During development
Check design changes quickly.
Change a magnet, lamination, or winding and get an estimate in seconds. Use detailed simulations for the strongest options.
AI modeling
Turn simulation data into faster decisions.
Your existing simulations and test results can train an AI model. It estimates how new designs will perform in seconds, so you can compare more options before running detailed simulations.
AI models
Get answers in seconds, not hours.
Simulations show how a design will perform, but each run can take hours. We use your existing simulation data to train an AI model.
Once trained, the model estimates results in seconds. Detailed simulations are then used to check the most promising designs.
The current process
Every new design starts another long simulation.
Simulation results are often used once and then stored. When the design changes, engineers run another simulation and wait again.
Today
Limited simulation time limits the search.
Each run needs time and computing power. Preparing and maintaining the parametric model also takes engineering work. Teams can therefore simulate only a small number of designs.
Other ideas are never compared, even when they could lead to a better result.
Use the result of every simulation again.
A trained model uses earlier results to estimate new designs in seconds. Engineers can compare more ideas before using the solver.
Full simulations are used for the most promising designs. They still provide the final engineering check.
With a model
Electric drives
Why this works well for electric motors.
Evaluating a motor requires electromagnetic, thermal, and mechanical simulations at several operating points. Each design can take hours to check.
Many projects ask similar questions with different targets, sizes, or materials. An AI model can reuse what earlier simulations have already shown.
During design search
Compare thousands of designs.
The model ranks many designs quickly. Only the most promising ones need a full simulation.
For product enquiries
Give customers a faster first answer.
A model trained on your product range can estimate feasibility, size, and performance before a detailed engineering study begins.
How it works
Build the model in four steps.
We start with the simulation and test data you already have. If important data is missing, we generate only what is needed.
- 01
Review existing data
We review your simulation results, design studies, and test measurements. We check what can be used and what is missing.
- 02
Generate missing data
If important areas are not covered, our automated process creates and simulates additional designs.
- 03
Train and validate the model
The model learns from the data. We then test its accuracy on designs that were not used for training.
- 04
Use it in your process
The model compares designs or provides fast estimates. Detailed simulations confirm the selected designs.
What you get
Compare more designs with less simulation time.
Results depend on your data, the designs you want to compare, and the accuracy you need. The figures below are project targets, not guaranteed results.
100×
faster estimates
Estimate results with the AI model before running a detailed simulation.
10×
more designs
Compare more design options and operating points in the same project.
50%
fewer full simulations
Use detailed simulations for selected designs instead of every early idea.
Reuse every result
Each new simulation becomes additional training data. The model can improve as the dataset grows.
Start a project
Find out what your data can support.
Tell us what you want to estimate and what data you already have. We will assess what can be built and which additional data is needed.
Book a call