Efficiency and Modularity of ALFVN (Adaptive Lightweight Finite Volume Numerics) Hydrodynamics Code
POSTER
Abstract
ALFVN is a versatile 3D hydrodynamics Python code which solves conservative equations of hydrodynamics using finite volume methods. It's written to demonstrate well-known hydrodynamical features such as turbulence and instabilities. We test ALFVN with standard test problems like the Linear Waves Test and the Sod Shock Tube. ALFVN is modular by design. The reconstruction step, Riemann solver, and the numerical integration method which dictates the overarching structure can all be swapped out individually while keeping the rest of the code intact. We hope to explore and compare various options available for each of these components. For reconstruction, we will experiment with donor cell, piecewise linear, and piecewise parabolic interpolations. For numerical integration, we will compare forward Euler, van Leer 2, and Runge-Kutta 2. We will also test LLF (Local Lax-Friedrich) and HLLE (Harten-Lax-van Leer-Einfeldt) methods for the Riemann solver. In each of these combinations, we supply ALFVN with the same initial condition and discuss the output in terms of their accuracy and efficiency. Since ALFVN is written to be easily run on a home computer for demonstration purposes, it benefits to know the limitations and runtime of various options in each component of a hydrodynamics code.
Presenters
-
Rachel Wang
University of California, Berkeley
Authors
-
Rachel Wang
University of California, Berkeley
-
Siddhant Solanki
University of California, Santa Barbara