Stability of non-convex gradient dynamics from the perspective of multiscale non-equilibrium thermodynamics

Adam Janečka, Michal PavelkaarXiv 2017 · arXiv

Abstract

Reformulating constitutive relations in terms of gradient dynamics—as the derivative of a dissipation potential—provides critical insights into the stability, metastability, and instability of complex fluid dynamics. This paper introduces the concept of "CR-stability" (Constitutive Relation stability), which is fundamentally linked to the convexity of the dissipation potential. When this potential becomes non-convex, the system undergoes "dissipative phase transitions," analogous to classical equilibrium phase changes but driven by dissipation rather than free energy. By constructing evolution equations for perturbations within a thermodynamically sound framework, this study explains observed phenomena like shear stress jumps in complex fluids and supercritical boiling curves.

Introduction: The Puzzle of Flow Asymmetry

In the study of non-Newtonian fluids, such as certain polymer or micellar solutions, researchers often encounter a puzzling asymmetry in experimental data. When the shear stress (the applied force) is controlled and increased slowly, the shear rate (the flow speed) typically increases continuously. However, if the shear rate is the controlled variable, the resulting shear stress can exhibit sudden, discontinuous jumps to higher values.

Standard fluid mechanics models can describe these "S-shaped" flow curves but often fail to explain the underlying stability or why the system selects a specific path. This research bridges that gap by applying multiscale non-equilibrium thermodynamics to define CR-stability, providing a mathematical framework to predict when and why these instabilities occur.

Methodology: Gradient Dynamics and Lifting

The authors utilize several advanced thermodynamic tools to analyze these systems:

  • Dissipation Potentials ($\Xi$): The constitutive relation is reformulated as the derivative of a dissipation potential. Stability is then tied to the convexity of this potential.
  • The GENERIC Framework: Using the "General Equation for Non-Equilibrium Reversible-Irreversible Coupling" ensures the model adheres to the second law of thermodynamics (entropy production).
  • Extended Hydrodynamics (Lifting): By "promoting" shear stress from a dependent variable to an independent state variable, the authors can write evolution equations for the stress itself. This allows the system to "relax" toward the equilibrium constitutive relation, revealing the dynamics of stability.
  • Generalized Legendre Transformations: Because non-convex potentials result in multivalued conjugates, a "slope transform" is employed to handle multiple possible states without information loss.

Experimental Validation

The theoretical framework was validated using experimental data from a study by Boltenhagen et al. (1997) on a 7.5/7.5 mM TTAA/NaSal solution. This specific complex fluid is known for bulk phase separation and "vorticity banding" under shear. The authors successfully fitted a modified power-law model (the Le Roux-Rajagopal model) to this data, demonstrating the framework's ability to capture real-world behavior.

Experimental data and model fit for a TTAA/NaSal solution showing the shear rate vs. shear stress relationship.

Key Insights: Dissipative Phase Transitions

The core finding is that jumps in shear stress are mathematically identical to classical phase transitions (like the transition from water to steam). While classical transitions are driven by non-convex free energy, these flow instabilities are driven by non-convex dissipation.

One of the most significant results is the application of the Maxwell Lever Rule. The authors demonstrate that the famous "equal area" construction used to find phase coexistence in equilibrium thermodynamics also applies to these dissipative systems. The jump in the flow curve occurs precisely where the shaded areas on the graph are equal, as shown in Figure 4.

The equal area construction (Maxwell lever rule) applied to the non-convex dissipation potential, explaining the stress jump.

This geometric approach explains the observed asymmetry: when controlling the shear rate, the thermodynamic force encounters a multivalued potential, necessitating a jump. Conversely, when controlling stress, the mapping remains unique and continuous.

Conclusion and Future Directions

The paper successfully models the discontinuous jumps observed in experiments while maintaining thermodynamic consistency. It links fluid stability to the Braun–Le Chatelier principle, suggesting that stable shear-thinning is a natural response of the system to minimize external perturbations.

Future work aims to move beyond the current "toy model" by incorporating objective derivatives for physical consistency in rotating flows and adding "weakly non-local" effects. These additions would allow for more precise modeling of the physical "bands" or layers that form within fluids during these dissipative transitions.

Summary updated

Abstract

Reformulating constitutive relation in terms of gradient dynamics (being derivative of a dissipation potential) brings additional information on stability, metastability and instability of the dynamics with respect to perturbations of the constitutive relation, called CR-stability. CR-instability is connected to the loss of convexity of the dissipation potential, which makes the Legendre-conjugate dissipation potential multivalued and causes dissipative phase transitions, that are not induced by non-convexity of free energy, but by non-convexity of the dissipation potential. CR-stability of the constitutive relation with respect to perturbations is then manifested by constructing evolution equations for the perturbations in a thermodynamically sound way. As a result, interesting experimental observations of behavior of complex fluids under shear flow and supercritical boiling curve can be explained.
BibTeX
@misc{janecka2017stability,
  author = {Adam Janečka and Michal Pavelka},
  title = {{Stability of non-convex gradient dynamics from the perspective of multiscale non-equilibrium thermodynamics}},
  year = {2017},
  eprint = {1701.06094},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1701.06094},
}

Loading related content...

Figure