This project aims to enhance simulation models at Danfoss Drives by developing data-driven components. High-fidelity physics-based models require substantial computational resources, which hinders their commercial viability and real-time application as digital twins. To address this, the project will: 1) develop methods for detecting bottlenecks or points that slow down simulations for a given model, and 2) develop data-driven surrogate models to reduce computational demands. These advancements will enable real-time, high-fidelity simulations, facilitating true digital twins and offering significant commercial benefits.