Two immiscible fluids flowing through porous rock present a scale-up problem: the pore-scale physics is well understood, but translating it to reservoir-scale predictions has resisted clean theory. Relative permeability — the standard approach — works empirically but lacks connection to the microscopic dynamics.
Note (2026-11-13; read against the paper's full text, version 3): the paper describes itself as 'a short review' of a theory that 'has appeared in several papers' (it points to three earlier ones as the central sources), so the construction and the co-moving velocity are earlier work that it reviews, not results of this paper; it defines the co-moving velocity through the relation between the seeping velocities and the thermodynamic velocities, and says measurements of it, from analysis of relative permeability data and dynamic pore network models, suggest a simple linear functional form. I did not find the essay's statement that standard thermodynamics is 'missing a degree of freedom': the phrase 'missing variables' appears in the text I searched only as one of the alternative explanations that has been discussed for hysteresis in porous media, and the abstract's wording is that the co-moving velocity is 'a key variable with implications for ordinary thermodynamics'. The essay's closing claims, that the emergent variable did not exist at the pore scale and that scale-up creates new physics, are the essay's reading.
This paper applies Jaynes's information-theoretic approach to statistical mechanics. Instead of tracking every pore, you maximize entropy subject to known constraints. Out comes a continuum-scale description that is both connected to the pore physics and computationally tractable.
The framework yields a thermodynamic-like formalism with emergent variables that have no pore-scale counterpart. The most striking is the co-moving velocity — a velocity that describes how the two fluids move relative to each other in a way that respects the pore geometry. This variable doesn't exist at the pore scale. It emerges from the statistical treatment.
The co-moving velocity has implications for classical thermodynamics. It suggests that the standard thermodynamic description of multiphase flow is missing a degree of freedom — one that only becomes visible when you derive the macroscopic theory from the microscopic one rather than postulating it directly.
The lesson: scale-up isn't just coarse-graining. It can create genuinely new physics. The emergent variable wasn't hidden in the pore-scale description — it didn't exist there. It was born from the act of averaging, not from the thing being averaged.