Analysis and virtual element discretisation of a Stokes/Biot--Kirchhoff bulk--surface model

1. Problem Statement

The paper addresses a coupled 3D–2D bulk–surface problem modeling the interaction between a free fluid and a thin poroelastic structure.

• Bulk Domain ($\Omega$): Contains a viscous, incompressible fluid governed by the Stokes equations.
• Surface Domain ($\Sigma$): Represents a thin, deformable, porous plate governed by the Biot–Kirchhoff equations. This model reduces the structural component to a 2D manifold while retaining key physical features (elasticity and poroelasticity).
• Coupling Mechanism: The interaction occurs at the interface $\Sigma$. The coupling enforces:
  • Kinematic continuity: Continuity of normal velocities between the fluid and the plate.
  • Dynamic balance: Balance of normal stresses (fluid pressure vs. plate stress) and tangential stresses governed by the Beavers–Joseph–Saffman–Jones condition (slip condition).
• Application Context: The model is motivated by immune isolation devices (e.g., for Type 1 diabetes treatment) where pancreatic islets are encapsulated in silicon nanopore membranes (SNM) to protect them from the immune system while allowing nutrient/insulin exchange.

2. Methodology

The authors propose a rigorous mathematical analysis and a numerical discretization strategy using the Virtual Element Method (VEM).

A. Mathematical Formulation

• Time Discretization: The problem is semi-discretized in time using a constant time step $\tau$, converting the time-dependent PDEs into a stationary coupled system.
• Weak Formulation: The continuous problem is cast as a double perturbed saddle-point problem. The solution space consists of:
  • Bulk velocity ($\mathbf{u}$) and pressure ($p$).
  • Surface pressure moment ($\varphi$) and plate deflection ($w$).
• Well-posedness Analysis: The existence and uniqueness of the continuous solution are proved using:
  • Fredholm's Alternative for compact operators.
  • Babuška–Brezzi theory for perturbed saddle-point problems with penalty terms.
  • Key steps involve establishing coercivity of the main operators and proving an inf-sup condition for the coupling operator.

B. Virtual Element Discretization (VEM)

The authors design a stable VEM scheme capable of handling general polyhedral (3D) and polygonal (2D) meshes.

• Spaces:
  • Stokes (Bulk): Enhanced 3D virtual element spaces ($\mathbf{V}_h^k$) that ensure divergence-free properties and handle complex geometries.
  • Poroelastic Plate (Surface): Conforming virtual element spaces for fourth-order problems ($W_h^k$ for deflection) and second-order problems ($R_h^l$ for pressure moment).
• Key Technical Innovation:
  • Construction of a novel Fortin interpolant with $\mathbf{H}^1$-regularity. This operator is crucial for proving the discrete inf-sup condition under a small mesh assumption, ensuring stability without requiring high regularity of the exact solution.
  • Use of polynomial projections ($\Pi^0, \Pi^\nabla, \Pi^{\nabla^2}$) to compute the discrete bilinear forms, making the method computable despite the non-polynomial nature of the basis functions.
  • Stabilization terms are added to ensure coercivity and consistency.

C. Implementation Strategy

• Splitting Scheme: Instead of solving the monolithic system directly, the authors implement an equivalent fixed-point (Picard) iteration.
  • The bulk Stokes problem and the surface poroelastic problem are solved sequentially, exchanging coupling variables (velocity and stress) at each iteration.
  • The paper proves that this splitting scheme is equivalent to the monolithic formulation under suitable assumptions and converges.
• Software: The method is implemented in the open-source VEM++ library (C++), utilizing hierarchical mesh structures to link 3D bulk elements with 2D surface faces.

3. Key Contributions

1. Rigorous Analysis: First complete theoretical analysis of a Stokes/Biot–Kirchhoff coupled system formulated as a double saddle-point problem, establishing well-posedness via Fredholm theory.
2. Stable VEM Scheme: Development of a compatible VEM for general polygonal/polyhedral meshes that handles the coupling between 3D Stokes flow and 2D fourth-order poroelastic plates.
3. Novel Interpolation: Construction of a Fortin interpolant tailored to the bulk-surface coupling, enabling the proof of the discrete inf-sup condition with only $\mathbf{H}^1$-regularity requirements.
4. Optimal Error Estimates: Derivation of optimal a priori error estimates in the energy norm, confirming convergence rates of $O(h^k)$ for velocity and pressure, and $O(h^{k-1})$ for surface variables.
5. Application Demonstration: Successful simulation of an immune isolation device using silicon nanopore membranes, validating the model's ability to capture fluid-structure interaction in biomedical contexts.

4. Results

• Convergence Rates: Numerical experiments on various polyhedral meshes (cubes, octahedra, Voronoi, etc.) confirm the theoretical convergence rates. The total error decays as $O(h^{k-1})$, consistent with the polynomial degree $k$ of the virtual elements.
• Stability: The fixed-point iteration scheme demonstrates robust convergence (typically within 4–5 iterations for the test cases) with a tolerance of $10^{-10}$.
• Application Simulation:
  • Simulated blood flow through a channel coupled with a silicon nanopore membrane (SNM).
  • Results showed realistic blood velocities ($\approx 0.34$ mm/s) and minute plate deflections ($\approx 10^{-10}$ mm) localized at inlet/outlet zones.
  • The simulation successfully captured the pressure-driven flow and the mechanical response of the porous membrane, validating the model for designing immune isolation devices.

5. Significance

This work bridges a critical gap in the numerical simulation of multiscale fluid-structure interactions involving lower-dimensional manifolds.

• Geometric Flexibility: By using VEM, the method overcomes the geometric limitations of traditional Finite Element Methods (FEM), allowing for the use of general polyhedral meshes which are often necessary for complex biomedical or industrial geometries.
• Computational Efficiency: The proposed splitting scheme allows for the reuse of existing solvers for Stokes and plate problems, reducing implementation complexity while maintaining the stability of a monolithic approach.
• Biomedical Impact: The specific application to immune isolation provides a validated tool for optimizing the design of encapsulation devices for Type 1 diabetes, potentially improving the longevity and efficacy of islet transplants.

In summary, the paper provides a mathematically sound, numerically stable, and practically applicable framework for simulating complex bulk-surface interactions in poroelastic systems, with a specific focus on advancing biomedical device design.