On the justification of Koiter's model for generalised membrane shells of the "first kind'' confined in a half-space

Technical Summary: Justification of Koiter's Model for Generalised Membrane Shells of the First Kind in a Half-Space

Problem Statement
This paper addresses the mathematical justification of Koiter's model for linearly elastic generalised membrane shells of the first kind when subjected to a confinement condition. Specifically, the study considers shells that must remain confined within a prescribed half-space, a scenario relevant to applications such as the motion of heart valves or elastic films in contact with rigid foundations.

The core challenge lies in the "obstacle problem" formulation. Unlike standard Signorini conditions which typically constrain only the boundary of the deformed body, the confinement condition considered here requires that all points of the deformed reference configuration remain within the half-space. The paper seeks to demonstrate that the two-dimensional limit model derived rigorously from the three-dimensional linearised elasticity equations (as the shell thickness $\varepsilon \to 0$) coincides with the two-dimensional model derived directly from Koiter's model under the same confinement constraints.

A significant difficulty arises because the set of admissible displacements for the limit model obtained from the 3D obstacle problem (denoted $U^\sharp_M(\omega)$) is defined over the abstract completion of a non-empty, closed, and convex set with respect to a specific semi-norm. In contrast, the set of admissible displacements for the limit model derived from Koiter's model (denoted $U^\sharp(\omega)$) is defined as the closure of a different set. While the variational inequalities governing the solutions are identical, the sets in which the solutions are sought do not, in general, coincide. The paper aims to prove that these sets are indeed equal under specific geometric conditions, thereby completing the justification of Koiter's model for this class of shells.

Methodology
The methodology follows a rigorous asymptotic analysis framework, building upon the foundational work of P.G. Ciarlet and collaborators.

1. Three-Dimensional Formulation: The study begins with the classical three-dimensional energy for linearised elasticity. The admissible displacements are sought in a Sobolev space constrained by the condition that the deformed configuration lies entirely within a half-space $H = \{x \in E^3; x \cdot q \ge 0\}$.
2. Scaling and Dimension Reduction: The problem is scaled onto a fixed domain $\Omega = \omega \times (-1, 1)$ independent of the thickness parameter $\varepsilon$. The authors perform an asymptotic analysis as $\varepsilon \to 0$.
3. Kirchhoff-Love Assumptions: A critical methodological step involves assuming the validity of Kirchhoff-Love assumptions (geometric constraints where normals to the middle surface remain normal after deformation). The authors show that under these assumptions, the global confinement condition is equivalent to the classical Signorini condition applied to the boundary faces. This equivalence is crucial for establishing the convergence of the displacement field to the limit space.
4. Abstract Completion: The analysis reveals that the limit solution lies in the abstract completion of the space of admissible displacements with respect to the norm induced by the linearised change of metric tensor. This leads to a two-dimensional variational inequality problem posed over an abstractly completed set.
5. Density Arguments: To bridge the gap between the 3D-derived limit and Koiter's limit, the paper employs density arguments. It constructs dense subsets of smooth functions within the admissible sets to prove that the two distinct sets of admissible limit solutions ($U^\sharp_M(\omega)$ and $U^\sharp(\omega)$) are identical.

Key Contributions and Results

• Convergence to a Two-Dimensional Limit (Theorem 5.1): The paper establishes that the unique solution to the three-dimensional obstacle problem for a generalised membrane shell of the first kind converges, as the thickness approaches zero, to the unique solution of a two-dimensional variational inequality. This limit model is posed over the abstract completion of a closed, convex set. The authors note that while the variational inequalities match those derived from Koiter's model, the underlying sets of admissible solutions differ in their definitions.
• Equivalence of Admissible Sets (Theorem 6.3): The primary novelty of the work is the proof that the set of solutions derived from the 3D obstacle problem coincides with the set derived from Koiter's model, provided specific geometric conditions are met. These conditions include:
  • The domain $\omega$ is a rectangle (or a special case where the boundary is fully clamped).
  • The portion of the boundary where displacement is fixed ($\gamma_0$) consists of whole edges of the rectangle.
  • A symmetry condition (Condition 19) holds regarding the extension of the basis vectors and the normal vector $q$ via reflection across the domain boundaries.
    Under these conditions, the density argument proves $U^\sharp_M(\omega) = U^\sharp(\omega)$, thereby unifying the two approaches.
• Equivalence of Confinement and Signorini Conditions (Lemma 4.2): The paper demonstrates that the global confinement condition (applying to all points in the shell) is equivalent to the classical Signorini condition (applying to the boundary) when Kirchhoff-Love assumptions are valid.

Significance and Claims
The paper claims to complete the justification of Koiter's model for linearly elastic generalised membrane shells of the first kind subjected to confinement in a half-space. Prior to this work, the identification of a two-dimensional limit model derived directly from the three-dimensional obstacle problem for this specific class of shells was an open problem.

The authors emphasize that their work resolves a discrepancy between two mathematical pathways:

1. Deriving the limit from the full 3D elasticity equations with obstacles.
2. Deriving the limit from Koiter's 2D model with obstacles.

While previous studies had justified Koiter's model for elliptic membrane shells and flexural shells, the generalised membrane case remained unresolved due to the complexities of the abstract completion of the solution space. By proving that the sets of admissible solutions coincide under reasonable geometric assumptions, the paper validates that Koiter's model correctly captures the asymptotic behavior of the 3D obstacle problem for these shells.

The authors remain modest regarding the geometric restrictions required for the density result (Theorem 6.3), acknowledging that the proof relies on the domain being a rectangle and specific boundary configurations. However, they argue that these assumptions cover non-trivial and practically relevant cases, such as portions of spherical caps, cylinders, and hyperboloids of revolution, as illustrated in the paper. The work also highlights that the assumption of Kirchhoff-Love conditions is an analytical necessity to bypass the lack of compactness in the abstract completion space, a technique consistent with established shell theory.