Classical surrogate models in engineering simulation loop
LiveCorehigh effect
Core capability
These methods replace expensive simulation runs with fast approximations, helping engineering teams evaluate more options in less time while keeping model behavior relatively interpretable.
How it works
The team runs a limited number of accurate simulations at carefully chosen points, and the system turns those results into a fast approximation model that can estimate many similar cases almost instantly.
Application here
Proven mathematical approximation methods estimate simulation results quickly within a defined parameter range.
Business impact
These models provide reliable, well-understood approximations and are often a practical starting point for engineering optimization.
Limitations
They become less efficient as the number of variables grows, and predictions degrade sharply outside the trained range.
In production
This is already a reliable production technique for speeding up design studies and reducing the number of expensive solver runs required.
Research
The next step is fast models that stay current automatically as new evidence comes in, reducing the need to rebuild the whole approximation each time conditions change.
Examples
Ansys optiSLang is the industry standard for classical surrogates (kriging, RSM, polynomial chaos expansion) in the engineering loop: after DOE runs the surrogate accelerates optimisation 10–100×. Used at Porsche and Rolls-Royce for multi-objective optimisation (Ansys customer stories) — .