Flux e-Machine Toolbox improvements

FEMT 2022.1 improvements

FEMT improvements:
  • Switch from User to Dynamic numerical memory mode and vice versa inside FeMT
  • Distributed computing available on Linux
  • Run tests through PBS (by manual Export / Import)
  • New extrapolation approach for efficiency maps

Switch from User to Dynamic numerical memory mode and vice versa inside FeMT

It is now possible to change the numerical memory mode used to run the test inside FeMT:
  • Either switching from User numerical memory mode to Dynamic numerical memory mode by setting the Numerical Memory (MiB) to 0
  • Or switching from Dynamic numerical memory mode to User numerical memory mode by setting the Numerical Memory (MiB) to a fixed amount of memory
    Warning: When switching from Dynamic numerical memory mode to User numerical memory mode, the user must know the numerical memory required to solve the project and therefore set a high enough value.


Note: With previous Flux versions, in order to change the numerical memory mode, the user had to regenerate the FeMT coupling component in Flux running in the chosen numerical memory mode.

Distributed computing available on Linux

On Linux, it is now possible to run a test by activating the local distribution.

First, you have to make sure that the distribution is properly configured.

See Parametric distribution in Flux (Linux) for more information.



Note: With previous Flux versions, only local distribution on Windows (via CDE) was possible.

Run tests through PBS (by manual Export / Import)

To save computation time, it is now possible to run tests through PBS.

Once the project is correctly defined, the different steps are:
  1. In FeMT: Export the required files for PBS
  2. In PBS: Run the computation through a cluster
  3. In FeMT: Import the results from PBS

And then postprocess the results as usual.



New extrapolation approach for efficiency maps

There is now a new map extrapolation method to obtain correct values near map borders and thus avoid obtaining non-physical values, such as negative Irms values.