---
title: Brinkman–Cahn–Hilliard Flow Models
url: https://www.emergentmind.com/topics/brinkman-cahn-hilliard-system
type: topic
---

# Brinkman–Cahn–Hilliard Flow Models

Searching arXiv for recent and foundational Brinkman–Cahn–Hilliard papers to ground the article.
The Brinkman–Cahn–Hilliard system is a class of diffuse-interface partial differential equation models coupling a Cahn–Hilliard-type phase evolution for an order parameter with a Brinkman-type momentum balance for the velocity field. In these models, the phase variable describes composition or volume-fraction contrast in a binary mixture or related multiphase medium, while the Brinkman equation accounts for viscous flow in porous or partially porous environments through a combination of viscous stress and linear drag. The coupling is bidirectional: the flow advects the phase field, and the phase field feeds back into the momentum equation through a capillary or Korteweg force such as $\mu\nabla\phi$ or $(\mu+\chi\sigma)\nabla\varphi$. Across the literature, Brinkman–Cahn–Hilliard systems appear in incompressible binary-fluid flow, porous-medium phase separation, tumor-growth models, nonlocal interaction models, higher-order curvature-driven phase separation, optimal control, stochastic extensions, and vanishing-viscosity limits toward Darcy or Hele–Shaw regimes [1402.6195], [1601.03272], [1811.06699], [2509.20282], [2601.06698].

## 1. Core mathematical structure

A standard local Brinkman–Cahn–Hilliard model couples
\[
\partial_t \phi + \nabla\cdot(\phi \mathbf u)=\Delta \mu,\qquad
\mu=-\Delta\phi+f(\phi),
\]
with a Brinkman momentum balance
\[
-\nabla\cdot(\nu D(\mathbf u))+\eta \mathbf u+\nabla p=\mu\nabla\phi,\qquad
\nabla\cdot\mathbf u=0,
\]
or equivalent variants differing by normalization, coefficient dependence, and source terms [1402.6195]. In this formulation, $\phi$ is the order parameter, $\mu$ the chemical potential, $\mathbf u$ the averaged velocity, and $p$ the pressure. The term $\mu\nabla\phi$ is the Korteweg force, while $\eta\mathbf u$ is the porous-medium drag characteristic of Brinkman flow [1402.6195], [2506.07128].

This baseline structure admits several important generalizations. In nonlocal models, the chemical potential takes the form
\[
\mu=a\varphi-J*\varphi+F'(\varphi),
\]
so the local gradient term is replaced by convolution with an interaction kernel $J$ [1601.03272], [1911.02811], [2502.04222]. In tumor-growth systems, additional fields such as nutrient concentration $\sigma$ enter the phase and momentum equations, and the velocity may be non-solenoidal because $\operatorname{div}(v)$ is prescribed by solution-dependent source terms [1811.06699], [1909.02289], [2204.13526], [2411.12505]. In higher-order curvature models, the phase subsystem is reformulated using auxiliary variables and the chemical potential itself becomes fourth order in the phase field, yielding an overall sixth-order evolution [2509.20282], [2512.07682], [2601.05820].

The principal mechanism is always the same: advection by Brinkman flow modifies phase separation, while diffuse-interface thermodynamics modifies momentum transport. This makes the system an intermediate model between pure Cahn–Hilliard diffusion, Darcy-type porous-medium flow, and Stokes-like viscous flow [1402.6195], [2506.07128].

## 2. Variational and energetic formulation

A defining feature of Brinkman–Cahn–Hilliard systems is their energetic structure. For the classical local model, the free energy is
\[
E(\phi)=\int_{\Omega}\left(\frac{\varepsilon^{2}}{2}|\nabla\phi|^2+F(\phi)\right)\,dx,
\]
and, under periodic or no-flux/incompressible settings, the continuous dissipation law takes the form
\[
\frac{d(E(\phi))}{dt}
=
-\int_{\Omega}\left(
M(\phi)|\nabla\mu|^2
+\frac1\gamma \eta(\phi)|\mathbf u|^2
+\frac{1}{2\gamma}\nu(\phi)|D(\mathbf u)|^2
\right)\,dx
\]
in one common normalization [2506.07128]. In the constant-coefficient CHB analysis, the corresponding identity is
\[
\frac{d}{dt}\left(\frac12 \|\nabla\phi\|^2+\int_\Omega F(\phi)\,dx\right)
+\|\nabla\mu\|^2+\nu\|\nabla\mathbf u\|^2+\eta\|\mathbf u\|^2=0
\]
[1402.6195].

For nonlocal systems, the energy functional is
\[
\mathcal E(\varphi)=
\frac14\int_\Omega\int_\Omega J(x-y)(\varphi(x)-\varphi(y))^2\,dx\,dy
+\int_\Omega F(\varphi)\,dx,
\]
equivalently
\[
\mathcal E(\varphi)=
\frac12(a\varphi,\varphi)-\frac12(\varphi,J*\varphi)+\int_\Omega F(\varphi)\,dx
\]
[1601.03272]. This formulation is analytically advantageous because the nonlocal chemical potential remains second-order in space rather than fourth-order, which is central to the nonlocal CHHS limit theory [1601.03272].

In curvature-driven models, the energy is of Willmore/functionalized-Cahn–Hilliard type:
\[
\mathcal E(\phi)
=
\frac12\int_\Omega\Bigl(-\Delta\phi+f(\phi)\Bigr)^2\,dx
+
\nu\int_\Omega\left(\frac12|\nabla\phi|^2+F(\phi)\right)\,dx,
\]
with
\[
w=-\Delta\phi+f(\phi),\qquad
\mu=-\Delta w+\bigl(f'(\phi)+\nu\bigr)w
\]
[2509.20282]. This is the mechanism behind the “curvature-induced phase separation” terminology: the chemical potential is the first variation of a higher-order free energy, so the phase dynamics becomes sixth-order after elimination of $w$ [2509.20282], [2512.07682].

Source terms can break strict gradient-flow structure. In the higher-order Brinkman system with
\[
\partial_t\phi+\mathbf v\cdot\nabla\phi-\operatorname{div}(m(\phi)\nabla\mu)=S(\phi),
\qquad
S(\phi)=-\sigma\phi+h(\phi),
\]
the spatial mean of the phase is not conserved, and the mean of $\mu$ must be estimated separately [2509.20282]. In tumor models, source terms in both divergence and phase equations similarly destroy mass conservation and complicate control of chemical-potential averages [1811.06699], [1909.02289].

## 3. Principal model variants

The Brinkman–Cahn–Hilliard literature is heterogeneous. The main variants differ in the order of the phase operator, local versus nonlocal interaction, compressibility, and auxiliary couplings such as nutrient or chemotaxis.

| Variant | Characteristic feature | Representative papers |
|---|---|---|
| Classical local CHB | Fourth-order phase field, Brinkman drag and viscosity | [1402.6195], [2506.07128] |
| Nonlocal CHB | $\mu=a\varphi-J*\varphi+F'(\varphi)$ | [1601.03272], [1911.02811], [2502.04222] |
| Tumor-growth CHB | Nutrient coupling, sources, often non-solenoidal velocity | [1811.06699], [1909.02289], [2204.13526], [2411.12505] |
| Sixth-order Brinkman–CH | Curvature effects via $w=-\Delta\phi+f(\phi)$ | [2509.20282], [2512.07682], [2601.05820] |
| Stochastic CHB | Multiplicative Wiener noise, dynamic boundary conditions | [2601.06698] |

The classical CHB system for incompressible binary flow in porous media treats $\mathbf u$ as divergence-free and typically imposes no-slip or traction-free velocity conditions together with Neumann conditions for $\phi$ and $\mu$ [1402.6195], [1601.03272]. Nonlocal CHB replaces the local interfacial term with long-range interaction through $J$, which leads to stronger well-posedness results for the corresponding Hele–Shaw limit than are known in the local case [1601.03272].

Tumor-growth variants add nutrient concentration $\sigma$, mass-production terms, chemotaxis, and compressibility. One model uses
\[
\operatorname{div}(v)=\Gamma_v(\varphi,\sigma),\qquad
-\operatorname{div}(T(v,p))+\nu v=(\mu+\chi\sigma)\nabla\varphi,
\]
with the phase equation containing both convection and source terms, and a quasi-static elliptic nutrient equation [1811.06699], [1909.02289]. In this setting, traction-free or frictionless velocity conditions are analytically important because Dirichlet conditions would impose the restrictive compatibility condition $\int_\Omega \Gamma_v(\varphi,\sigma)\,dx=0$ [1811.06699], [1909.02289].

The higher-order sixth-order system studied in three dimensions introduces the unknowns $(\mathbf v,p,\phi,\mu,w)$ and the coupled equations
\[
-\operatorname{div}\mathbf T(\phi,\mathbf v,p)+\lambda(\phi)\mathbf v=\mu\nabla\phi+\mathbf u,\qquad
\operatorname{div}\mathbf v=0,
\]
\[
\partial_t\phi+\mathbf v\cdot\nabla\phi-\operatorname{div}(m(\phi)\nabla\mu)=S(\phi),
\]
\[
-\Delta w+f'(\phi)w+\nu w=\mu,\qquad
-\Delta\phi+f(\phi)=w
\]
[2509.20282]. This formulation makes the sixth-order structure analytically tractable by reducing it to a mixed second-order system.

## 4. Existence, uniqueness, and regularity theory

The foundational local CHB analysis established global weak well-posedness for $d=2,3$ with constant coefficients under polynomial-growth assumptions on $f$ and showed uniqueness under stronger regularity assumptions on the potential [1402.6195]. In that framework, weak solutions satisfy
\[
\phi \in C([0,T];H^1)\cap L^2(0,T;H^3),\qquad
\mathbf u \in L^2(0,T;\mathbf V),
\]
with mass conservation and the energy identity as central a priori tools [1402.6195]. The same work proved eventual higher regularity, existence of a global attractor, and convergence of trajectories to single equilibria via the Łojasiewicz–Simon inequality [1402.6195].

For the nonlocal CHB system,
\[
\varphi_t + \nabla \cdot (\mathbf u \varphi)=\Delta \mu,\qquad
\mu=a\varphi-J*\varphi+F'(\varphi),
\]
global weak existence follows under assumptions on $J$, $F$, viscosity, permeability, and forcing [1601.03272]. With constant viscosity, weak solutions are unique, and bounded initial data yield $\varphi,\mu\in L^\infty(\Omega\times(0,T))$ [1601.03272]. In the two-dimensional regular-potential setting, this theory can be upgraded to strong solutions with
\[
\varphi \in L^\infty(0,T;H^2(\Omega)),\qquad
\varphi_t \in L^\infty(0,T;H)\cap L^2(0,T;V),\qquad
\mathbf u \in L^\infty(0,T;H^1(\Omega)^2)
\]
[1911.02811].

Singular-potential theory is substantially more delicate. For tumor-growth CHB systems with source terms and non-solenoidal velocity, weak and stationary solutions have been established for both double-obstacle and logarithmic potentials [1909.02289]. A key difficulty there is that source terms destroy the usual conservation of the mean phase, so one must replace the standard mass-conservation argument by a proof that the spatial mean of $\varphi$ remains strictly inside $(-1,1)$; this then yields control of the mean of $\mu$, which is essential for singular-potential analysis [1909.02289]. A related tumor-growth model with Dirichlet boundary condition for the chemical potential was shown to admit weak solutions for regular, logarithmic, and double-obstacle potentials; the Dirichlet condition is analytically decisive because it allows direct control of $\mu$ in $H_0^1(\Omega)$ from $\nabla\mu$ without requiring a separate estimate of its mean [2204.13526].

The sixth-order Brinkman–Cahn–Hilliard system with curvature effects admits weak solutions in three dimensions under variable shear viscosity, variable drag, and nonconservative source terms [2509.20282]. In the divergence-free variational framework,
\[
\mathbf v\in L^2(0,T;\mathbf V_\sigma),\quad
\phi\in H^1(0,T;V^*)\cap L^\infty(0,T;W)\cap L^2(0,T;H^5(\Omega)),
\]
\[
\mu\in L^2(0,T;V),\quad
w\in L^\infty(0,T;H)\cap L^2(0,T;H^3(\Omega)\cap W),
\]
and the main existence theorem provides a uniform a priori bound of this full mixed system [2509.20282]. Uniqueness and continuous dependence are proved only when mobility and shear viscosity are constants, precisely because difference estimates become tractable only in the absence of nonlinear coefficient variations [2509.20282].

## 5. Singular potentials, degenerate mobility, and separation phenomena

Singular free energies and degenerate mobilities play a special role because they enforce physically meaningful bounds on the phase and modify regularity mechanisms. In the nonlocal CHB system with logarithmic potential
\[
F_{\log}(r)=(1+r)\log(1+r)+(1-r)\log(1-r),
\]
and degenerate mobility
\[
m(r)=1-r^2,
\]
the main recent result is a separation property in two dimensions: if $\|\varphi_0\|_{L^\infty}<1$ and $|\bar\varphi_0|<1$, then for any $\tau>0$ there exists $\delta\in(0,1)$ such that
\[
\sup_{t\ge \tau}\|\varphi(t)\|_{L^\infty(\Omega)}\le 1-\delta
\]
[2502.04222]. Equivalently, after any positive waiting time the solution stays uniformly away from the pure phases $\pm1$.

This result is nontrivial because the singularity of $F'$ near $\pm1$ and the vanishing of $m$ at $\pm1$ create opposite effects: singularity repels the solution from pure phases, while degeneracy weakens diffusion there [2502.04222]. The proof uses a De Giorgi iteration adapted to the nonlocal Brinkman-coupled setting, together with a crucial uniform $L^1$ bound on $F'(\varphi)$ and an entropy test built from a function $M$ satisfying $mM''=1$ [2502.04222]. The transport term caused by the Brinkman velocity is harmless in the iteration because incompressibility gives exact cancellation when truncation functions are used [2502.04222].

Singular potentials also dominate tumor-growth CHB theory. The double-obstacle and logarithmic cases ensure $\varphi$ remains in the physically relevant interval $[-1,1]$ or $(-1,1)$, which is important when $\varphi$ is interpreted as a difference of volume fractions [1909.02289], [2204.13526]. In modified chemotaxis-Brinkman-Cahn–Hilliard models, the Flory–Huggins logarithmic potential
\[
F(r)=(1+r)\log(1+r)+(1-r)\log(1-r)-\lambda r^2
\]
is used together with nonlinear chemotactic sensitivity; the singular potential yields a uniform bound on $\varphi$, which is essential for controlling the cross-diffusion structure [2411.12505].

## 6. Limits, control, computation, and extensions

A major structural theme is the vanishing-viscosity limit from Brinkman to Darcy or Hele–Shaw dynamics. In the classical local CHB system, letting the Brinkman viscosity parameter tend to zero yields the Cahn–Hilliard–Hele–Shaw system, and in two dimensions there is even a quantitative estimate comparing CHB and CHHS solutions [1402.6195]. For the nonlocal CHB system, weak solutions of CHHS are obtained as limits of CHB solutions under $\nu\to 0$, and bounded weak solutions of the nonlocal CHHS problem are unique under constant permeability [1601.03272]. In tumor-growth CHB with nutrient coupling, vanishing viscosities lead to a Cahn–Hilliard–Darcy system with pressure satisfying an elliptic problem and velocity determined algebraically from capillary and nutrient forcing [1811.06699]. In the sixth-order curvature model, the Darcy limit is rigorously established by a sequence of viscosity coefficients $\eta_n$ with $\|\eta_n\|_{L^\infty(\mathbb R)}\to 0$, producing a reduced variational law
\[
\int_\Omega \lambda(\phi)\mathbf v\cdot\boldsymbol\zeta
=
\int_\Omega(\mu\nabla\phi+\mathbf u)\cdot\boldsymbol\zeta
\qquad
\forall \boldsymbol\zeta\in \mathbf H_\sigma
\]
[2509.20282].

Optimal control theory has recently become an active part of the Brinkman–Cahn–Hilliard literature. For the sixth-order system with constant mobility and viscosity, distributed forcing in the Brinkman equation has been treated as the control variable, and first-order necessary optimality conditions have been derived both for smooth quadratic costs and for sparsity-promoting functionals containing an $L^1$ term [2512.07682]. In that setting the control-to-state map is Fréchet differentiable, the adjoint system can be written explicitly, and the optimality condition takes the form of a variational inequality; for $\kappa>0$, the nonsmooth term yields sparse controls through subdifferential conditions [2512.07682]. A closely related higher-order optimal velocity control problem for a curvature-regularized Brinkman–Cahn–Hilliard system also derives adjoint-based first-order conditions and pointwise projection formulas for box-constrained controls [2601.05820]. In two-dimensional nonlocal CHB with regular potential, distributed forcing control has likewise been analyzed using strong-solution theory, a linearized system, and an adjoint system [1911.02811].

Numerically, recent work has focused on high-order, unconditionally energy-stable discretizations. A class of variable-step R-IMEX-BDF$(k)$ schemes based on a generalized scalar auxiliary variable approach with relaxation was developed for the CHB system with periodic boundary conditions [2506.07128]. These schemes are proved unconditionally energy-stable in the modified scalar energy variable, remain stable without adjacent-step-ratio restrictions, and support adaptive and hybrid-order time-stepping strategies tuned to the multiscale dynamics of phase separation [2506.07128]. The numerical experiments include manufactured-solution convergence tests, spinodal decomposition, and buoyancy-driven flow, and show that the hybrid adaptive algorithm effectively combines short-time third-order accuracy with long-time second-order robustness [2506.07128].

The model class has also expanded beyond deterministic bulk equations. A stochastic Cahn–Hilliard–Brinkman model with dynamic boundary conditions and independent multiplicative Wiener noises in bulk and surface phase equations has been formulated and weak martingale solutions have been proved in dimensions two and three [2601.06698]. This stochastic extension preserves the deterministic Brinkman–Cahn–Hilliard coupling but destroys exact mass conservation, which introduces substantial new difficulties in controlling chemical-potential means and boundary phase variables [2601.06698].

Overall, the Brinkman–Cahn–Hilliard system is best understood not as a single PDE model but as a family of coupled diffuse-interface flow systems organized around a common mechanism: Cahn–Hilliard thermodynamics for phase segregation coupled to Brinkman momentum transport for porous or porous-like flow. The main directions of current research are the treatment of singular and higher-order energies, rigorous Brinkman-to-Darcy limits, nonlocal and chemotactic couplings, optimal control, robust energy-stable numerics, and stochastic generalizations [1402.6195], [1601.03272], [1811.06699], [2509.20282], [2512.07682], [2506.07128], [2601.06698].

Source: https://www.emergentmind.com/topics/brinkman-cahn-hilliard-system