Core Concepts
A simple and efficient convex optimization-based post-processing procedure can enforce bounds, conservation, and high-order accuracy for numerical schemes solving the Cahn-Hilliard-Navier-Stokes system.
Abstract
The content discusses a simple and efficient approach to enforce the bound-preserving property of high-order accurate numerical schemes for phase field models, without destroying conservation and accuracy.
Key highlights:
Many numerical methods, especially high-order accurate schemes, do not preserve bounds, which is critical for physical meaningfulness and robustness of numerical computation.
The authors propose a convex optimization-based post-processing procedure to enforce bounds, conservation, and high-order accuracy for numerical solutions of the Cahn-Hilliard-Navier-Stokes (CHNS) system.
The optimization problem is formulated as a nonsmooth convex minimization, which can be efficiently solved using the generalized Douglas-Rachford splitting method with optimal algorithm parameters.
The authors analyze the asymptotic linear convergence rate of the Douglas-Rachford splitting and provide a simple formula to choose nearly optimal algorithm parameters based on the number of out-of-bounds cells.
Numerical tests on a 3D CHNS system demonstrate that the proposed limiter is high-order accurate, very efficient, and well-suited for large-scale simulations, taking at most 20 iterations to converge.