Physics-AI Surrogates: FusionMap and GrainMap

Two surrogate-training engines now power part-scale prediction. FusionMap sweeps thousands of high-fidelity FusionCore thermal runs across the process window and learns them into fast thermal surrogates; GrainMap does the same for GrainPath microstructure.

Together they realise the learn-once, predict-at-part-scale approach — the trained models drive StressForge and GrainPath predictions in minutes, with no meso-scale solve in the loop. FusionMap and GrainMap run as internal training engines and are not exposed as standalone tools.

Part-Scale Residual Stress with StressForge

StressForge completes the additive-manufacturing pipeline with part-scale residual stress, warping, and distortion prediction for LPBF builds. It reuses the geometry and voxel mesh from a linked PathWeaver solution and an inherent-strain (ε*) field from a FusionCore, GrainPath, or Surrogate source — with no meso-scale solve left in the loop.

Simulations run on AWS Batch across S–XL compute sizes, with progressive layer activation and an optional build-plate release for springback. Results include von Mises and principal stresses, directional components, distortion, yield utilization, recoater-crash risk, and build-plate uplift.

Solidification Microstructure with GrainPath

GrainPath predicts as-built microstructure using a cellular-automaton grain solver run over representative volume elements (RVEs) sampled from a completed FusionCore thermal field.

Each RVE resolves grain morphology and crystallographic texture, including columnar-to-equiaxed transitions driven by the local thermal gradient and solidification velocity. The overview maps grain diameter, columnar fraction, and thermal cluster across the part; clicking an RVE reveals its full grain structure in inverse-pole-figure colour.

Meso-Scale Melt-Pool Thermal Simulation with FusionCore

FusionCore introduces high-fidelity meso-scale thermal simulation of LPBF — resolving the moving melt pool, solidification gradients, and phase evolution across multi-layer builds on graded meshes.

Laser and process parameters accept comma-separated lists for parametric sweeps, and each thermal solve auto-chains microstructure, residual-stress, and defect-risk predictions. Results include the temperature time series, melt-pool width and depth, and thermal-gradient and cooling-rate fields.

LPBF Scan-Path Generation with PathWeaver

PathWeaver launches the additive-manufacturing pipeline, generating optimised laser scan strategies directly from part geometry — build orientation, layer slicing, and hatching for bulk, contour, and up- and down-skin regions.

The output is a time-stamped multi-track, multi-layer scan strategy plus a voxelised per-region plan (bulk, contour, up-skin, down-skin) that feeds the downstream thermal, microstructure, and residual-stress solvers as surrogate inputs.

Rigid Impact Simulation with Peridynamics

Rigid Impact with Peridynamics has been introduced to simulate interactions between a rigid impactor and a deformable body. This analysis type is designed for high-speed, transient impact scenarios.

The formulation uses an explicit time integration scheme, making it well-suited for dynamic contact and impact problems where inertia and wave propagation effects are dominant.

Multiple 3D impactor geometries

SolidNetics now offers multiple predefined three-dimensional impactor shapes for rigid impact simulations. Available geometries include sphere, cuboid, ellipsoid, cylinder, and capsule. These options allow users to model a wide range of realistic impact scenarios without needing custom geometry, accelerating setup and analysis.

Predefined Crack Support for Peridynamic Analysis

SolidNetics now includes predefined crack capabilities for peridynamic fracture analysis. Users can initialize cracks within the model to study crack growth, interaction, and failure mechanisms under applied loading.

This feature provides additional control for fracture studies while retaining the advantages of peridynamics in handling discontinuities.

Fracture Analysis with Peridynamics (PD)

Fracture Analysis with Peridynamics has been added to SolidNetics, leveraging the inherent strengths of peridynamic theory for crack initiation and propagation modeling. This capability avoids the need for predefined crack paths or remeshing.

The analysis supports incremental loading strategies suitable for quasi-static fracture simulations and uses an implicit solver to ensure numerical stability and accuracy.

Stress Analysis with Physics-Informed Neural Networks (PINNs)

SolidNetics now enables stress analysis using Physics-Informed Neural Networks (PINNs) as an alternative to traditional finite element methods. Users can configure key network parameters including the number of layers, neurons per layer, and weighting factors for PDE residuals and boundary condition loss terms.

PINN simulations are executed on high-performance AWS GPU instances, allowing efficient training and solution of complex three-dimensional elasticity problems directly in the cloud.

Multi-Run Project Management and Execution Control

Projects in SolidNetics now support multiple simulation runs under a single project configuration. Each run is tracked independently, and post-processing can be performed on any selected run.

A dedicated Status section lists all runs and provides real-time access to run status, execution logs, and management actions. Users can monitor progress, terminate active runs, or delete completed runs as needed.

Pre-Run Mesh and Boundary Condition Visualization

SolidNetics now provides pre-run visualization of the computational mesh, including nodes and elements, prior to launching a simulation. Boundary conditions applied at the nodal level are also displayed graphically.

This capability enables users to verify mesh quality and confirm boundary condition assignments before execution, reducing setup errors and improving overall model reliability.

Expanded CAD Support: STEP and IGES

In addition to STL files, SolidNetics now supports industry-standard CAD formats including STEP and IGES. This enhancement improves interoperability with mainstream CAD tools and reduces the need for external geometry conversion.

Imported CAD models are automatically processed and prepared for meshing and analysis, allowing users to transition seamlessly from design to simulation.

Cloud-Native SolidNetics Platform on AWS

SolidNetics is now fully deployed on Amazon Web Services (AWS) as a cloud-native engineering simulation platform. The software operates entirely in the cloud, eliminating local installation and enabling scalable compute resources for demanding solid mechanics analyses.

Users can upload CAD geometry in STL format, generate meshes using the built-in mesh generator, execute simulations on cloud compute instances, and perform interactive post-processing directly within the platform. This end-to-end workflow streamlines model setup, execution, and visualization in a single environment.