GNN Surrogate Modeling for Crash Simulation

An applied ML research workflow for predicting mesh deformation, acceleration, and safety-relevant HIC15 from time-dependent CAE simulations.

Industrial AI / Graph Neural Networks / Scientific ML

An applied GNN surrogate-model research and engineering workflow for crash simulation prediction, evaluation, and validation.

CAE InputSimulation MeshLarge-scale mesh snapshots
PrepGraph RepresentationSpatial & historical features
ModelGNN SurrogatePyTorch / Commercial toolkit
OutputsDeformation & AccelerationAutoregressive rollout
SafetyHIC15 EvaluationInteractive diagnostics

Anonymized technical surrogate-model architecture overview

Overview

High-fidelity crash simulations generate large, graph-structured mesh datasets across multiple timesteps. This project investigated whether GNN-based surrogate models could provide faster approximate predictions of deformation and acceleration while preserving the engineering signals required for safety assessment.

My work focused on the applied ML and research workflow around the surrogate model rather than replacing the underlying physics solver. The project was executed as an industrial CAE collaboration with engineering software collaboration.

Why this was difficult

01

Imbalanced Node Dynamics

Most mesh nodes showed little deformation, allowing generic losses to be dominated by easy far-field regions while missing high-impact localized gradients.

02

Localized Safety Signals

Safety-relevant acceleration was concentrated in a small subset of nodes around the impact zone, making standard mean error metrics misleading.

03

Downstream Metric Sensitivity

Acceleration peaks strongly affected downstream HIC15 calculations, where minor trajectory smoothing caused catastrophic safety misclassification.

04

Autoregressive Error Accumulation

Multi-timestep rollout introduced error accumulation and initial-state mismatch as previous predictions were fed back into subsequent GNN steps.

05

Deceptive Aggregate Loss

A model could perform well on aggregate validation loss while still failing around the critical impact zone during engineering evaluation.

06

Multi-Objective Conflict

Different physical outputs required different evaluation and optimization strategies, as joint optimization of position and acceleration degraded peak accuracy.

My contribution

Data and workflow engineering

Built preprocessing and export workflows for graph-structured simulation data, covering mesh and metadata ingestion, feature attachment, dataset preparation, prediction remapping, and unified analytics outputs.

Model development

Designed spatial and historical features, custom graph-loss variants, and autoregressive rollout workflows for deformation and acceleration prediction using PyTorch and a commercial graph-learning toolkit.

Engineering evaluation

Developed evaluation logic beyond generic validation loss, including directional deformation errors, critical-region analysis, acceleration trajectories, HIC15 error, and safety-oriented diagnostics.

Experiment and diagnostic tooling

Automated repeatable experiment workflows and built interactive dashboards for comparing model trials, inspecting mesh-level errors, and analyzing deformation, acceleration, training behavior, and HIC outcomes.

System workflow

  1. 01Prepare simulation data

    Parse mesh and metadata, validate snapshots, decimate elements, and attach model features.

  2. 02Build graph inputs

    Construct node, edge, global, spatial, and historical features from CAE mesh topology.

  3. 03Train surrogate models

    Run configurable GNN experiments and custom physics-aware loss variants.

  4. 04Roll predictions forward

    Feed predicted deformation and acceleration back into subsequent timesteps autoregressively.

  5. 05Evaluate engineering outputs

    Compare deformation, acceleration trajectories, critical regions, and HIC15 safety values.

  6. 06Diagnose and iterate

    Use Streamlit dashboards and trial comparisons to isolate error propagation and failure patterns.

Technical decisions

Decision 01

Spatially aware features

spatial-feature experiment
Problem
Prediction errors were concentrated around localized impact regions while far-field mesh nodes dominated the loss calculation.
Decision
Incorporated spatial features representing the topological and geometric relation between each mesh node and the impact location.
Result
Reduced combined acceleration and deformation MAE by approximately 15% in corresponding feature experiments.
Decision 02

Task-specific loss design

custom-loss experiment
Problem
A standard uniform loss allowed the large number of non-deforming nodes to dominate training and weakened supervision of localized peak responses.
Decision
Separated deformation and acceleration supervision, applying higher loss weighting to localized high-response impact zones while maintaining lower weighting on far-field nodes.
Result
Reduced HIC prediction error by 27% in the evaluated custom-loss experiment.
Loss Weighting Comparison / Bonnet Mesh Silhouette
Standard LossUniform MSE
All weights = 1.0
Equal weights → Local error diluted
Weighted LossLocalized Loss Function
w=0.1w=0.1w=1.0 (impact zone)
Higher weights on impact zone → HIC error ↓27%
Decision 03

Autoregressive timestep rollout

selected rollout experiment
Problem
Treating each timestep independently failed to model temporal state transitions, while multi-step rollouts risked error propagation.
Decision
Implemented an autoregressive rollout workflow with recurrent state updates and targeted analysis of initial-state mismatch.
Result
In a selected rollout experiment, deformation MAE decreased by 13% and X-direction deformation R² improved from 0.75 to 0.86.

Results

These results come from separate experiments and model configurations. They should be interpreted individually, not as the performance of one combined final model.

Selected spatial-feature experiment~15% reductionCombined acceleration & deformation MAE
Baseline
vs. baseline feature configuration
Custom loss-weighting experimentHIC error ↓27%Safety metric relative error
Baseline
vs. joint loss baseline
Selected autoregressive rollout experiment0.75 → 0.86Deformation R² score (X-direction)
Baseline
vs. baseline rollout setup
Selected autoregressive rollout experimentDeformation MAE ↓13%Timestep rollout error
Baseline
vs. baseline rollout setup

Anonymized Model Diagnostics

I built interactive Streamlit, Plotly, and PyVista diagnostics to replace manual comparison across simulation files, prediction exports, training logs, and engineering plots.

Streamlit / Plotly / PyVista. Conceptual reconstruction of the implemented analysis workflow.

  • Truth vs. prediction acceleration curves
  • Deformation error by direction and timestep
  • Safety metric calibration and relative-error distribution
  • Over-prediction vs. under-prediction analysis
  • Trial and loss-function comparison
  • 3D mesh-level error visualization using PyVista
  • Impact-region and critical-deformation analysis
Anonymized Model Diagnostics Console
MESH DEFORMATION / TRUTH VS PREDICTION
Truth MeshImpact Zone
GNN PredictionImpact Zone
ACCELERATION TRAJECTORY (NORMALIZED)
1.0g0.0g
Peak Response
0ms50ms100ms
CAE Truth GNN Prediction
Deformation R² (X)0.75 → 0.86Selected rollout experiment
HIC Error↓ 10%Selected rollout experiment

Lessons & takeaways

  • Aggregate ML metrics can hide engineering-critical failure modes in localized impact zones.
  • Different physical outputs require different loss and evaluation strategies rather than a naive joint loss.
  • Rollout models must be evaluated for error propagation over time, not only single-step accuracy.
  • Interactive 3D & curve diagnostics were essential for connecting ML results with engineering interpretation.

Links

Next project03EV Charger DetectionIndustrial AI / Computer Vision

Project image

Interactive demo