Implementation of Viscous Effects in RELAP5-3D Thermal Hydraulics Code

Author:  Raymond C. Wang
Institution:  University of California, Berkeley

The Reactor Excursion Leak Analysis Program (RELAP5-3D) is a world class thermal hydraulics safety analysis code developed at the Idaho National Laboratory (INL) to address safety concerns in light water nuclear reactors. The purpose of this project is to implement viscous effects into the current RELAP5-3D code. Viscous effects have been thoroughly studied and implemented in many Computational Fluid Dynamics (CFD) codes. However, because of the original purpose of RELAP5-3D, viscous effects on fluid dynamics were not implemented during initial code development. As demand for coupling RELAP5-3D with CFD codes increases, implementing viscous effects in RELAP5-3D is becoming more important both for calculation accuracy and code coupling stability. The momentum flux equations used in RELAP5 resemble the Navier-Stokes Equations (NSE) but doesn't include the viscous contributions. Therefore, a double central finite difference method is used to discretize second order differentials of the viscous terms in both Cartesian and cylindrical geometries. A total of 20 new terms were introduced and the original RELAP5 wall boundary condition was changed from free slip to no slip. The original code architecture was not changed during any of the coding to avoid introducing new limitations. Two test models were used to analyze the performance using water as working fluid. The 3D cylindrical pipe model was used to test if a Poiseuille velocity profile could be generated for laminar flow regimes. The rectangular duct model was used to test for functionality of the viscous terms in Cartesian geometry. A comparison between the calculations from the modified and unmodified RELAP5-3D for the cylindrical pipe was made. The modified calculations converged in approximately 300 seconds to a classic parabolic Poiseuille flow profile with an error from analytical results of 2.28%, whereas the unmodified calculations converged in 600 seconds to a profile with plug flow characteristics. These results demonstrate a significant improvement in both convergence time and accuracy. Thus, implementation of viscous effects in RELAP5-3D is not only achievable, but also beneficial to convergence speed. Further development of the viscous effect can expand capabilities to include modeling turbulent viscous effects, two phase effects, or effects using nearly implicit advancements.

The Journal of Young Investigators is not affiliated with the US Department of Energy. This paper was written by a student intern with the Department of Energy and does not constitute a declarative position of either the Department of Energy or the Journal of Young Investigators.