total loss of electrical power / 12 / 6
Table 3 – Number of PDS (900 MW PWR level 2 PSA V2)
These transients are calculated with a version, named SCAR, of the SIPA 2 simulator that includes the thermal-hydraulics code CATHARE 2. In the context of PSA, use of a simulator with an extended level of qualification, allows a better modelling of plant systems and human actions and allows calculation of a large number of transients.
Advantages of this approach, in comparison with a simplified interface (and a limited number of calculations) are:
· to obtain a better evaluation of accident kinetics and to improve evaluation of delay before radioactive releases,
· to consolidate some level 1 PSA assumptions (a feed-back of the results for level 1 PSA is done),
· to define more precise conditions for severe accidents phenomena calculations,
· to provide a large panel of “best-estimated” transients that can be used in other contexts (accident management, safety analysis).
5. APET - quantification of physical phenomena with uncertainties
The different physical phenomena that might occur during a severe accident are explicitly represented in the APET. They have been organized in «physical models» so that :
· each physical model represents a set of physical phenomena tightly coupled because of feedback processes or temporal dependencies,
· two separate physical models are linked by a limited number of variables which can be transmitted by the APET.
Application of these principles has led to define the following physical models:
· accident progression before core degradation (BCD model)
· accident progression during core degradation (DCD model)
· induced SGTR in case of core melting with a pressurized RCS (I SGTR model)
· hydrogen combustion during core degradation (H2 model)
· advanced core degradation (ACD model)
· in-vessel steam explosion and mechanical consequences (IVE model)
· ex-vessel stem explosion and mechanical consequences (EVE model)
· direct containment heating (DCH model)
· accident progression after vessel rupture (melt-corium concrete interaction) (MCCI model)
· combustion during MCCI (H2 –CO model)
· containment mechanical behavior (CMB model).
The concatenation of the different physical models done in the APET, as illustrated in Figure 1 represents the accident progression from the physical point of view. APET has also been separated in 4 phases: before core degradation, during core degradation, vessel rupture, after vessel rupture.
Figure 1 - Physical models of the APET
In the IRSN approach, construction of the APET physical models is based, as far as possible, on results obtained by validated codes calculations. Expert’s judgments are used for result interpretation and when direct code calculations are not possible.
The Figure 2 presents the codes used for each APET physical model.
Figure 2 – Codes used in 900 MWe PWR level 2 PSA
For the construction of physical models, 2 methods are used:
· response surfaces (IVE, EVE, DCH, MCCI, ACD models)
· tables of results (DCD model).
The first method is applied when scenarios effects and discontinuities are not too important. The IRSN approach consists of the construction of response surfaces that calculate downstream variables (results) in function of upstream variables (state or uncertain variables). The method leads to an intensive sensitivity analysis which gives information on uncertainties. Details are provided in a second paper of the workshop.
For core degradation progression (DCD model), strong scenario effects and discontinuities have to be taken into account (valve opening, RCS cooling with steam generator, accumulators discharge, water injection into RCS…). Preliminary studies have shown that the construction of a response surface is a very difficult task and finally the surface is poorly statistically representative of the physical calculation results. So other approach has been preferred.
Calculated transients are then defined by the following way (Figure 3):
· for each PDS from level 1 PSA interface, a system-thermal-hydraulics transient (SCAR) is defined,
· each system-thermal-hydraulics transient is continued by two ASTEC V0.4 core degradation calculations (taking into account or not actions recommended by severe accident management guides).
Figure 3 – definition of calculated core degradation transient
Calculated transients are characterized by identification variables values. A table of results can be established (Table 4).
Transientnumber / Identification variables values / DCD downstream (results) variables values
Table 4 – table of result for core degradation calculations
The table 5 gives examples of DCD downstream variables
Core decovering instant / Vessel breach instantBeginning of fuel damage (1100 °C) instant / Primary pressure at vessel breach instant
Automatic CHRS start / Vessel head temperature
Automatic CHRS start instant / % zirconium oxidation
Melt-corium falling instant / Containment pressure at vessel breach instant
Table 5 DCD Examples of downstream variables
Use of transients results by the APET (DCD model) is made by the following way :
· for each considered scenario (depending on systems availability, human actions, residual power…), the APET has to choose a representative transient ; choice is done according to the identification variables values by a selection tree,
· downstream (results) variables values useful for accident progression evaluation are then extracted from the tables of results for the selected transient (table 4),
· for some of the downstream (results) variables values, uncertainties are taken into account (hydrogen mass in containment, melt-corium mass available for in-vessel steam explosion e.g).
6. A specific model for containment leakage through reactor building penetrations
A detailed study of the containment leakage, due to pre-existing leakage or to isolation failure during the accident, has been performed. A software, named BETAPROB, has been developed in order to identify all the possible leakage paths from the containment to the environment through reactor building penetrations and auxiliary buildings and to quantify the couples (probabilities, section) of containment leakage.
A model is constructed with the following assumptions:
· a description of the systems which penetrate into the reactor building (hydraulics components, valves, pump, sumps) and the rooms of the auxiliary building (ventilation, filtration level) is constructed with BETAPROB,
· failure probabilities, coming from operation experience feedback, are affected to the different components (l, failure in operation and g, failure on demand),