On shifting the thermal explosion threshold by a vortical flow in dimension two

Technical Summary: On shifting the thermal explosion threshold by a vortical flow in dimension two

Problem Statement
This paper investigates a natural generalization of the classical Frank-Kamenetskii model for thermal explosion, specifically within a two-dimensional bounded domain $\Omega \subset \mathbb{R}^2$. The study focuses on the stationary temperature distribution $u$ in a combustion vessel with a cold boundary, governed by the semi-linear elliptic equation with an advection term:
$$\begin{cases} -\Delta u + A \mathbf{v} \cdot \nabla u = \lambda f(u) & \text{in } \Omega, \\ u = 0 & \text{on } \partial\Omega, \end{cases}$$
where $\lambda > 0$ is the Frank-Kamenetskii parameter (measuring reaction strength), $A \in \mathbb{R}$ is the amplitude of the flow, and $\mathbf{v}$ is a prescribed incompressible vortical flow satisfying $\mathbf{v} = \nabla^\perp \psi$ and the no-penetration condition $\mathbf{v} \cdot \nu = 0$. The reaction term $f$ is assumed to be a positive, convex, increasing $C^2$ function with sufficiently fast growth at infinity.

The central physical phenomenon is the existence of a critical threshold $\lambda^*(A)$. For $\lambda < \lambda^*(A)$, the system admits a minimal classical solution representing a stable thermal equilibrium. For $\lambda > \lambda^*(A)$, no classical solution exists, corresponding to a thermal explosion (thermal runaway). The paper addresses the question of whether a small-amplitude vortical flow can adjust this explosion threshold. Specifically, it asks if one can construct a flow such that $\lambda^*(A)$ changes linearly with $A$ (i.e., $\lambda^*(A) = \lambda^*(0) + \theta A + o(A)$ with $\theta \neq 0$) by reversing the flow direction, provided the domain is not a disk.

Methodology
The authors employ a combination of qualitative analysis of elliptic boundary value problems, perturbation theory, and spectral analysis. The methodology proceeds through several key stages:

1. Existence and Regularity of Extremal Solutions: The authors first establish that for the given hypotheses, the extremal solution $u^*$ (the pointwise limit of minimal solutions as $\lambda \nearrow \lambda^*(A)$) is a classical solution ($C^{2,\alpha}$). This extends known results for the case $A=0$ to the case with advection. The proof relies on constructing integral inequalities mimicking the semi-stability condition, utilizing the principal eigenfunction of the linearized operator, and applying specific estimates on the reaction term $f$.
2. Uniqueness and Degeneracy: It is proven that $u^*$ is the unique classical solution at $\lambda = \lambda^*(A)$ and that the linearization of the problem about $u^*$ is degenerate (the principal eigenvalue is zero).
3. Differentiability of the Threshold: Using the implicit function theorem and Fredholm alternative arguments, the authors demonstrate that the critical parameter $\lambda^*(A)$ is a differentiable function of the flow amplitude $A$. An explicit formula for the derivative $\dot{\lambda}^*(A)$ is derived involving the extremal solution $u^*$, the adjoint eigenfunction $\tilde{\phi}_1$, and the flow field.
4. Geometric Analysis of the Derivative: The core of the main result involves analyzing the derivative at $A=0$. The expression for $\dot{\lambda}^*(0)$ is shown to depend on the integral of the product of the stream function $\psi$ and the Jacobian of the extremal solution $u$ and the principal eigenfunction $\phi_1$ of the unperturbed problem ($A=0$).
5. Symmetry Breaking: The authors prove that if the Jacobian $\nabla u \cdot \nabla^\perp \phi_1$ vanishes identically, the domain $\Omega$ must be a disk (via an overdetermined boundary value problem argument). Consequently, if $\Omega$ is not a disk, there exists a region where this Jacobian is non-zero, allowing the construction of a stream function $\psi$ that yields a non-zero derivative.

Key Contributions and Results

• Regularity of Extremal Solutions: The paper establishes that under the assumption of sufficiently fast growth of the reaction term, the extremal solution for the Gelfand problem with vortical flow in 2D is always classical ($C^{2,\alpha}$). This confirms that the "weak" solutions often discussed in higher dimensions do not arise in this 2D setting with these specific flow conditions.
• Differentiability of the Explosion Threshold: The authors prove that the explosion threshold $\lambda^*(A)$ is differentiable with respect to the flow amplitude $A$. They provide the explicit formula:
  $$\dot{\lambda}^*(A) = \frac{\int_\Omega (\mathbf{v} \cdot \nabla u^*) \tilde{\phi}_1 \, dx}{\int_\Omega f(u^*) \tilde{\phi}_1 \, dx}$$
  where $\tilde{\phi}_1$ is the positive eigenfunction of the adjoint linearized operator.
• Main Theorem (Threshold Shifting): The primary result (Theorem 1.1) states that if the domain $\Omega$ is not a disk, there exists a regular incompressible vortical flow such that the explosion threshold shifts linearly with the flow amplitude. Specifically, $\lambda^*(A) = \lambda^*(0) + \theta A + o(A)$ with $\theta \neq 0$. This implies that by reversing the direction of the flow, one can either increase or decrease the explosion threshold.
• Geometric Constraint: The paper rigorously shows that this ability to shift the threshold via a small flow is impossible if the domain is a disk, as the relevant Jacobian term vanishes identically in that symmetric case.

Significance
The paper claims significance in providing a constructive theoretical framework for controlling thermal explosion limits in combustion vessels through fluid dynamics. It answers a natural question regarding the non-monotonic dependency of the critical parameter on flow amplitude observed in numerical studies. By proving that the threshold can be adjusted by reversing the flow direction (provided the geometry is not a perfect disk), the work offers a mathematical basis for the design of combustion devices where flow manipulation is used to enhance safety (raising the threshold) or control reaction rates. Furthermore, the detailed characterization of the regularity and uniqueness of extremal solutions in the presence of advection contributes to the broader understanding of Gelfand-type problems.

The authors note that the "fast growth" condition on $f$ is primarily used to ensure the regularity of the extremal solution and can be relaxed to include typical nonlinearities like $f(u) = e^u$ or $f(u) = (1+u)^p$ ($p>1$). The results are strictly theoretical, relying on the properties of elliptic operators and spectral theory, without proposing new experimental setups.