On the Existence of Boundary Layer Separation for Incompressible Fluid Flow in the Half-Space

Technical Summary: On the Existence of Boundary Layer Separation for Incompressible Fluid Flow in the Half-Space

Problem Statement
This paper investigates the phenomenon of boundary layer separation for incompressible fluid flow in the half-space $\mathbb{R}^n_+$. Specifically, the authors analyze the Stokes system (the linearized Navier–Stokes equations) and the full Navier–Stokes system under localized, non-zero boundary data with zero initial conditions. The central focus is the sign change of the velocity field near the boundary, which implies the vanishing of the normal derivative of the tangential velocity and the subsequent formation of a boundary layer separation point. While previous works have established the existence of singular solutions with unbounded velocity gradients or unbounded velocities under no-slip conditions, this study addresses the specific dynamics of separation points under sufficiently regular, localized boundary data where normal derivatives remain bounded.

Methodology
The analysis relies on the explicit representation formula for solutions to the Stokes system in the half-space with non-zero boundary data. The authors decompose the solution into five distinct terms based on the Poisson kernel and related operators (Gaussian, Newtonian, and Riesz transforms).

1. Decomposition and Normal Derivative Analysis: By utilizing the solution representation and specific identities for the kernels, the authors derive a formula for the normal derivative of the tangential velocity components ($D_{x_n}w_i$) on the boundary. This formula reduces to the sum of two primary terms: one involving a spatial convolution with the boundary data and another involving a singular time-integral quantity, denoted as $M(t)$.
2. The Integral Quantity $M(t)$: The core of the analysis centers on the integral:
   $$M(t) := \int_{-\infty}^t \frac{\phi(t) - \phi(s)}{(t-s)^{3/2}} ds$$
   where $\phi(s)$ represents the temporal profile of the boundary data. The authors demonstrate that the sign of $M(t)$ is the decisive factor in determining the sign of the normal derivative $D_{x_n}w_i$ on the boundary.
3. Pressure Gradient Analysis: Using the representation formula for pressure, the authors analyze the tangential derivatives of the pressure ($D_{x_i}p$) to establish the development of adverse pressure gradients, a physical precursor to separation.
4. Perturbation Argument for Navier–Stokes: To extend results to the nonlinear Navier–Stokes equations, the authors employ a perturbation argument. They construct solutions as a sum of the Stokes solution and a perturbation term, utilizing Hölder continuity estimates and iterative schemes to prove the existence of solutions that retain the qualitative separation behavior of the linear case for sufficiently small boundary data amplitudes.

Key Contributions and Results

• Existence and Non-Existence Criteria: The paper establishes a rigorous criterion for the existence of boundary layer separation points based on the sign of $M(t)$.
  • Existence: If $M(t)$ is negative over a specific time interval (Assumption 1.4), a boundary layer separation point exists. The separation point is not stationary; it moves away from the origin and reaches infinity in finite time.
  • Non-Existence: If $M(t)$ is strictly positive (Assumption 1.5), boundary layer separation does not occur; the velocity maintains its direction near the boundary.
• Dynamics of the Separation Point: The authors characterize the motion of the separation point. They prove that for large spatial distances, the time of separation $t^*_i(x')$ is a strictly increasing function of the distance from the origin. Furthermore, as the spatial coordinate tends to infinity, the time of separation converges to a critical time $t^*_0$, defined by the first time $M(t)$ becomes non-positive.
• Uniqueness and Monotonicity: Under the assumption that the temporal boundary data is unimodal (increasing then decreasing), the separation point is unique at each time instance.
• Pressure Gradient Sign Change: The study proves that the tangential pressure gradient undergoes a sign change. Initially negative, it becomes strictly positive at a later time, corresponding to the development of an adverse pressure gradient.
• Navier–Stokes Extension: The paper constructs solutions to the Navier–Stokes equations that exhibit the same qualitative behavior (separation or non-separation) as the Stokes solutions, provided the boundary data amplitude is sufficiently small.

Significance and Claims
The authors claim that this work provides a mathematical foundation for understanding boundary layer separation in the context of the Stokes system, a model often considered too linear to exhibit complex separation phenomena. By identifying a specific singular integral determined by the boundary data as the governing factor, the paper bridges the gap between linear theory and the physical observation of flow reversal.

The significance lies in:

1. Rigorous Definition: Providing a precise mathematical definition of a boundary layer separation point based on the sign change of the velocity near the wall.
2. Linear Mechanism: Demonstrating that even the linear Stokes system can exhibit flow reversal mechanisms connected to boundary layer separation, challenging the notion that such phenomena are exclusively nonlinear.
3. Predictive Criteria: Offering a concrete, computable condition (the sign of $M(t)$) to predict whether separation will occur based solely on the temporal profile of the boundary data.
4. Extension to Nonlinearity: Validating that these linear mechanisms persist in the full Navier–Stokes equations under perturbative conditions, suggesting that the identified integral condition is robust.

The paper concludes that the behavior of the separation point and the associated pressure gradients are intrinsically linked to the nonlocal effects captured by the integral $M(t)$, offering a new perspective on the dynamics of incompressible flow near boundaries.