Navier-Stokes-Fourier Equations

• The Navier-Stokes-Fourier system is a mathematical model for compressible, viscous, and heat-conducting fluids governed by conservation laws.
• It couples mass, momentum, and energy equations with constitutive laws for pressure, stress, and heat flux to ensure thermodynamic consistency.
• Recent advances analyze the stability of strong solutions and extend measure-valued formulations to quantify uncertainty and address blow-up phenomena.

The Navier-Stokes-Fourier (NSF) system describes the evolution of a compressible, viscous, heat-conducting fluid under classical thermomechanical laws. The system couples conservation of mass, linear momentum, and internal energy via constitutive relations for pressure, viscous stress, and heat flux, typically modeled as Newtonian and Fourier laws. The framework is foundational in continuum thermomechanics, underpinning rigorous analysis of stability, statistical solutions, and uncertainty quantification. In particular, recent advances have established stability of strong solutions, robust pushforward representation of measure-valued (statistical) solutions, and analytic tools for extending solution concepts beyond blow-up time in the sense of Feireisl–Lukáčová-Medviďová (Feireisl et al., 2022). The NSF system serves as a canonical model for compressible flows subject to thermal effects, supporting deep connections to stochastic analysis, statistical mechanics, and numerical approximation theories.

1. Mathematical Formulation and Constitutive Laws

The NSF system is posed on a three-dimensional torus $\Omega = T^3$ with periodic boundary conditions. Primary fields are the density $\rho(t, x) > 0$, absolute temperature $\vartheta(t, x) > 0$, and velocity $u(t, x) \in \mathbb{R}^3$. The PDEs governing the system comprise:

• Continuity equation (mass conservation):

$\partial_t \rho + \nabla_x \cdot (\rho u) = 0.$

• Momentum balance (Newton's law):

$$\partial_t (\rho u) + \nabla_x \cdot (\rho u \otimes u) + \nabla_x p(\rho, \vartheta) = \nabla_x \cdot \mathbb{S}(\nabla_x u) + \rho g.$$

• Internal energy balance (first law, with heat conduction):

$$\partial_t (\rho e(\rho, \vartheta)) + \nabla_x \cdot (\rho e(\rho, \vartheta) u) + \nabla_x \cdot q(\nabla_x \vartheta) = \mathbb{S}(\nabla_x u) : \nabla_x u - p(\rho, \vartheta) \nabla_x \cdot u + Q,$$

where $g(t,x)$ is a given volume force, $Q(t, x)$ an external heating, and the closure relations for a Newtonian caloric ideal gas are:

• Equation of state: $p(\rho, \vartheta) = \rho \vartheta$, $e(\rho, \vartheta) = c_v \vartheta$ with $c_v > 1$.
• Viscous stress: $$\mathbb{S}(\nabla_x u) = \mu(\nabla_x u + \nabla_x^T u - \frac{2}{3} (\nabla_x \cdot u) I) + \lambda (\nabla_x \cdot u) I$$ with shear viscosity $\mu > 0$, bulk viscosity $\lambda \ge 0$.
• Heat conduction: $q(\nabla_x \vartheta) = -\kappa \nabla_x \vartheta$ with $\kappa > 0$.

Admissibility of initial data $$D = (\rho_0, \vartheta_0, u_0; g, Q; c_v, \mu, \lambda, \kappa)$$ requires $\inf \rho_0 > 0$, $\inf \vartheta_0 > 0$, $c_v > 1$, $\mu > 0$, $\kappa > 0$ (Feireisl et al., 2022).

2. Well-Posedness and Strong Solution Theory

Let $3 < q \le 6$ and trajectory space $X_T$ built from regularity classes of initial data and time-evolved fields. A triple $(\rho, \vartheta, u)$ is a strong solution on $[0, T]$ if:

• $\rho > 0$, $\rho \in C([0, T]; W^{1, q}(\Omega))$, $\partial_t \rho \in C([0, T]; L^q(\Omega))$.
• $$u \in C([0, T]; W^{2, 2}(\Omega; \mathbb{R}^3)) \cap L^2(0, T; W^{2, q}(\Omega; \mathbb{R}^3))$$, $$\partial_t u \in L^2(0, T; L^2(\Omega; \mathbb{R}^3))$$.
• $\vartheta > 0$, $$\vartheta \in C([0, T]; W^{2, 2}(\Omega)) \cap L^2(0, T; W^{1, 2}(\Omega))$$, $\partial_t \vartheta \in L^2(0, T; L^2(\Omega))$.

The classical result (cf. Cho–Kim) assures existence of a unique strong solution up to a maximal time $T_{\max}(D)$, with blow-up detected by divergence of relevant Sobolev norms:

$$\lim_{t \to T_{\max}^-} \left( \|\rho(t)\|_{W^{1, q}} + \|u(t)\|_{W^{2, 2}} + \|\vartheta(t)\|_{W^{2, 2}} + \left\|\frac{1}{\rho(t)}\right\|_{C} \right) = \infty.$$

(Feireisl et al., 2022)

3. Stability of Solutions and Data Dependence

A central result demonstrates stability of strong solutions with respect to data perturbations. For sequences $D_n \to D$ in the product space $\mathcal{D}$:

• $$\liminf_{n \to \infty} T_{\max}(D_n) \ge T_{\max}(D)$$.
• For every $T < T_{\max}(D)$, solution trajectories $$(\rho, u, \vartheta)[D_n] \to (\rho, u, \vartheta)[D]$$ weak-* in $X_T$.

This yields convergence in $C([0, T] \times \Omega; \mathbb{R}^5)$ on any subinterval $[0, T]$ prior to possible singularity formation (Feireisl et al., 2022). The proof exploits uniform energy-type bounds and Sobolev compactness structures.

4. Statistical Solutions and Measure-Valued Framework

Statistical solutions arise by randomization of data $D$, distributed according to a Borel probability measure $V$ on $\mathcal{D}$. Denote

$U_t(D) := (\rho, u, \vartheta)[D](t, \cdot)$

and define the time-$t$ marginal as push-forward $V_t = (U_t)_{\#} V$. The statistical solution is then the family $\{V_t\}_{t\ge0}$ of probability measures on phase space $X_+$, satisfying a Markov semigroup property:

$$M_t: V \mapsto V_t = M_t(V), \quad M_{t+s} = M_t \circ M_s.$$

Crucial properties:

• $V_t$ puts zero mass on blow-up events for $t < T_{\max}$, by lower semicontinuity of $T_{\max}$.
• The semigroup structure persists in the autonomous case (time-independent $g, Q$).
• Disintegration of $V$ with respect to physical parameters yields decomposable family of statistical solutions.

Push-forward construction avoids further selection arguments—Polish metric topology and continuity of solution operator suffice for compactness and tightness (Feireisl et al., 2022).

5. Extension Beyond Blow-Up and the Statistical Measure Approach

The analytic machinery supports extension of individual strong solutions beyond blow-up time by setting $U_t \equiv 0$ for $t \ge T_{\max}(D)$, and augmenting the phase space $X_+$ with a zero element. The endowment of a Polish metric $d$ enforces $U_t(D) \to 0$ as $t \nearrow T_{\max}(D)$. Solution flow $D \mapsto U_t(D)$ remains continuous in this topology. This enables rigorous definition of statistical solutions "beyond blow-up", applicable to uncertainty quantification and numerical schemes (Feireisl et al., 2022).

Monte Carlo methods: Empirical distributions from Monte Carlo samples of random $D^n$ converge almost surely in the sense of the Banach-valued strong law (Ledoux–Talagrand) to the statistical solution $V_t$, linking theory with probabilistic numerical methods.

6. Analytical and Probabilistic Implications

This framework establishes a mathematically sound basis for statistical solutions of compressible NSF systems—measure-valued, consistent for random or uncertain data, supporting robust propagation up to (and beyond) singularity formation. The tightness and continuity properties facilitate rigorous uncertainty quantification and reliability analysis in physical or engineering simulations. The methods also provide a template for extending similar constructions to more complex multi-physics PDE systems, stochastic and measure-valued formulations, and for connecting deterministic and probabilistic solution concepts in nonlinear thermomechanics. The statistical approach is compatible with empirical algorithms, and avoids delicate selection issues present in other measure-valued PDE theories.

7. Conceptual Table: Solution Structures and Extensions

Solution Type	Definition	Extension Beyond Blow-Up
Strong solution	Pointwise, regular in Sobolev spaces	Not defined beyond $T_{\max}$
Statistical solution	Pushforward of data measure via solution operator	Set $U_t \equiv 0$ after blow-up
Empirical measure	Monte Carlo sample average of deterministic solutions	Converges to statistical soln

The statistical solution formalism enables interpretation and analysis of NSF dynamics under uncertainty, accommodates loss of regularity, and ensures stability with respect to input data.

---

The developments in (Feireisl et al., 2022) rigorously unify classical stability, modern measure-valued/statistical solution theory, and uncertainty quantification for compressible Navier-Stokes-Fourier systems, establishing a robust analytic foundation for theoretical and applied investigations.