Best Practice Guidelines for the use of CFD in Nuclear Reactor Safety Applications

Abstract

In May 2002, an 'Exploratory Meeting of Experts to Define an Action Plan on the Application of Computational Fluid Dynamics (CFD) Codes to Nuclear Reactor Safety Problems' was held at Aix-en-Provence, France. One of three recommended actions was the formation of this writing group to report on the need for guidelines for use of CFD in single phase Nuclear Reactor Safety (NRS) applications. CSNI approved this writing group at the end of 2002, and work began in March 2003. A final report was submitted to GAMA in September 2004, summarizing existing Best Practice Guidelines (BPG) for CFD, and recommending creation of a BPG document for Nuclear Reactor Safety (NRS) applications. The present document is intended to provide an internally complete set of guidelines for a range of single phase applications of CFD to NRS problems. However, it is not meant to be comprehensive; it is recognized that for any specific application a higher level of specificity is possible on questions of nodalization, model selection, and validation. This document should provide direct guidance on the key considerations in known single phase applications, and general directions for resolving remaining details. The intent is that it will serve as a template for further application specific (e.g. PTS, induced break) BPG documents that will provide much more detailed information and examples. The document begins with a summary of NRS related CFD analysis in countries represented by the authors. Chapter 3 deals with definition of the problem and its solution approach. This includes isolation of the portion of the NRS problem most in need of CFD, and use of a classic thermal hydraulic (TH) safety code to provide boundary conditions for the CFD based upon less detailed simulation of the balance of plant. Chapter 4 provides guidance in choosing between various options, and also discusses use of a transient calculation with tightly coupled CFD and TH codes. Chapter 5 discusses selection of physical models available as user options. As is appropriate for single phase CFD, most of the emphasis is on selection of turbulence models. Recommendations are provided for high level selection between Reynolds Averaged Navier Stokes (RANS), Large Eddy Simulation (LES), and hybrid approaches such as Detached Eddy Simulation (DES). Chapter 7 focuses on the numerical approximations available to solve the flow equations. Guidelines are provided for nodalization, and for choice of discrete approximations to the differential equations. Guidance is also given on convergence of iterative solutions, and numerical techniques for following free surfaces. Chapter 7 discusses general assessment strategy. Chapter 8 covers approaches to limiting errors associated with discretization and numerical solution methods (verification). This step is a necessary precursor to quantifying errors associated with physical models (validation) as described in Chapter 9. Guidance on documentation is provided in Chapter 10. Chapter 11 provides some examples of NRS applications; the first two examples are boron dilution and pressurized thermal shock; the third example explores the use of Fluent for simulation of dry cask storage of spent fuel (this example is highly suited to single phase CFD analysis)