Case Study: Convion cuts design cycles from months to under an hour with SimScale Physics AI

Convion reduces design cycles from months to under 1 hour with SimScale

Convion Ltd, a Finnish developer of solid oxide fuel cell and electrolyzer systems, faced a significant challenge in optimizing a complex fluidic device for its high-temperature hydrogen production systems. Using traditional CFD-driven design optimization was a slow process, with a single iteration taking hours and a full exploration of the design space requiring months. Convion needed a faster way to recover process gases and maximize flow versus pressure recovery while also optimizing for new packaging constraints.

By adopting SimScale's Physics AI-driven optimization process, Convion compressed its design cycle from months to under one hour. The solution involved running a large-scale design of experiments campaign on SimScale to train an AI surrogate model, which could then predict the performance of new design variants in milliseconds. This allowed Convion to identify a non-intuitive, high-performing geometry that met all targets while occupying 50% less volume. The validated AI model was deployed as a shared tool, enabling the wider engineering team at Convion to instantly test and verify new geometries.

Convion Ltd