🔢 mathematics

Viscosity in error upper bound for a consistent splitting scheme of the Navier-Stokes equations

This paper demonstrates that the consistent splitting scheme for the Navier-Stokes equations proposed by Huang and Shen lacks robustness at low viscosities (high Reynolds numbers) because its error bound contains negative powers of viscosity, a finding supported by refined theoretical analysis and numerical experiments that identify the explicit treatment of the convection term as the source of instability.

Original authors: M Nader Alhomsi, Jiahong Wu, Xiaoming Zheng

Published 2026-06-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: M Nader Alhomsi, Jiahong Wu, Xiaoming Zheng

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to simulate how water flows through a pipe on a computer. The math behind this is incredibly complex, involving equations that describe how speed (velocity) and pressure interact. To solve these equations, scientists use "schemes"—basically, a set of step-by-step instructions the computer follows to guess the answer.

This paper investigates a specific set of instructions proposed by Huang and Shen. The authors of this new paper, Nader Alhomsi, Jiahong Wu, and Xiaoming Zheng, wanted to see if these instructions work well when the fluid is very slippery (low viscosity) or very thick (high viscosity).

Here is a breakdown of their findings using simple analogies:

1. The "Slippery Floor" Problem (Viscosity)

In fluid dynamics, viscosity is like the "thickness" or "stickiness" of a fluid. Honey has high viscosity; water has low viscosity.

The Original Study: The original creators of the scheme tested their instructions assuming the fluid was "sticky" (viscosity = 1). Under these conditions, the scheme worked perfectly.
The New Discovery: The authors of this paper asked, "What happens if we make the fluid super slippery, like water or air?" They kept the viscosity as a variable (a symbol) in their math instead of fixing it at 1.
The Result: They found that as the fluid gets more slippery (viscosity approaches zero), the "error" in the computer's guess explodes. It's like trying to walk on a floor that is becoming increasingly icy; the original instructions don't have enough grip to keep you from falling. The math shows that the error bound contains "negative powers of viscosity," which is a fancy way of saying: as the fluid gets thinner, the error gets infinitely huge.

2. The "Splitting" Strategy

The scheme they are testing is called a "splitting scheme." Imagine you are trying to navigate a maze.

The Strategy: Instead of looking at the whole maze at once, the scheme breaks the problem into two parts:
1. Diffusion: How the fluid spreads out (like ink in water).
2. Convection: How the fluid moves forward (like a river current).
The Flaw: The scheme treats the "spreading" part carefully (implicitly) but treats the "moving" part (convection) loosely (explicitly).
The Analogy: Think of driving a car. The "spreading" part is like braking and steering, which the scheme handles well. The "moving" part is like accelerating. The scheme tries to guess how fast you will go based on your current speed without checking the road ahead. At low speeds (high viscosity), this guess is fine. But at high speeds (low viscosity/high Reynolds number), guessing your acceleration without checking the road leads to a crash.

3. The "Triple-Exponential" Warning

The authors did a deep dive into the math to see exactly how bad the error gets. They found a terrifying pattern.

The error doesn't just get bigger; it grows at a rate that is triple-exponential.
The Metaphor: Imagine a single bacterium that splits into two every minute. That's exponential. Now imagine a bacterium that splits into two, and each of those splits into two, and each of those splits into two, all happening simultaneously.
The authors found that to keep the computer simulation from crashing as the fluid gets slippery, you would need to take time steps so incredibly small that they are practically zero. It's like trying to take a photo of a hummingbird's wing with a camera that takes one picture every trillion years. The scheme becomes useless for fast-moving, slippery fluids.

4. The "Blow-Up" Experiment

To prove their math wasn't just theory, they ran a computer experiment using a "perturbed Kovasznay flow" (a specific, well-known test case for fluid flow).

The Test: They ran the simulation with different levels of "stickiness" (Reynolds numbers).
The Outcome:
- The "Robust" Solver: They compared their scheme against a "fully implicit" solver (a very careful, slow, but safe method). This solver stayed smooth and stable, even at high speeds.
- The "GSAV" Scheme: When they used the scheme in question, it worked fine at low speeds. But as soon as they increased the speed (lowered viscosity), the numbers in the computer went wild. The velocity and pressure values shot up to infinity (blow-up), and the simulation crashed.
The Conclusion: By removing the "moving" part of the equation (convection) and just testing the "spreading" part, the scheme worked fine. This proved that the explicit treatment of the convection term (the loose guessing of acceleration) is the specific culprit causing the crash.

Summary

This paper is a "stress test" for a popular computer algorithm used to simulate fluids.

The Verdict: The algorithm is not robust. It works great for thick, slow fluids, but it fails catastrophically for thin, fast fluids.
The Reason: It tries to guess the future speed of the fluid too loosely. When the fluid is slippery, that guess is wrong, and the error compounds until the simulation explodes.
The Fix Needed: To make this scheme work for high-speed flows, the "guessing" part needs to be tightened up, or the whole method needs to be changed to handle the "moving" part more carefully.

The authors did not suggest new medical applications or future engineering projects; they simply identified a mathematical flaw in a specific tool and proved why it breaks under certain conditions.

Technical Summary: Viscosity in Error Upper Bound for a Consistent Splitting Scheme of the Navier–Stokes Equations

Problem Statement
This paper investigates the robustness of the second-order consistent splitting scheme for the Navier–Stokes equations proposed by Huang and Shen [5]. While the original analysis in [5] established unconditional stability and error bounds, it relied on the assumption that the viscosity $\nu$ is fixed to unity. The central problem addressed here is whether the scheme remains robust as $\nu \to 0$ (high Reynolds number). Following the definition in [2], a scheme is considered robust if the error upper bound for the velocity contains no negative powers of the viscosity. The authors aim to determine if the Huang–Shen scheme satisfies this criterion by re-analyzing the error bounds while retaining $\nu$ as a symbolic parameter throughout the derivation.

Methodology
The authors extend the proof methodology of Huang and Shen [5] by systematically tracking the dependence of all constants on the viscosity $\nu$ . Key methodological steps include:

Refinement of Stokes Pressure Estimates: A critical component of the error analysis involves the Stokes pressure estimate. The authors refine a theorem from [8] (cited as Lemma 4 in [5]) to explicitly quantify the constant $C_S(\varepsilon)$ in the inequality $\|\nabla p_s(u)\|^2 \leq (\frac{1}{2} + \varepsilon)\|\Delta u\|^2 + C_S(\varepsilon)\|\nabla u\|^2$ . They prove that as $\varepsilon \to 0^+$ , $C_S(\varepsilon)$ behaves as $O(\varepsilon^{-3})$ , a dependency crucial for the subsequent error analysis.
Generalized Stability Analysis: The stability analysis for the time-dependent Stokes equations is extended to arbitrary $\nu > 0$ . This involves re-deriving the energy estimates for the scheme (3a)–(3e) and determining the minimum integer value of the stabilization parameter $k$ required for unconditional stability.
Generalized Error Analysis: The error analysis for the full Navier–Stokes equations is generalized to arbitrary $\nu$ . This requires careful handling of the nonlinear convection term, which is treated explicitly in the scheme. The authors employ discrete Gronwall inequalities and track the accumulation of constants dependent on $\nu$ through three nested applications of the Gronwall lemma.
Numerical Verification: To corroborate the analytical findings, the authors conduct numerical experiments on a perturbed Kovasznay flow. They compare the GSAV-Spectral scheme against a fully implicit FEM-Newton solver (treated as a robust reference) and the time-dependent Stokes counterpart of the same scheme.

Key Contributions and Results

Reduction of Parameter $k$ : The authors demonstrate that the stability and convergence requirements for the parameter $k$ can be reduced from $k \geq 5$ (as stated in [5]) to $k \geq 4$ . This reduction improves the truncation error of the scheme, as verified numerically in Example 1.
Non-Robustness of the Scheme: The primary theoretical result is that the error upper bound for the velocity in the $H^1$ $H^{1}$ norm depends on negative powers of the viscosity. Specifically, the constant $C_{5.3}$ $C_{5.3}$ in the error bound $\|\nabla e^{n+1}\| \leq C_{5.3} \delta t^2$ $∥\nabla e^{n + 1} ∥ \leq C_{5.3} δ t^{2}$ exhibits a triple-exponential growth with respect to $\nu^{-1}$ $ν^{- 1}$ as $\nu \to 0^+$ $ν \to 0^{+}$ .
- The constant $C_1$ (related to velocity stability) scales as $O(\nu^{-3} \exp(\nu^{-4}))$ .
- The constant $C_2$ (related to velocity error) scales as $O(\exp(\exp(\nu^{-4})))$ .
- The final constant $C_5$ (related to the GSAV consistency) scales as $O(\exp(\exp(\exp(\nu^{-4}))))$ .
  Consequently, the time-step constraint $\delta t$ required to guarantee convergence must shrink at least as fast as $O(\exp(-\exp(\exp(\nu^{-4}))))$ , rendering the scheme non-robust for small viscosities.
Localization of Failure: The numerical experiments confirm that the scheme "blows up" (intermediate velocities and pressures diverge) at high Reynolds numbers ( $Re \geq 2 \times 10^4$ ). Crucially, when the convection term is removed (Stokes problem), the scheme remains stable for all tested Reynolds numbers. Similarly, a fully implicit Newton solver does not exhibit blow-up. This comparison localizes the source of non-robustness to the explicit treatment of the convection term within the splitting scheme.

Significance and Claims
The paper claims to provide the first rigorous stability and error analysis for a second-order consistent splitting scheme that explicitly tracks viscosity dependence. The authors conclude that while the Huang–Shen scheme offers unconditional stability in a weak sense and improved accuracy with $k=4$ , it is fundamentally non-robust for the Navier–Stokes equations at high Reynolds numbers. The failure is intrinsic to the explicit handling of the convection term, which introduces error bounds that explode as viscosity decreases. The work highlights the necessity of implicit treatment for convection terms or alternative formulations if robustness at high Reynolds numbers is required. The authors do not propose a new scheme to fix this issue but rather delineate the limitations of the existing consistent splitting approach.

1. The "Slippery Floor" Problem (Viscosity)

2. The "Splitting" Strategy

3. The "Triple-Exponential" Warning

4. The "Blow-Up" Experiment

Summary

More like this