---
title: Core-Annular WRIBL Model
url: https://www.emergentmind.com/topics/core-annular-wribl-model
type: topic
---

# Core-Annular WRIBL Model

Searching arXiv for recent and foundational papers on core-annular WRIBL and related reduced models.
The **core-annular weighted residual integral boundary layer (WRIBL) model** is a reduced-order, one-dimensional representation of axisymmetric two-phase flow in a cylindrical tube, designed for **two hydrodynamically active immiscible Newtonian fluids** separated by a deformable interface. In the formulation rederived algorithmically in 2025, the inner fluid is the **core** phase and the outer fluid is the **annular** phase; the interface is located at \(r=d(z,t)\), and the principal dynamical variables are the interface position \(d(z,t)\) and the phase flow rates \(Q_c(z,t)\) and \(Q_a(z,t)\) [2507.05103]. The model belongs to the WRIBL class because it derives **one-dimensional evolution equations** by averaging the long-wave Navier–Stokes equations over the thin-gap direction with carefully chosen weight functions, thereby retaining **inertial effects, viscous coupling between phases, capillarity, longitudinal viscous diffusion, and nonlinear curvature** at \(O(\epsilon^2)\) while remaining substantially cheaper than direct simulation [2507.05103]. In the literature surrounding core-annular flows, it occupies an intermediate position between full diffuse-interface or sharp-interface simulations and simpler lubrication or slender-jet models; a closely related long-wave two-fluid model for a corrugated tube provides a lubrication-type limiting case rather than a genuine WRIBL formulation [2007.03145], while phase-field simulations provide high-fidelity benchmarks for laminar, linear, and nonlinear regimes [1902.07351].

## 1. Physical setting and reduced variables

The model concerns **axisymmetric core-annular flow** in a cylindrical tube of radius \(R\), with two immiscible Newtonian fluids of densities \(\rho_c,\rho_a\), viscosities \(\mu_c,\mu_a\), and interfacial tension \(\gamma\). In cylindrical coordinates \((r,z)\), the interface is at
\[
r=d(z,t),
\]
so that \(d\) is the nondimensional interface position and the annular film thickness is \(1-d\) [2507.05103].

The long-wave parameter is
\[
\epsilon=\frac{R}{\Lambda}\ll 1,
\]
where \(\Lambda\) is the axial length scale [2507.05103]. The nondimensionalization uses
\[
r^* = R,\qquad z^* = \Lambda,\qquad t^* = \frac{R}{U},\qquad d^* = R,
\]
\[
u_i^* = U,\qquad v_i^* = \epsilon U,\qquad p_i^* = \frac{\mu_i U}{\epsilon R},
\]
with \(U\) a characteristic axial velocity [2507.05103]. The principal dimensionless groups are
\[
Re_i = \frac{R U \rho_i}{\mu_i},\qquad \Pi_\mu=\frac{\mu_c}{\mu_a},\qquad Ca=\frac{\gamma}{\mu_a U},
\]
and
\[
\mathcal{B}_i = \frac{b_i R^2}{\mu_i U},
\]
with \(b_i\) the dimensional axial body force per unit volume [2507.05103].

A defining feature of the core-annular WRIBL formulation is that **both phases are retained dynamically**. The reduced model does not treat the outer phase as a passive gas or an algebraically eliminated lubricating film. Instead, it evolves the interface position together with the **cross-section averaged flow rates**
\[
Q_c(z,t),\qquad Q_a(z,t),
\]
so that the one-dimensional dynamical system represents both core and annulus as momentum-carrying layers [2507.05103].

This suggests a central conceptual distinction from lubrication-type limits. In a slender-jet or lubrication reduction, one phase may be quasi-eliminated through a local closure. In the WRIBL model, by contrast, the annulus remains an active participant through its own flow-rate dynamics, interfacial stress transmission, and weighted residual projection [2507.05103].

## 2. Long-wave equations and interfacial conditions

Retaining terms up to \(O(\epsilon^2)\), the axisymmetric continuity equation in each phase is
\[
\frac{1}{r}\partial_r (r v_i) + \partial_z u_i = 0.
\]
The axial momentum equation is
\[
\epsilon Re_i\left(\partial_t u_i + v_i\partial_r u_i+ u_i\partial_z u_i\right) = -\partial_z \left(p_i |_{d}\right) -\epsilon^2 \partial_z \left(\partial_z u_i  |_{d}\right) +\frac{1}{r}\partial_r \left(r\partial_r u_i\right) +2\epsilon^2 \partial_{zz} u_i+\mathcal{B}_i .
\]
These forms arise after using the \(O(\epsilon^2)\) radial momentum balance to eliminate \(\partial_z p_i\) in favor of interfacial values [2507.05103].

At the interface \(r=d(z,t)\), the velocity continuity conditions are
\[
u_a=u_c,\qquad v_a=v_c.
\]
Tangential stress continuity, to \(O(\epsilon^2)\), is
\[
\partial_r u_a- \Pi_{\mu} \partial_r u_c = \left[2\epsilon^2 \partial_z d\left(\partial_z u_a- \partial_r v_a\right)-\epsilon^2 \partial_z v_a\right] -\Pi_{\mu}\left[2\epsilon^2 \partial_z d\left(\partial_z u_c-\partial_r v_c\right)-\epsilon^2 \partial_z v_c\right].
\]
The normal stress balance is
\[
p_a - \Pi_{\mu} p_c =  - Ca\, \kappa + 2\epsilon^2\left(\partial_r v_a- \partial_r u_a \partial_z d \right) -2\epsilon^2  \Pi_{\mu}\left(\partial _r v_c- \partial_r u_c \partial_z d \right),
\]
with curvature
\[
\kappa = \frac{1}{d}-\frac{\epsilon^2(\partial_z d)^2}{2d}-\epsilon^2 \partial_{zz} d.
\]
The kinematic condition is
\[
\partial_t d = v_a - u_a\partial_z d.
\]
At the centerline \(r=0\),
\[
v_c=0,\qquad \partial_r u_c=0,
\]
and at the wall \(r=1\),
\[
v_a=0,\qquad u_a=0.
\]
These equations remain two-dimensional and free-boundary; WRIBL reduces them to one-dimensional form [2507.05103].

The derivation assumes an \(O(\epsilon^2)\) model and additionally states that terms of order \(O(Re_i\epsilon^2)\) are neglected. It also assumes \(u_i'\), \(v_i'\) are at most \(O(\epsilon)\) [2507.05103]. A plausible implication is that the model is intended for moderate inertial regimes where long-wave inertial transport is important but higher-order inertial corrections do not dominate.

## 3. Weighted residual construction

The derivation introduces a decomposition of the axial and radial velocities into quasi-steady local profiles plus corrections:
\[
u_i(r,z,t)=\hat u_i(r;d,Q_i)+u_i'(r,z,t),\qquad
v_i(r,z,t)=\hat v_i(r;d,Q_i)+v_i'(r,z,t).
\]
Here \(\hat u_i\) is the quasi-steady, quasi-developed profile parametrized by the local interface position \(d\) and local phase flow rate \(Q_i\), and \(u_i'\) is a correction [2507.05103]. The leading profile is chosen so that it exactly reproduces the local phase flow rates:
\[
\int_0^d \hat u_c\, r\,dr = Q_c,\qquad \int_d^1 \hat u_a\, r\,dr = Q_a.
\]

At leading order, the axial profiles satisfy
\[
\frac{1}{r}\partial_r\left(r\partial_r \hat u_c\right)=A_c,\qquad \frac{1}{r}\partial_r\left(r\partial_r \hat u_a\right)=A_a,
\]
with \(A_c,A_a\) independent of \(r\), subject to centerline, wall, and interfacial matching conditions [2507.05103]. The general solutions are
\[
\hat u_c = \frac{A_c r^2}{4}+C_{11}+C_{12}\ln r, \qquad \hat u_a = \frac{A_a r^2}{4}+C_{21}+C_{22}\ln r,
\]
and regularity at the centerline removes the logarithm from the core phase [2507.05103].

The annular profile can be written exactly, although the explicit form is lengthy. For symbolic work it is rewritten as
\[
\hat{u}_a = r^2\big(A_1(d)Q_c+A_2(d)Q_a\big) +\log(r)\big(A_3(d)Q_c+A_4(d)Q_a\big) +\big(A_5(d)Q_c+A_6(d)Q_a\big),
\]
\[
\hat{u}_c = r^2\big(B_1(d)Q_c+B_2(d)Q_a\big) +\big(B_5(d)Q_c+B_6(d)Q_a\big),
\]
where the coefficient functions are exact shorthand for rational-logarithmic functions of \(d\) [2507.05103].

From continuity, the radial velocities are
\[
\hat v_c = -\frac{1}{r}\int_0^r \partial_z \hat u_c\, r\,dr, \qquad \hat v_a = \frac{1}{r}\int_r^1 \partial_z \hat u_a\, r\,dr.
\]
Enforcing continuity of \(\hat v_i\) at the interface yields
\[
\partial_z Q_a + \partial_z Q_c = 0,
\]
which is the exact one-dimensional phase-flow constraint [2507.05103].

The weighted residual step then averages the momentum residual with phase-dependent weight functions \(w_i(r)\) using
\[
\langle p|q\rangle = \Pi_\mu \int_0^d p_c q_c\, r\,dr + \int_d^1 p_a q_a\, r\,dr.
\]
The weights are chosen so that integration by parts closes the dominant transverse viscous term exactly and eliminates the unknown corrections \(u_i'\) [2507.05103]. They satisfy
\[
\frac{1}{r}\partial_r(r\partial_r w_c)=C_c,\qquad \frac{1}{r}\partial_r(r\partial_r w_a)=-1,
\]
with
\[
w_a=0\quad \text{at } r=1,\qquad \partial_r w_c=0\quad \text{at } r=0,\qquad \partial_r w_a=\Pi_\mu \partial_r w_c\quad \text{at } r=d.
\]
For the flow-rate equation, the integral constraint is
\[
\int_0^d w_c\, r\,dr = -\int_d^1 w_a\, r\,dr.
\]
This choice allows the interfacial pressure jump to enter through the normal-stress balance and removes explicit pressure-gradient dependence from the averaged momentum equation [2507.05103].

This construction is characteristic of WRIBL methodology. The reduced equations are not obtained by direct asymptotic elimination alone, but by a weighted projection designed to preserve dominant transverse diffusion and to encode cylindrical geometry and two-phase coupling in closed coefficient functions.

## 4. Governing equations of the core-annular WRIBL model

The exact interface evolution law follows from continuity integrated across the annulus:
\[
d\,\partial_t d = \partial_z Q_a,
\]
or equivalently,
\[
d\,\partial_t d = -\partial_z Q_c,
\]
using \(\partial_z Q_a+\partial_z Q_c=0\) [2507.05103].

The principal second-order WRIBL averaged-momentum equation is
\[
\Pi_{\mu i}Re_i\biggl( S_{ij}\partial_t Q_j + F_{ijk} Q_j\,\partial_z Q_k + G_{ijk}Q_jQ_k\,\partial_z d \biggr) = \Pi_{\mu}C_cQ_c-Q_a -Ca\,(\partial_z\kappa)\,I +\bigl[\Pi_{\mu}\mathcal{B}_c-\mathcal{B}_a\bigr]I
\]
\[
\qquad +J_jQ_j(\partial_z d)^2 +K_j(\partial_z Q_j)(\partial_z d) +L_jQ_j\,\partial_z^2 d +M_j\partial_z^2Q_j.
\]
Repeated indices are summed, and
\[
\Pi_{\mu a}=1,\qquad \Pi_{\mu c}=\Pi_\mu.
\]
The coefficient sets
\[
S_{ij},\ F_{ijk},\ G_{ijk},\ I,\ J_j,\ K_j,\ L_j,\ M_j
\]
are functions of \(d\) only [2507.05103].

The curvature used in the capillary term is
\[
\kappa = \frac{1}{d}-\frac{\epsilon^2(\partial_z d)^2}{2d}-\epsilon^2 \partial_{zz} d,
\]
so that
\[
\partial_z\kappa = -\frac{\partial_z d}{d^2} -\epsilon^2\partial_z\!\left(\frac{(\partial_z d)^2}{2d}\right) -\epsilon^2 \partial_{zzz}d.
\]
Accordingly, the capillary forcing term \(-Ca(\partial_z\kappa)I\) contains both the classical cylindrical Rayleigh–Plateau contribution and the \(O(\epsilon^2)\) nonlinear-curvature and axial-curvature corrections [2507.05103].

A second weight choice yields a diagnostic pressure equation:
\[
\Pi_{\mu i}Re_i\biggl( \tilde S_{ij}\partial_t Q_j +\tilde F_{ijk}Q_j\partial_z Q_k +\tilde G_{ijk}Q_jQ_k\partial_z d \biggr) = \Pi_{\mu}\tilde C_cQ_c-Q_a -2\Pi_{\mu}\,\partial_z p_c|_d\,\tilde I +Ca(\partial_z\kappa)\tilde I +\bigl[\Pi_{\mu}\mathcal{B}_c+\mathcal{B}_a\bigr]\tilde I
\]
\[
\qquad +\tilde J_jQ_j(\partial_z d)^2 +\tilde K_j(\partial_z Q_j)\partial_z d +\tilde L_jQ_j\partial_z^2 d +\tilde M_j\partial_z^2Q_j.
\]
The paper notes that the full phase pressures are radially uniform at this order because \(\partial_r p_i=0\) follows from the \(O(\epsilon^2)\) radial momentum equation [2507.05103].

The physical interpretation of the principal terms is explicit. The left-hand side
\[
\Pi_{\mu i}Re_i\left(S_{ij}\partial_tQ_j + F_{ijk}Q_j\partial_zQ_k + G_{ijk}Q_jQ_k\partial_z d\right)
\]
represents weighted inertial response, convective nonlinearities, and nonparallel geometric corrections. On the right-hand side, \(\Pi_\mu C_cQ_c-Q_a\) is the leading transverse-viscous closure; \(-Ca(\partial_z\kappa)I\) is capillary forcing; \([\Pi_\mu \mathcal B_c-\mathcal B_a]I\) is differential forcing; and the \(J_j,K_j,L_j,M_j\) terms encode \(O(\epsilon^2)\) geometric, interfacial-stress, and longitudinal-diffusion corrections [2507.05103].

## 5. Relation to lubrication and slender-jet models

The core-annular WRIBL model should be distinguished from lubrication-type two-phase reductions. A particularly relevant comparison is the long-wave model for an **axisymmetric viscous core thread** surrounded by a **less viscous annular fluid layer** inside a **cylindrical tube with axial wall corrugation** [2007.03145]. That model derives one-dimensional equations for the core radius \(S(z,t)\) and a core axial velocity \(w(z,t)\):
\[
S_t+\frac{1}{2}Sw_z+wS_z=0,
\]
\[
\bar R_e(w_t+ww_z)=3\frac{(S^2w_z)_z}{S^2}-\kappa_z-2\lambda_0\frac{G(S,d)}{S^2}w,
\qquad
\kappa=\frac1S-\epsilon^2S_{zz},
\]
with wall coupling through
\[
G(S,d)=\frac{S^2+d^2}{S^2-d^2-(S^2+d^2)\ln(S/d)}.
\]
In that formulation, the annular phase is not evolved through its own flow-rate equation; rather, annular stresses are collapsed into a local geometric resistance \(G(S,d)\), under the singular scaling \(\lambda=\epsilon^2\lambda_0\) and negligible annular inertia [2007.03145].

This model is therefore **not** a classical WRIBL model and **not** an explicit integral-boundary-layer system. It is more accurately described as a **slender-jet / lubrication-type active-interface model** with a local interface evolution law, a core axial momentum balance, capillary pressure closure, and wall-mediated annular drag [2007.03145]. Its relevance to WRIBL lies in structural adjacency: it provides explicit interfacial stress balances, capillary closure, and a wall-coupling mechanism that could serve as a creeping-flow or weak-annulus-viscosity limit for a more general core-annular WRIBL model [2007.03145].

The thin-annulus limit in the corrugated-tube study reinforces this distinction. Under
\[
d=1+\delta(1+\sigma'\phi),\qquad S=1+\delta-\delta h,
\]
with \(\delta\ll1\), the reduced model yields a modified Hammond-type thin-film equation [2007.03145]. This is explicitly a lubrication equation, not a WRIBL system.

A plausible implication is that the WRIBL framework should be viewed as a broader class capable of recovering lubrication or slender-jet closures in special limits while retaining two-phase flow-rate dynamics, systematic \(O(\epsilon^2)\) inertial and viscous-diffusion effects, and a more explicit representation of cylindrical two-layer momentum exchange.

## 6. Benchmarking against sharp-interface and diffuse-interface studies

Phase-field simulations of core-annular pipe flow provide a separate but closely related reference frame for assessing reduced models [1902.07351]. That work studies upward core-annular flow in a vertical circular pipe using a diffuse-interface CHNS formulation, but it also supplies an analytic **sharp-interface laminar solution**, linear stability benchmarks, and nonlinear axisymmetric wave dynamics that are directly relevant to WRIBL validation [1902.07351].

For a steady coaxial core-annular base state with interface radius
\[
\eta=\frac{R_I}{R},
\]
the centerline velocity of laminar CAF is
\[
U = (f-\rho_o g)\frac{R_I^2}{4\mu_o} + (f-\rho_w g)\frac{R^2-R_I^2}{4\mu_w} - (\rho_o-\rho_w)g \frac{R_I^2}{2\mu_w}\log\frac{R}{R_I},
\]
and the analytic sharp-interface velocity profile is
\[
u_z(r)= \begin{cases} \left(\dfrac{1}{\eta^2}-\dfrac{r^2}{\eta^2}-2(K_f-1)\log r\right)\dfrac{1}{A}, & 0<r\le \eta,\\[1.2ex] \left(1-\hat\mu \dfrac{r^2}{\eta^2}\right)\dfrac{K_f}{A}, & \eta<r\le 1. \end{cases}
\]
Here
\[
K_f=\frac{Fr\,f-1}{Fr\,f-\hat\rho},\qquad
A=\hat\mu K_f + \frac{1}{\eta^2}-1 + 2(K_f-1)\log\left(\frac{1}{\eta}\right).
\]
For matched density,
\[
f=\frac{1}{Re}\, \frac{4\hat\mu}{1-\eta^2+\eta^2 \hat\mu},
\]
and the total volume flux is
\[
\dot V=\frac{\pi}{2}\, \frac{1-\eta^4(1-\hat\mu)}{1-\eta^2(1-\hat\mu)}.
\]
These expressions provide exact steady-state closure targets for a reduced model [1902.07351].

The phase-field study identifies three instability classes summarized from Hu and Joseph: at low \(Re\), capillary instability dominates; for \(Re\gtrsim 50\), instability is predominantly interfacial-friction-driven; and for \(Re\gtrsim 200\), it becomes predominantly inertial [1902.07351]. This taxonomy is valuable for WRIBL interpretation because the weighted residual model explicitly retains the mechanisms most relevant to capillary and viscosity-stratified interfacial dynamics, and at least the long-wave branch of inertial instabilities.

In the nonlinear regime, the CHNS simulations reproduce axisymmetric **bamboo waves** and provide comparison targets such as hold-up ratio, wave speed, and centerline velocity reduction [1902.07351]. For the CAF3 benchmark, the leading eigenvalue is complex and the laminar state undergoes a **Hopf bifurcation** to a traveling wave of the form
\[
\boldsymbol u(r,z,\theta,t)=\boldsymbol u(r,z-c t,\theta,0),
\]
with computed wave speed \(c=0.83\) compared with a sharp-interface result \(c=0.85\) [1902.07351]. The paper also reports a diffuse-interface linear growth rate of about \(0.22\) versus a sharp-interface result \(0.248\) for the same case, with nonlinear effects becoming important around \(t\approx 20\) and a bamboo-wave pattern formed around \(t\approx 30\) [1902.07351].

These benchmarks are significant for WRIBL models because they test exactly the kinds of observables the reduced system is intended to capture: phase-flux partition, interfacial-wave growth, traveling-wave speed, and transport degradation in a periodic pipe. At the same time, the phase-field study warns that laminar high-viscosity-contrast CAF has diffuse-interface error
\[
\varepsilon=\mathcal O\!\left(Cn\log\hat\mu^{-1}\right),
\]
so sharp-interface analytic solutions remain the preferred reference where available [1902.07351].

## 7. Derivation workflow, numerical use, and scope

The 2025 derivation makes a methodological contribution by showing that the core-annular WRIBL model of Dietze and Ruyer-Quil can be reconstructed with **SymPy** through a sequence of symbolic substeps [2507.05103]. The workflow defines \(r,z,t\), constants such as \(\Pi_\mu\), unknown fields \(d(z,t)\), \(Q_c(z,t)\), \(Q_a(z,t)\), and symbolic radial functions \(\hat u_i(r)\); solves the ODEs for \(\hat u_i\) and the weight functions with `dsolve` and `solve`; abstracts cumbersome exact coefficients into functions \(A_i(d),B_i(d),D_i(d),E_i(d)\); computes \(\hat v_i\) by symbolic integration; assembles the weighted residual equation term by term; eliminates \(\partial_t d\) using
\[
d\,\partial_t d=\partial_zQ_a = -\partial_zQ_c;
\]
extracts the coefficients by symbolic expansion; and exports them for numerical use, including replacement of `log` with `np.log` for NumPy/SciPy implementation [2507.05103].

For time integration, the model is reformulated using
\[
Q_a = Q_t(t)-Q_c,
\]
where \(Q_t\) is the total flow rate, depending only on time, and the pressure equation is integrated over the periodic domain to obtain an ODE
\[
d_t Q_t = \Phi(d,Q_c,Q_t).
\]
The resulting numerical system comprises a PDE for \(d(z,t)\), a PDE for \(Q_c(z,t)\), and an ODE for \(Q_t(t)\) [2507.05103]. The discretization reported is:
- second-order central finite differences in \(z\),
- a uniform grid with 500 points,
- time integration using SciPy’s `solve_ivp` with **LSODA** [2507.05103].

The initial condition for Rayleigh–Plateau calculations is
\[
d = d_0 + 10^{-3}\cos\!\left(\frac{2\pi z R}{\Lambda}\right),
\]
on a periodic domain of nondimensional length \(\Lambda/R\) [2507.05103]. The paper studies, among other cases, an air-core–mucus-annulus configuration with strong capillary-driven hump growth and eventual liquid-bridge or plug formation, and an air-core–water-annulus gravity-driven case in which the hump saturates into a stable annular collar or unduloid that translates down the tube [2507.05103].

The scope and limitations are stated clearly. The model assumes:
- long-wave geometry \(\epsilon=R/\Lambda\ll1\),
- axisymmetry,
- Newtonian, immiscible fluids,
- retention of terms up to \(O(\epsilon^2)\),
- Reynolds numbers small enough that \(O(Re_i\epsilon^2)\) terms are neglected,
- radially uniform pressure at this order,
- an interface that remains a **single-valued function** \(r=d(z,t)\) [2507.05103].

It captures well:
- inertia,
- interfacial capillarity including nonlinear curvature,
- longitudinal viscous diffusion,
- two-phase coupling,
- collar or plug formation onset and pre-coalescence dynamics [2507.05103].

Its limitations are equally important. Once the interface touches the wall or centerline and topology changes, the single-valued thin-film description breaks down. The basic WRIBL model cannot continue beyond true coalescence or dryout events without additional regularization; for post-plug dynamics, it may be augmented with a **disjoining pressure** to create pseudo-plugs or precursor films [2507.05103]. It is also not intended for short-wave, strongly non-slender, or fully three-dimensional non-axisymmetric regimes [2507.05103].

Within those bounds, the core-annular WRIBL model occupies a precise role in the hierarchy of two-phase pipe-flow models. It is more general and dynamically richer than lubrication closures that quasi-eliminate one phase [2007.03145], yet substantially cheaper than full CHNS or sharp-interface DNS while retaining the essential multiscale physics needed for capillary instability, annular collar dynamics, and coupled phase-momentum transport [2507.05103, 1902.07351].

Source: https://www.emergentmind.com/topics/core-annular-wribl-model