The present project was implemented through two contracts, one with a Consortium of EU organisations: Framatome ANP, Forschungszentrum Rossendorf (FZR) & University of Pisa (contract 61936; project R2.02/02A); and the other with the Russian nuclear design and engineering organisation, EDO Gidropress (contract 97056; project R2.02/02B). The objectives and achievements of the overall project are described below.
All Russian nuclear power plant units were constructed according to design of the sixties and seventies, in accordance with the standards and technical documentation of these times. On the basis of worldwide experience in the operation of nuclear power plants and analysis of the causes and consequences of Nuclear Power Plants (NPP) accidents, the nuclear power plant safety requirements have changed substantially in time.
The response of the reactor core to transients leading to asymmetric conditions at core inlet strongly depends on the degree of mixing of the coolant from the different loops of the reactor before reaching the core, i.e. in the downcomer and lower plenum of the reactor pressure vessel. It has been shown that the ‘ideal’ assumptions (from the point of view of modelling simplicity) of “perfect mixing” and of “absence of mixing” can lead to a wide variation of the predicted consequences in the reactor core; the former leading to non-conservative results, the latter possibly leading to significant overestimation of the consequences.
A more realistic method of modelling this mixing process is necessary. Preferably, this should be sufficiently simple that it can be incorporated in a code intended for use as an engineering tool.
The specific objective of this project was to bring to the Russian nuclear power companies and to linked Russian research/design institutes a set of validated experience and capabilities to address this issue according to the best international practices.
To achieve this objective a specific experimental program and extensive pre and post-test calculations were foreseen in order to provide material for validating the capability of the developed system codes against the obtained experimental data, representing expected physical situations during NPP postulated initiating events.
A “realistic” method of the modelling mixing process in reactor vessel is necessary to understand the phenomena correctly and preferably it should be incorporated in a code intended for use as an engineering tool. Such code needs to be an industrial fast running tool for safety analyses.
A complete working program was set up including an initial research of what was done for other existing NPPs and actual tendencies, definition of physical situations to be addressed, definition of experiments and extensive pre and post test calculations with both selected system codes and Computational Fluid Dynamics (CFD). The material was an important basis for system code validation and/or highlighted the need for future activities.
The following steps were then undertaken:
- identifying the status of the art in core mixing
- reviewing of existing tools and their capabilities
- defining of the “code validation matrix”
- defining the list of experiments to be performed
- choosing the “system code” to be validated
- executing of experiments
- performing pre and post-test calculations by selected system codes and by CFD
- comparing simulated results with the previously obtained experimental data
- Within the scope of the project was to validate the selected system codes, in order to use them for future licensing purposes.
At time of the licensing (80s) the Safety Analysis Report (SAR) considered only the Steam Line break and the pump seizure accidents and the mixing models were empirical based on experiments or modelled “no mixing”. Later, (90s) the Boron Dilution Transients started to be taken into account, and CFD became available, allowing deeper understanding of mixing phenomena and of experimental results. The current general tendency is:
- a wider use of CFD codes to extrapolate test results to reactor conditions,
- a wider use of CFD for new analysis and for existing NPPs (applications were done in France, Germany and Russia for four loop plants, i.e. EPR, Konvoi, VVER 1000)
- for SAR applications, the development of more accurate mixing models for system codes as CFD are still too expensive in resources
- still using system codes with simplified approaches
This initial research confirmed the need to have a “realistic” method of modeling the mixing process in the reactor vessel, in order to both predict in a less conservative way the reference transients and to take into account the new transients not considered at the time of licensing.
In different countries are used 1D system codes (MANTA, RELAP5, ATHLET) using “cross connection junctions” or “pre-defined mixing matrices” or “sector adaptive mixing model” as simplified approaches. Such an engineering mixing modeling approach requires a case-specific in-advance adaptation of their parameters based on experiments and CFD calculations and has limited applicability to blind calculations of transients for which no experimental background exists. However, when it is used to predict a well defined and verified range of transients, its application is fast running and sufficient for nuclear reactor safety assessment.
Concerning Russian approaches for modeling mixing, the program module KAMERA has to be mentioned. Such program module KAMERA belongs to the class of coarse mesh 3D thermo-hydraulic codes. It has been implemented into three different Russian system codes (DKM, TRAP-KS and KORSAR/GP) which are to be used in future Russian licensing procedures.
Expected Initiating Events for VVER 1000 with potential for asymmetric temperature or boron distribution at core inlet were listed and were grouped as follows:
- events leading to asymmetric temperature distribution at core inlet,
- events leading to diluted slug formation in loop seals with potential for reaching core inlet in case of inadvertent pump start-up or under re-establishment of natural circulation,
- asymmetric boron injection or inadvertent injection of unborated water and asymmetric transport to core of diluted water.
From a physical point of view, the asymmetric condition at core inlet takes place for three different flow regimes, listed as follows:
- start-up of an idle pump: the corresponding loop seal is filled by non borated water and pump start-up transports very fast such water to reactor inlet;
- re-establishment of natural circulation: in long term SBLOCA, loop seal is filled by condensate (with no boron) and re-establishment of natural circulation transports slowly such non borated water to reactor inlet,
- stable pump operation: flow distribution remains constant, water condition in one loop change (temperature decrease) and are transported to reactor inlet.
For the above events, the coolant distribution is governed by the three different forces:
start-up of an inactive pump: inertial forces characterized by Strouhal number,
re-establishment of natural circulation: gravitational forces characterized by Froude number,
steady state pump operation: friction forces characterized by Reynolds number.
The code validation matrix was defined reflecting the knowledge or the lack of information for the physical situations expected in above postulated initiating events. A success criterion "system code will predict mixing phenomena with increased detail with respect to current version before implementation of mixing model and, at the same time, predicted mixing will be lower than experimentally observed for all performed experiments" was stated to judge on reached improvement for modelling mixing phenomena. Such criterion reflected the need for having a simplified and therefore fast running, industrial tool for SAR purposes.
Since there were three different flow patterns for the expected initiating events, experiments were divided in three “groups”. For each “group”, different initial conditions and/or start-up of circulation were imposed, in order to take into account additional parameters like different slug volumes, different density ratio, number of operating pumps. The total number of experiments was ten as it was foreseen in the project.
Concerning the system code(s), DKM, TRAP-KS and KORSAR/GP were selected. All three code packages used (or can use) the module KAMERA to represent multi channel core. Since the module KAMERA is : fast running, sufficient for nuclear reactor safety assessment, has the required modeling capabilities and is already implemented in Russian system codes it was therefore selected for validation. Such choice led to a total of four different possibilities for system code, corresponding to:, code package DKM using KAMERA, code package TRAP-KS using KAMERA, code package KORSAR/GP, used without module KAMERA: in this case the multi channel behavior is predicted on the basis of code itself, code package KORSAR/GP, used with module KAMERA. Extensive calculations with enlarged number of choice were expected to give a maximum of information on both module KAMERA and its results if used in different approaches.
Ten experiments were successfully performed and all were performed successfully and in schedule, with continuous assistance provided from the Consultant. Each test was repeated at least five times, in order to improve statistical accuracy of measured results and the obtained data, except some measurements, was of good quality.
The main findings for the three groups of experiments can be summarized as follows:
- first group, start-up of in idle pump: part of the tracer feeds the cold leg in reverse flow and does not reach the core inlet; the remaining part of tracer appears in the opposite side with respect to the injection loop; the transient is so fast that only moderate mixing takes place, concentration of tracer at core inlet is approximately still 80% of injected perturbation in cold leg.
- second group, re-establishment of natural circulation: the maximum concentration of tracer at core inlet does not exceed 60% of injected perturbation in cold leg and tracer concentration at core inlet is rather uniform. For these slow transients a very important effect of density difference (between slucg and bulk) was found.
- third group, stable pump operation: accumulation of tracer at core inlet takes place and maximum tracer concentration is found three transit times after beginning of the transient. At core inlet there is a well defined formation of sectors corresponding to loops with operating pumps.
Extensive pre and post test calculations were performed for each of the ten experiments, either by CFD codes or by selected system code packages. Calculation results were deeply compared with obtained experimental data.
The main findings from system code calculations can be summarized as follows:
- system codes gave a very good prediction of flow pattern for all cases with at least two pumps in operation, while they gave an incorrect prediction of flow pattern for all the other cases (stable one pump operation, start-up of one idle pump, re-establishment of natural circulation),
- a tendency for much mixing was observed. At the same time work performed in the project highlighted the need of modeling improvement: post test calculations did not exhibit such too high mixing,
- proper and acceptable prediction of section of core affected by the perturbation, and this was found true also for cases where flow pattern is wrongly predicted: i.e., prediction of affected core section is wrong in location but good in size (which is the most relevant parameter for safety analysis purposes).
The main findings from CFD calculations can be summarized as follows:
Performed work with CFD codes allowed deep understanding of mixing phenomena. This was particularly important to understand the observed major effect of density difference experiments.
For start-up on one pump, CFD predicted flow pattern and allowed following the tracer distribution inside vessel and the occurrence of perturbation at core inlet on the opposite side with respect to a cold leg injection.
A particular powerful application was the analysis of the density difference experiments where tracer concentration at core inlet was found strongly dependant on density effect. The tracer distribution predicted by CFD allowed completion of experimental data and made possible clear understanding of tracer distribution in lower plenum or in the upper part of downcomer for higher and lower density tracer respectively.
For cases with stable pump operation CFD, predicted flow pattern confirmimg the formation of sector with planes without transverse velocity.
A general tendency to slightly underestimate mixing was found and predicted tracer concentrations at core inlet were slightly over predicted with respect to the experimental values.
The use of CFD proved not yet to be at industrial level due to the excessive calculation time, but proved powerful tool to investigate key parameter for mixing phenomena and therefore tuning industrial system codes.
The analysed system codes proved a fast, industrial tool for performing complete plant calculations for events leading to asymmetric conditions at core inlet. They were found to predict very well cases with stable operation of at least two pumps: for such cases, their application for SAR purposes requires moderate penalization in order to have a reliably bounding approach for SAR purposes.
However for all the other cases of interest ‘start up of one idle pump’, re establishment of natural circulation’ and ‘stable one pump operation’ system code predicted a wrong flow pattern: their application for SAR purposes for these transients is not therefore possible at present status of the art. Nevertheless, taking into account a relatively good prediction of degree of mixing and in section of core affected by the perturbation, there is potential for using system code for SAR purposes if a ‘sufficiently bounding approach’ is developed.
If such methodology can be developed, and this can assure the prediction of a “higher perturbed area” and a “lower degree of mixing”, the predicted asymmetric perturbation at core inlet can be used for successive neutronic analysis.
If the additional assumption of locating the most powerful assembly in the zone of maximum perturbation is systematically used, the final obtained results will be ‘bounding’ for the key parameters:
- maximum neutronic power,
- maximum cladding and fuel temperatures,
- maximum generated energy during power pulse.
In conclusion, there is important potential to use available system codes for SAR purposes also for cases where prediction of flow pattern is wrong provided that a “sufficient” conservatism is used.
The work performed in the project allows today to define this “sufficient” conservatism using obtained experimental data as well as the experience gained by the CFD application to mixing processes.
CFD codes were confirmed not yet an industrial tool due to excessive calculation time; at the same time they proved a powerful tool to investigate specific key parameter for mixing phenomena and therefore tuning industrial system codes.
Future tendency of a wider industrial use of CFD codes is therefore confirmed.
One additional important output of the project related to this point was the deep experience gained in setting-up the nodalization schemes and the deep analysis of results.
Two major points for future activities were highlighted, the investigation of ‘swirling effect’ and the benefit added by an ‘improved instrumentation’.
Concerning ‘swirling effect’:
- experiments never showed occurrence of swirling effect,
- additional experimental work is required to highlight this important phenomenon enhancing mixing and the reasons originating it,
- code validation matrix has to be updated to take into account this phenomenon.
Concerning facility instrumentation:
- existing probes at core inlet proved sufficient,
- existing probes measuring value of tracer before core inlet are not sufficient to follow precisely tracer mixing from vessel inlet to core inlet,
- additional probes at vessel inlet and in downcomer are desirable to properly measure tracer inside vessel and investigate in vessel mixing.
- calibration procedure of the measuring probes should be improved making possible an “in-situ” calibration without necessity to take out the probes from the reactor model during calibration.