FZR has 10 years of experience using ANSYS CFX software for nuclear reactor applications and, more recently, chemical process simulations. These CFD activities are very closely connected to experimental investigations at facilities in the institute. Using advanced two-phase flow measurement techniques, a comprehensive database for CFD code verification is being created. Thus, in addition to single-phase flow nuclear reactor engineering applications, CFD can be used for development and verification of physical models for two-phase flows. FZR is participating in the German initiative on CFD applications in nuclear reactor safety research in close cooperation with the Gesellschaft für Anlagen- und Reaktorsicherheit (GRS, the leading German nuclear safety expert organization) and ANSYS, Inc. The institute has set aside a significant portion of FZR’s 100-processor Linux cluster for its group of young and highly motivated CFD scientists.
FZR coordinated the experimental and analytical investigations of turbulent mixing inside pressurized water reactors (PWRs) performed within the EC project FLOMIX-R (fluid mixing and flow distribution in the reactor circuit). The purpose of this project was to describe coolant mixing phenomena, particularly for severe accident scenarios. Such scenarios include steam line breaks and boron dilution, in which mixing of coolant from different loops moderates the inflow of water with insufficient boron content or temperature into the reactor core. These changes can lead to reactor power excursions (rapid increase in reactor power) due to positive reactivity effects. An example of a typical boron dilution scenario occurs during start-up of the first main coolant pump after a slug of low borated water has formed in one of the cooling loops, in which mixing is largely forced by the momentum introduced by the pump starting.
Another safety issue arises during emergency core cooling (ECC) situations, when cold water is injected into a hot cooling loop. In this case, buoyancy-driven mixing is influenced by density differences in the fluid and is typical for so-called pressurized thermal shock (PTS) scenarios. When a streak of cold ECC water touches the reactor pressure vessel (RPV) wall, unacceptable thermal stresses can occur.
Measurement data from several sets of mixing experiments, using advanced measurement techniques with enhanced temporal and spatial resolution, improved the basic understanding of turbulent mixing and provided data for CFD code verification.
Selected experiments were then simulated using ANSYS CFX and applying best practice guidelines (BPGs), a set of systematic procedures for quantifying and reducing numerical errors. BPGs were applied when considering computational grid resolution and time step, turbulence models, internal geometry modeling, boundary conditions, numerical schemes and convergence criteria. These investigations highlighted the importance of grid quality and the need to minimize numerical diffusion by using second-order discretization. In fact, first-order schemes sometimes were found to provide physically incorrect results.
ANSYS CFX was well able to predict the experimental flow patterns and mixing phenomena for both buoyancy-driven and momentum-driven flows. Two-equation turbulence models, like k-ω or SST, were found to be suitable for momentum-driven slug mixing, while Reynolds stress models provided better results for buoyancy-driven mixing.
Comprehensive multiphase flow models, advanced turbulence models, second order discretization and scalable parallel performance all combine to make ANSYS CFX a valuable tool at FZR. ANSYS CFX software has been instrumental in the development and verification of best practice guidelines for the use of CFD in nuclear safety analysis.
Acrylic model of the reactor pressure vessel.
ANSYS CFX results for the start-up of the first pump after 9s. The mixing scalar represents the dimensionless boron concentration or temperature for a pump start-up scenario.
Streamlines representing the fluid flow 23s after ECC injection took place, illustrating the flow pattern in a buoyancy-driven mixing case.
Plateau-averaged mixing scalar in the downcomer of the pressure vessel at azimuthal positions, in which all loops are in operation with the nominal flow rate (185 m^3/h per loop) and tracer is injected into one loop. The result is a sector formation of the tracer in the downcomer in the corresponding quadrant.
Time-dependent tracer distribution of the buoyancy-driven mixing experiment and calculation at an indicated local position of the cold leg.