Abstract
Recent years have seen growing application potential for Lattice-skin Plate Structures in advanced manufacturing fields such as aerospace and automotive engineering. For multiscale performance evaluation of such structures, conventional homogenization methods for lattice-filled volume structures are often used for equivalent analysis. However, in finite-thickness Lattice-skin Plate Structures, periodic boundary conditions imposed along the three orthogonal directions of the representative cell cannot adequately capture the boundary effect of the free surfaces in the thickness direction, which introduces bias into the prediction of effective properties.
To reduce this bias, this study develops and open-sources a homogenization method for Lattice-skin Plate Structures, forming an open-source computational framework for this class of structures. Representative numerical examples show that the framework can stably extract effective plate/shell stiffness matrices and can be extended to predict multiphase material properties and analyze steady-state heat conduction. The tool provides an open and reusable analysis foundation for the high-fidelity design of multifunctional lightweight structures.
Framework
The companion page follows the manuscript and uses the original figures from the paper. The central idea is to replace direct three-dimensional simulation of a finite-thickness lattice-skin structure with an equivalent plate or shell model while preserving the physically meaningful membrane, bending, and coupling responses captured by the ABD stiffness matrix.
Plate and shell reduction
The method targets finite-thickness lattice-skin structures whose mechanical behavior is more faithfully described by equivalent plate and shell kinematics than by volume-only periodic homogenization.
Loop-free tensor mapping
The implementation relies on multidimensional tensor slicing to construct local-to-global mappings efficiently and keep the core homogenization routine compact.
Reusable computation pipeline
The same framework supports stiffness extraction, multiphase material settings, and steady-state heat conduction analysis in a consistent open-source workflow.
edofMat based on multidimensional tensor slicing.
Code
The manuscript provides a compact 99-line implementation. The following excerpt highlights the stress recovery and ABD extraction routine, where np.einsum is used to avoid explicit loops over tensor contractions.
# Core computation: microscopic stress recovery and effective ABD extraction
ABD = np.zeros((6, 6))
# Accumulate over the eight hexahedral Gauss points
for i_gp in range(8):
# Physical stress recovery: Sigma = C * (E_macro - B * U)
Sigma = np.einsum('ij,kjl->kil', C, E_macro -
np.einsum('ij,kjl->kil', Bs[i_gp], U[edofMat, :]))
# Zeroth-order in-plane integration for the A and B matrices
ABD[0:3, :] += np.sum(Sigma[:, [0, 1, 5], :], axis=0) * detJ / plate_area
# First-order integration through the thickness for the D matrix
ABD[3:6, :] += np.sum(
Sigma[:, [0, 1, 5], :] * z_active[:, None, None],
axis=0
) * detJ / plate_area
# Symmetrization to satisfy Maxwell-Betti reciprocity
ABD = (ABD + ABD.T) / 2.0
return ABD
Results
The paper validates the framework with representative thin-walled metamaterial cases, including TPMS and BCC sandwich structures, thickness size effects, multimaterial settings, and static-response comparisons against full-scale finite element models.
Citation
If this code or the homogenization framework is useful in your research, please consider citing the paper.