SolidNetics / Additive ManufacturingLPBF · Physics-AI Surrogates · CloudPlatform live
Physics-AI · Metal Additive Manufacturing
Print it right the first time.
SolidNetics is the cloud Physics-AI platform for metal additive manufacturing. Surrogates trained on thousands of high-fidelity simulations carry a part from scan path through melt-pool, microstructure and residual stress to a certification-ready report — then tune the process to print it faster.
Physics-AI surrogatesLaser Powder Bed FusionNo install
Scan path through to qualification, plus two internal surrogate engines.
5M
Element ceiling
Per part-scale residual-stress solve, on elastic cloud compute.
4
CAD formats in
STL, STEP, IGES and OBJ — repaired and face-tagged on import.
0
Local installs
Every solver runs in the browser. No licence server, no workstation.
§ 01
Learn once. Predict at part scale.
Simulate → Learn → Predict → Optimize
Full-physics AM simulation is accurate but far too slow to run over a whole part. SolidNetics trains physics-informed surrogate models on thousands of high-fidelity simulations, so part-scale residual stress, distortion and microstructure that once took days run in minutes — fast enough to close an optimization loop around.
01
Simulate the ground truth
FusionCore resolves melt-pool thermal history; GrainPath resolves solidification microstructure. This is the high-fidelity physics the AI learns from.
FusionCoreGrainPath
02
Learn the surrogate
FusionMap and GrainMap sweep thousands of those runs across the process window and fit physics-informed surrogate models to the results.
FusionMapGrainMap
03
Predict at part scale
StressForge evaluates the trained surrogates per voxel to predict residual stress and distortion across the entire part — no meso-scale solve in the loop.
StressForge
04
Optimize the process
Because prediction is now cheap, ProcessPilot can search the process window — cutting residual stress and distortion while holding build rate.
Every stage consumes the contract the stage before it emits — geometry to trajectory, trajectory to melt pool, melt pool to microstructure and residual stress, and all of it to a qualification verdict. No manual handoff, no re-exporting between tools.
S1 · Scan-Path GeneratorEnterprise
PathWeaver
Turns a part and a process strategy into a complete, time-stamped laser trajectory — the single contract every downstream AM solver reads. Slicing, region decomposition, multi-laser zoning, hatch generation and timing.
Placeholder
Fig. S1Stripe / island decomposition
Readspart geometry · process strategy
Emitstime-stamped laser trajectory
Slicing & region decompositionStripe / island / chessboardMulti-laser zoningHatch & timing
High-fidelity meso-scale thermal simulation of LPBF — resolving melt-pool geometry, solidification gradients and phase evolution along the trajectory PathWeaver laid down.
The high-throughput engine that sweeps FusionCore across the process window and trains the thermal surrogates StressForge runs on. Not exposed as a standalone tool — it is how the platform learns.
Predicts solidification microstructure — grain morphology, texture and columnar-to-equiaxed transition — from the thermal-gradient and solidification-velocity fields FusionCore emits.
Placeholder
Fig. S3Cellular-automaton grain growth
ReadsG & R fields · material
Emitsgrain morphology · texture · CET map
Grain morphologyColumnar ↔ equiaxedTexture from G & R
The microstructure counterpart of FusionMap: runs thousands of GrainPath simulations to train the surrogates that bring grain prediction to part scale. Not exposed as a standalone tool.
GIF loop
GrainMap — grain statistics across the process window
1600 × 1000 · ~6 s silent loop<br>grain-size map sweeping P–v space
Fig. Int-2Awaiting capture
Readsthousands of GrainPath runs
Emitstrained microstructure surrogate
High-throughput sweepsMicrostructure training dataSurrogate fitting
S4 · Residual StressEnterprise
StressForge
Part-scale residual stress, warping and distortion — evaluating the trained surrogates per voxel to drive an inherent-strain solve across the whole part, at up to 5M elements.
Consolidates every upstream prediction into defect-probability maps, property estimates and pass/fail gates — the certification-ready report that ends the chain.
Closes the loop. Drives the trained Physics-AI surrogates across the LPBF process window to minimise residual stress and distortion while maximising build rate — returning optimized, region-aware process parameters.
GIF loop
ProcessPilot — searching the LPBF process window
1600 × 1000 · ~8 s silent loop<br>Pareto front converging over P–v space
What actually changes when the whole chain lives in one cloud platform and the expensive physics has been learned once, up front.
Property
SolidNetics
Conventional AM CAE
scan_path
Generated in-platform from the part and strategy
Imported from the machine, or approximated
melt_pool
Resolved meso-scale, then learned as a surrogate
Re-solved every run, or skipped entirely
residual_stress
Part-scale in minutes, via surrogate inherent strain
Days of chained meso-to-macro solves
microstructure
Grain morphology and CET, predicted at part scale
Rarely coupled to the build simulation at all
element_ceiling
5M per part-scale solve, on elastic cloud compute
Whatever the workstation holds in RAM
geometry_in
STL, STEP, IGES, OBJ — repaired and face-tagged
STL, repaired by hand
qualification
Defect maps, property estimates and pass/fail gates
Assembled by hand from raw field output
optimization
Closed-loop search over the process window
Manual DOE, or trial builds on the machine
install
None — every solver runs in a browser tab
Workstation install and a licence server
updates
Continuous — you are always on the latest solver
Annual release, migrated by hand
§ 04
Where it has to be right the first time
6 sectors
The sectors betting real parts on metal AM — the failure mode each one is fighting, and what SolidNetics does about it.
#SectorFailure modeSolidNetics response
01
AerospaceBrackets · nozzles · turbines
Flight-critical parts must be fully traceable. Residual stress and hidden defects cause distortion and rejected builds, and qualifying by repeated trial prints is slow and expensive.
Predicts melt-pool, microstructure, residual stress and defect probability before printing, and produces certification-ready reports that shorten qualification.
02
Medical & ImplantsPatient-specific titanium
Every implant is a one-off geometry that cannot be trial-printed, yet it still has to meet exact porosity, fatigue and dimensional targets.
Simulates microstructure and residual stress for each patient geometry, holding mechanical performance and dimensional accuracy without iterative builds.
03
Tooling & MouldsConformal cooling
Conformal-cooling tools warp and crack under residual stress. A failed insert costs powder, machine hours and lead time on an expensive build.
Predicts distortion and residual stress up front so geometry can be pre-compensated — right-first-time tools, fewer rebuilds.
04
AutomotiveLightweight & EV
Performance and EV parts need fast iteration, but porosity and build failures stall the move from prototype to series production.
Surrogate-accelerated simulation screens process parameters in minutes, cutting scrap and speeding qualification at production rate.
05
EnergyHeat exchangers · turbines · nuclear
Components run in extreme thermal and pressure conditions where a defect or a residual-stress hotspot threatens structural integrity.
Generates defect-probability maps and residual-stress fields so integrity and long-term reliability can be confirmed before the build.
06
Research & AcademiaProcess–structure–property
Process–structure–property studies need high-fidelity thermal and microstructure data that is slow and expensive to produce experimentally.
Cloud physics and surrogate solvers generate G/R, microstructure and residual-stress datasets at scale — no lab time, no workstation.
§ 05
From CAD to a qualified build
Three steps · entirely in the browser
No installation, no licence server, no hardware to provision.
Step 01
Upload & set the process
Import a CAD model and define the LPBF process strategy — orientation, laser parameters, scan strategy — through surface-based selection with automatic normal detection.
Step 02
Generate on the cloud
Scan path and simulation domain are built on scalable cloud infrastructure. No local hardware, no meshing bottleneck, no overnight queue on a shared box.
Step 03
Run & qualify
Solve melt-pool, microstructure and residual stress across CPUs and GPUs, then explore defect maps and qualification results online.
Also in SolidNetics
The Core physics module
Beyond additive: SolidNetics started as a general solid-mechanics platform, and those solvers still ship with every account — finite elements, physics-informed neural networks and peridynamics, on the same cloud infrastructure.
Predict melt-pool, microstructure, residual stress and defect risk before you commit powder and machine time — and ship certification-ready reports from a single cloud platform.
Placeholder
