[Dataset] Solver-Agnostic Implementation of Atom-Informed Thermal Conductivity Fields in Continuum Heat-Flow Simulations

SCACS and Abaqus data files This dataset contains the SCACS point-data files and corresponding Abaqus input files for the three silicon models investigated in the associated publication. It also includes the Python utility Paraview_to_Abaqus.py for processing point data exported from ParaView and converting the corresponding VTU mesh to the Abaqus INP format. The files are organized into the following folders: SCACS/: SCACS point-data files and the data-conversion utility. ABAQUS/: Converted Abaqus input files for steady-state and transient heat-transfer simulations using isotropic and anisotropic thermal-conductivity fields. The text-based files can be opened using any text editor. Visualization and post-processing of the SCACS files are best performed using ParaView, whereas visualization and execution of the Abaqus models require Abaqus/CAE. Model descriptions M1: Twin-grain-boundary silicon nanowire. M2: Amorphous–crystalline silicon interface. M3: Silicon nanopillar structure containing four pillars. SCACS point-data files The SCACS data are stored as XDMF/HDF5 file pairs. The .xdmf file contains the metadata required to describe the mesh and associated fields, while the corresponding .h5 file contains the numerical data. Both files should be retained in the same directory. M1.xdmf and M1.h5: Twin-grain-boundary silicon nanowire. M2.xdmf and M2.h5: Amorphous–crystalline silicon interface. M3.xdmf and M3.h5: Four-pillar silicon nanopillar structure. Abaqus input files Model M1 Mesh size: 1,682,384 elements and 162,000 nodes. M1_Dirichlet_SS_Aniso.inp: Steady-state solution with anisotropic conductivity. M1_Dirichlet_Transient_Aniso.inp: Transient solution with anisotropic conductivity. M1_Dirichlet_SS_Iso.inp: Steady-state solution with isotropic conductivity. M1_Dirichlet_Transient_Iso.inp: Transient solution with isotropic conductivity. Model M2 Mesh size: 5,417,280 elements and 511,584 nodes. M2_Dirichlet_SS_Aniso.inp: Steady-state solution with anisotropic conductivity. M2_Dirichlet_Transient_Aniso.inp: Transient solution with anisotropic conductivity. M2_Dirichlet_SS_Iso.inp: Steady-state solution with isotropic conductivity. M2_Dirichlet_Transient_Iso.inp: Transient solution with isotropic conductivity. Model M3 Mesh size: 462,372 elements and 504,621 nodes. M3_Dirichlet_SS_Aniso.inp: Steady-state solution with anisotropic conductivity. M3_Dirichlet_Transient_Aniso.inp: Transient solution with anisotropic conductivity. M3_Dirichlet_SS_Iso.inp: Steady-state solution with isotropic conductivity. M3_Dirichlet_Transient_Iso.inp: Transient solution with isotropic conductivity. ParaView-to-Abaqus data-conversion utility The Python utility Paraview_to_Abaqus.py, provided in the SCACS/ folder, processes SCACS point data exported from ParaView. It accepts comma-separated (.csv) and text (.txt) files and automatically detects comma, tab, or whitespace delimiters. The utility also: removes duplicate nodal entries; extracts nodal identifiers and coordinates; extracts the available conductivity components; optionally normalizes or scales the conductivity fields; optionally scales the nodal coordinates; exports individual and combined text files; provides three-dimensional visualization using PyVista; and converts a corresponding VTU mesh to the Abaqus INP format using MeshIO. The point-data processing and mesh conversion are separate operations. The VTU-to-INP conversion transfers the supported mesh geometry and element connectivity to the Abaqus input-file format. The conductivity values extracted from the CSV or TXT file are exported separately and are not automatically inserted into the generated INP file as spatially varying material properties. Expected point-data fields The input point-data file may contain the following ParaView fields: Point ID: nodal identifier; _0, _1, and _2: nodal coordinates; k_xx_used: xx component of the thermal-conductivity tensor; k_yy_used: yy component of the thermal-conductivity tensor; k_zz_used: zz component of the thermal-conductivity tensor; and k_used: optional scalar conductivity associated with the selected loading direction. During processing, these fields are renamed as follows: Point ID → ID _0 → X _1 → Y _2 → Z k_xx_used → k_xx k_yy_used → k_yy k_zz_used → k_zz ParaView may export multiple entries for a node shared by several elements. The utility groups the data using the nodal identifier and retains one coordinate and conductivity entry for each unique node. If an explicitly labeled nodal-identifier column is unavailable, the first column in the input file is used as the identifier. Normalization and scaling For visualization, each conductivity component can optionally be normalized by its maximum value: k_ii_normalized(x_a) = k_ii(x_a) / max[k_ii(x_b)] where i is x, y, or z, and the maximum is evaluated over all nodes. Each component is normalized independently. This preserves its spatial distribution but does not preserve the relative magnitudes of k_xx, k_yy, and k_zz. Independently normalized fields should therefore be used primarily for visualization and should not be interpreted as the original physical conductivity tensor. The directional conductivity components can also be multiplied by a user-defined scaling factor: k_ii_scaled(x_a) = s_k k_ii(x_a) If normalization and scaling are both selected, scaling is applied to the normalized field: k_ii_output(x_a) = s_k k_ii(x_a) / max[k_ii(x_b)] The nodal coordinates may be scaled independently according to: x_a_scaled = s_x x_a Coordinate scaling can be used to convert the model dimensions to the unit system employed in Abaqus. Generated output files The utility can generate individual text files for the following available quantities: ID.txt X.txt Y.txt Z.txt k_xx.txt k_yy.txt k_zz.txt k_used.txt Each individual file contains a single column without a header and is written to ten decimal places. Filename suffixes indicate whether normalization, conductivity scaling, or coordinate scaling was applied. A combined tab-delimited file containing the nodal identifiers, coordinates, and all available conductivity fields can also be generated. The combined file includes column headers and is intended for inspection, visualization, and post-processing. Three-dimensional visualization When all three coordinate fields are available, the processed data can be visualized using the interactive PyVista viewer included in the utility. The nodal coordinates are represented as a three-dimensional point cloud, and the available conductivity quantities are assigned as point-associated scalar fields. A tetrahedral volume is reconstructed from the point cloud using three-dimensional Delaunay triangulation. The exterior surface is colored according to the selected conductivity field, and interactive controls allow the displayed field to be switched among k_xx, k_yy, k_zz, and k_used. The Delaunay reconstruction is intended only for visual inspection. Because it reconstructs connectivity from the nodal coordinates, it may not reproduce the original finite-element connectivity for nonconvex geometries, internal voids, or complex interfaces. The original connectivity is retained in the corresponding VTU mesh. VTU-to-INP conversion For an input data file named model.csv or model.txt, the utility searches for a corresponding file named model.vtu or model.VTU in the same directory. MeshIO is then used to convert the supported mesh geometry and element connectivity to model_mesh.inp. Successful conversion depends on whether the cell types contained in the VTU file are supported by MeshIO and have corresponding Abaqus element definitions. Software requirements The data-conversion utility requires: Python 3 Pandas PyVista MeshIO ParaView for visualization and post-processing of the original SCACS data Abaqus/CAE for visualization, modification, and execution of the Abaqus models