Sumitomo Riko speeds automotive component design with Ansys AI
Pittsburgh, PA - Sumitomo Riko is deploying Ansys SimAI (part of Synopsys, Inc., NASDAQ: SNPS) to accelerate time-to-solution and improve efficiency across the design and manufacturing of high-performance automotive rubber components.
SimAI learns from new or legacy simulation data to create AI models that predict performance. The company is applying it to computation-heavy work such as anti-vibration design and exploration, battery cooling, magnetic field analysis, and mixing heat transfer analysis.
Why it matters for product development
Designing reliable rubber components requires hundreds of multiphysics simulations. Pre-processing steps-like defining geometric parameters-are slow and demand specialist skills.
By training AI models directly from historical simulation data without parameterizing geometry, teams can move faster while keeping accuracy close to high-fidelity simulation.
How Sumitomo Riko is doing it
Engineers are training SimAI models on past simulation datasets for products like vibration isolators and hoses. These models generate performance predictions in under five minutes, saving over an hour per new design while maintaining comparable accuracy to detailed FEA.
In one example, a high-fidelity Ansys Mechanical analysis of rubber bush straining was approximated by SimAI in about five minutes-enabling much faster iteration without reconfiguring model parameters.
Early results
Initial testing showed more than a 10x speedup for predicting mechanical performance of specific rubber bushes used in suspension systems. Faster turnaround allows tighter design loops and more informed tradeoffs before prototyping.
Leadership emphasized two priorities: automating workflows across the product lifecycle and driving broader AI adoption across teams. Removing the need for parameterized geometries makes collaboration easier across disciplines. They also noted that high-quality data combined with simulation is key to balancing speed and accuracy in early design.
Where it applies
- Anti-vibration design and design space exploration
- Battery cooling and thermal management
- Magnetic field analysis
- Mixing and heat transfer analysis
Practical steps for product teams
- Start with a focused use case where simulation is a bottleneck (e.g., a repeated mechanical analysis with long pre-processing).
- Leverage historical simulation runs to train initial AI models; document boundary conditions, materials, and meshing choices.
- Validate against trusted high-fidelity results with a holdout set; track error metrics that matter to your acceptance criteria.
- Integrate into your existing CAE workflow (pre/post, PLM, and design review checkpoints) to avoid disrupt-ing downstream processes.
- Set guardrails: define the model's valid input range, auto-flag out-of-distribution inputs, and require spot-checks for critical loads.
- Measure ROI: time saved per iteration, throughput increase, and rework reduction from earlier decision-making.
What this signals for PD leaders
AI surrogates trained on your own simulation data can compress iteration cycles without forcing teams to re-parameterize complex geometries. The outcome: faster concept loops, earlier risk reduction, and better data for tradeoff decisions-especially in areas like vibration, thermal, and electromagnetic analysis.
Learn more
- Ansys AI overview
- Build AI skills by job role for your product and engineering teams
Your membership also unlocks:
