Shell theory comparison

Linear KLKoiterReissner–MindlinNaghdi
DOFs/node3 ($u_1,u_2,u_3$)3 ($u_1,u_2,u_3$)5 ($u_1,u_2,u_3,\varphi_1,\varphi_2$)5 ($u_1,u_2,u_3,\varphi_1,\varphi_2$)
Directorimplicit: $\mathbf{n} = \mathbf{a}_1\times\mathbf{a}_2/|\cdot|$implicit: $\mathbf{n} = \mathbf{a}_1\times\mathbf{a}_2/|\cdot|$additive: $\mathbf{d} = \mathbf{G}_3+\varphi_1\mathbf{T}_1+\varphi_2\mathbf{T}_2$, $|\mathbf{d}|\neq 1$Rodrigues: $\mathbf{d} = \cos|\varphi|\,\mathbf{G}_3+\mathrm{sinc}|\varphi|(\varphi_1\mathbf{T}_1+\varphi_2\mathbf{T}_2)$, $|\mathbf{d}|=1$
Membrane strainlinear: $\tfrac{1}{2}(\mathbf{A}_\alpha\cdot\mathbf{u}_{,\beta}+\mathbf{A}_\beta\cdot\mathbf{u}_{,\alpha})$Green–Lagrange: $\tfrac{1}{2}(a_{\alpha\beta}-A_{\alpha\beta})$linear: $\tfrac{1}{2}(\mathbf{A}_\alpha\cdot\mathbf{u}_{,\beta}+\mathbf{A}_\beta\cdot\mathbf{u}_{,\alpha})$Green–Lagrange: $\tfrac{1}{2}(a_{\alpha\beta}-A_{\alpha\beta})$
Bending strain$\kappa_{\alpha\beta} = -u_{3,\alpha\beta}$ (flat ref.)$\kappa_{\alpha\beta} = b_{\alpha\beta}-B_{\alpha\beta}$$\tfrac{1}{2}(\mathbf{A}_\alpha\cdot\mathbf{d}_{,\beta}+\mathbf{A}_\beta\cdot\mathbf{d}_{,\alpha})-B_{\alpha\beta}$$\tfrac{1}{2}(\mathbf{a}_\alpha\cdot\mathbf{d}_{,\beta}+\mathbf{a}_\beta\cdot\mathbf{d}_{,\alpha})-B_{\alpha\beta}$
Transverse shear$\gamma=0$ (Kirchhoff)$\gamma=0$ (Kirchhoff)$\gamma_\alpha = \mathbf{A}_\alpha\cdot\mathbf{d}$$\gamma_\alpha = \mathbf{a}_\alpha\cdot\mathbf{d}$
Finite rotationsnoyessmall only ($|\varphi|\ll 1$)yes
C¹ for bendingyesyesnono
In FerriteShells_KL functions_RM functions

The key distinction between RM and Naghdi is which base vectors appear in the strain measures: RM uses the reference base vectors $A_\alpha$ (linearised around the reference configuration), while Naghdi replaces them with the current $a_\alpha$ everywhere, giving fully nonlinear strains. The director parametrisation (non-unit additive vs unit Rodrigues) is a separate but related choice — in practice the two always appear together.

Koiter has no director DOFs; the normal is always implicit from the surface geometry, so the Kirchhoff constraint (zero shear) is built in and C¹ continuity is required for bending.

Classical RM (additive director, $\|\mathbf{d}\|\neq 1$) is not implemented; the _RM functions go directly to the geometrically exact Naghdi form via Rodrigues parametrisation.