Analysis of a Stabilized CNLF Method with Fast Slow Wave Splittings for Flow Problems

Nan Jiang, Hoang TranComput. Methods Appl. Math 2015 · DOI

TL;DR

This paper introduces a new, unconditionally stable numerical method (STAFASL) for simulating fluid flows, particularly those with Coriolis forces like those found in geophysical applications. By adding a stabilization term, the method overcomes the restrictive time step limitations of existing techniques, offering a more robust and efficient way to model complex fluid dynamics.

What Problem Does This Paper Address?

Simulating fluid flows, especially those governed by the Navier-Stokes equations with additional forces like the Coriolis effect (common in atmospheric and oceanic modeling), presents significant computational challenges. Traditional numerical methods, like the Crank-Nicolson Leap-Frog (CNLF) scheme, often suffer from restrictive time step conditions (CFL conditions) that limit their practical application. These limitations arise from the need to accurately capture both fast and slow-moving wave phenomena. Furthermore, existing methods can be unstable in certain "unstable modes" of the flow, potentially leading to unphysical results or divergence. This paper aims to develop and analyze a stabilized version of a fast-slow wave splitting method that addresses both the time step restriction and the stability issues.

Methods

The authors propose a Stabilized Crank-Nicolson Leap-Frog (STAFASL) method. This method builds upon a Fast-Slow Wave Splitting (FASL) approach, which decomposes the fluid flow problem into components representing fast and slow waves. The core innovation is the addition of a stabilization term to the discrete equations. This term, specifically $2\Delta t \Lambda_h(u^{n+1} - u^{n-1}), \Lambda_h(v^n)$, is designed to eliminate the CFL time step restriction that plagues the standard FASL method. The paper provides a rigorous mathematical analysis of the stability and convergence of both the FASL and the proposed STAFASL methods, employing energy methods and discrete Gronwall inequalities. They also analyze the behavior of the "unstable mode" of the flow, demonstrating that their stabilized method controls it.

Datasets

Not discussed in the paper. The paper focuses on theoretical analysis and numerical experiments with exact solutions and a benchmark problem.

Key Insights and Ideas

  • Unconditional Stability is Achievable: The primary insight is that by carefully adding a stabilization term, it's possible to create a numerical scheme that is stable regardless of the time step size, a significant improvement over traditional methods.
  • Controlling the "Unstable Mode": A key challenge in fluid simulations is the behavior of certain flow modes that are not naturally damped by standard schemes. The authors show that their stabilization technique can control these modes, preventing unphysical growth.
  • Fast-Slow Splitting Benefits: The underlying fast-slow wave splitting is crucial for separating different time scales in the flow, allowing for more targeted numerical treatment.
  • Coriolis Force Impact: The presence of the Coriolis force, while a simplification of geophysical flows, introduces specific stability challenges that the proposed method is designed to handle.

Key Results

  • The proposed STAFASL method is proven to be unconditionally stable for the Oseen equation with a Coriolis term, meaning it does not require a restrictive CFL time step condition.
  • For the Oseen equation with a Coriolis term, the unstable mode of the standard CNLF method is shown to be asymptotically stable under the FASL time step restriction.
  • Numerical experiments confirm the theoretical findings, demonstrating that STAFASL successfully eliminates time step restrictions and provides stable simulations, unlike the standard FASL method which exhibits instability at larger time steps.
  • Convergence rates for STAFASL are shown to be second-order, comparable to FASL, but with superior stability.

Limitations and Future Work

  • Full NSE Unstable Mode: The paper notes that while they prove control of the unstable mode for the linearized Oseen problem, controlling the unstable mode for the full Navier-Stokes equation with rotation is still an open question.
  • Approximation of Initial Steps: STAFASL, like CNLF, is a three-level method, requiring approximations for the first two time steps. The accuracy of these initial steps can affect the overall convergence rate.
  • Computational Cost: While STAFASL removes the time step restriction, the added stabilization term might increase the computational cost per time step. The paper doesn't extensively detail this trade-off.
  • Future Work: The authors suggest that their work is a step towards more complex applications and that many open questions remain, particularly regarding the full NSE.

Summary updated

BibTeX
@article{jiang2015analysis,
  author = {Nan Jiang and Hoang Tran},
  title = {{Analysis of a Stabilized CNLF Method with Fast Slow Wave Splittings for Flow Problems}},
  journal = {Comput. Methods Appl. Math.},
  year = {2015},
  doi = {10.1515/cmam-2015-0010},
}
Figure