• Aucun résultat trouvé

Two-phase thermal-hydraulics

Dans le document EDP Open (Page 50-53)

Separate-effects or Analytical Experiments versus Integral Experiments

Chapter 3 Research on Loss-of-coolant

3.1. Two-phase thermal-hydraulics

French research in this area, carried out for the most part through cooperation betweenEDF,CEA, Framatome and IPSN, resulted in the development of theCATHARE12 simulation code[1]. This code provides detailed behavioral models of the waterflowing through the reactor coolant and secondary systems of a pressurized water reactor, from normal operating conditions up to the limits of standard design-basis conditions, i.e. fuel damage. In order to cover the widest possible range of thermal-hydraulic conditions, the liquid and vapor phases are processed separately, using a set of six equations (conserva-tion of mass, momentum and quantity of energy). Initially, modeling was mostly one-dimensional, although in some parts of the reactor, such as the reactor core,flows may not be unidirectional, owing to non-uniform power distribution or changes in geometry during the accident. In the 2000s, the code was upgraded to include the possibility of modeling multidimensionalflows and describing the behavior of droplets during core reflooding separately from that of the continuous liquid and vapor phases[2]. As it travels at high speed along the cladding, the vapor produces droplets due to a shear phenome-non; these circulate within the vaporfilm and contribute to heat transfers.

Designing this type of software to obtain a detailed description of thermal-hydraulic phenomena calls for precise knowledge of the physical laws governing mass, momentum and energy transfers at the interfaces between each phase and between these phases and the walls. For this reason, many analytical tests were performed as part of the development of this code. Most of them took place between the 1980s and 1990s in special, highly instrumented facilities built byCEAon its Grenoble site. The following is a non-exhaustive list of experimental facilities:

– CANON and SUPERCANON, for the study of depressurization and vaporization of water, first in a circular configuration, then in a geometrical configuration representative of a fuel assembly;

– MOBY DICK and SUPER MOBY DICK (SMD), for the study of two-phaseflows passing through openings representative of the breaks studied in the reactor coolant system;

– OMEGA and APHRODITE (EDF/Chatou), for the study offilm boiling around the fuel rods;

– DEBORA, for the study offlows under boiling conditions;

– SMD, for the study of friction at the liquid-vapor interface;

– PERICLES-2D, for the study of core uncovery and reflooding; in particular, it implements a mockup of three assemblies of different power levels;

– ROSCO, for the study of the veryfirst moments of core reflooding, which are characterized by an unstableflow rate, as observed under experimental conditions during large-scale experiments on reactor models (see further on).

12. Advanced Thermohydraulics Code for Water Reactor Accidents.

Research on Loss-of-coolant Accidents 17

In the 1980s, CEA, with support from EDF, Framatome and IPSN, designed the BETHSY facility on its Grenoble site, with the aim of checking whether theCATHAREcode was able to reliably predict the behavior of a reactor in an accident situation. BETHSY was a mock-up of the reactor coolant system of a 900 MWe reactor. It was a full-scale model in terms of the height of the different components, while volumes were represented on a 1/100 scale (Figure 3.1). It consisted of three loops, each equipped with a pump and a steam generator, together with the secondary system components considered essential for thermal-hydraulic studies. The facility was designed for pressures of 17.2 MPa in the reactor coolant system and 8 MPa in the secondary system. The reactor core was represented on a 1/100 scale by 428 electrically heated rods with stainless steel cladding.

Their power output was 3 MW, i.e. roughly 10% of the nominal power output of a reactor on the scale considered, thus allowing simulation of the residual heat in the core just after rod drop. All the engineered safety systems were reproduced, including the high- and low-pressure injection systems, the accumulators and the secondary system safety valves. Breaks could be simulated at various points in the reactor coolant system: in the cold leg, in the hot leg, at the top of the pressurizer, and in the steam generator. More than 1000 measuring channels monitored changes in key parameters during the tests (temperature, pressure, rate and direction offlow, void ratio, etc.).

Figure 3.1 The BETHSY loop (900 MWe PWR, three loops). © Georges Goué/IRSN-Source CEA (left), CEA (right).

In all, more than 80 tests were performed between 1987 and 1998. They did not focus only on large-break LOCAs (complete pipe break). Other accident scenarios were studied, including those involving small or intermediate breaks, or nitrogen injection in the reactor coolant system after total drainage of the accumulators, or loss of cooling during an outage when the reactor coolant system is partially drained. The facility was also used to perfect control procedures (as part of the new state-oriented approach) to be imple-mented by licensees during the different phases of an accident to bring the reactor to a safe state.

In 1995, the international community selected test 6.9c as the basis for theOECD/

NEA international benchmark exercise on simulation software, called "International Standard Problem (ISP) 38". This involved the study of a loss-of-coolant scenario during an outage, when pressurizer and the steam generator outlet plenum manways are open for maintenance work. ISP 38 brought together research organizations and licensees from 18 countries and compared the results of five different software codes with experimental results. It revealed that the main physical phenomena were effectively reproduced by the codes considered, although variations were observed due to faults in modeling multidimensional effects, particularly in the low-pressure range.

Similar facilities were built overseas in the 1970s and the early 1980s to study the thermal-hydraulic behavior of various nuclear reactor types in accident situations. The most important of these are listed below:

– Loss-of-Fluid Test, or LOFT, United States. A scale model of a 1000 MWe reactor (volume and power 1/50, height 1/2), comprising two loops, the only facility of its type with a nuclear core for heating water;

– Primärkreislauf, or PKL, Germany. A scale model of a 1300 MWe KWU reactor (volume and power 1/145, height full scale), comprising four loops and 314 electrically heated rods;

– Large Scale Test Facility, or LSTF, Japan. A scale model of a 1100 MWe reactor (volume and power 1/48, height full scale), comprising two loops;

– LOBI, European CommunityJoint Research Center, Ispra, Italy. A scale model of a 1300 MWe KWU-Siemens reactor (volume and power 1/700, height full scale). It comprised two loops and 64 electrically heated rods;

– PSB-VVER13(Russia). A scale model of a 1000 MWe VVER-type reactor (volume and power 1/300, height full scale) comprising four loops;

– Parallel Channel Test Loop, or PACTEL, Finland. A scale model of a 400 MWe VVER-type reactor (volume and power 1/305, height full scale), comprising three loops and 144 electrically heated rods.

International agreements, generally negotiated in connection with research work supported by OECD/NEA, gaveCATHARE development teams access to many exper-imental results obtained during various research programs carried out at these facilities.

13. Vodo-Vodianoï Energuetitcheski Reaktor.

Research on Loss-of-coolant Accidents 19

In all, nearly 300 tests were analyzed and used to qualify the simulation code on a broad spectrum of cooling accidents.

Most of these facilities were shut down and decommissioned. Some uncertainties remained, however. In some configurations, for example, water slugs containing no boric acid can form as a result of steam condensing in the steam generator tubes during a LOCA, for example. They can then be transported to the core where they are liable to initiate a criticality accident. In addition, performance checks had to be carried out on the passive cooling systems, based on natural convection, that were contemplated for use in some Generation III reactors. The studies aimed at eliminating these uncertainties called for the use of Computational Fluid Dynamics (CFD) codes, which are multidimensional codes that provide very detailedflow descriptions. It was, of course, important to check the predictive capabilities of these advanced tools through large-scale tests using systems that were as realistic as possible. For this purpose, the OECD/NEA backed new research projects in the PKL and LSTF facilities, in most cases with French involvement. These included:

– the PKL (2004–2007), PKL-2 (2007–2011) and PKL-3 (2012–2015) projects aimed at studying phenomena such as boric acid dilution in various situations, and natural convection in the event of loss of cooling during an outage, or to reproduce beyond-design-basis situations involving delayed safety injections for assessing safety margins;

– the ROSA14(2005–2009) and ROSA-2 (2009–2012) projects for studying thermal stratification and natural convection phenomena, and testing new cooling proce-dures in accident situations.

Tools simulating the behavior of the nuclear steam supply system were needed in order to train teams likely to be required in the event of nuclear emergency. The SIPA simulator was developed by IPSN in the 1990s usingCATHAREmodules. It has since been replaced by the SOFIA15simulator, which was jointly developed byAREVAandIRSN. It is used by IRSN in particular to produce accident scenarios used during national emergency exercises.

Dans le document EDP Open (Page 50-53)