Material Identification
The Material Identification tool is based on the Altair Compose environment and allows determining from measurements the parameters and coefficients required to create a material in Flux.
Introduction
- Non-linear magnet described by HcB, HcJ and Br module;
- Isotropic analytic saturation + knee adjustment;
- Isotropic hysteretic, Jiles-Atherton model;
- Isotropic hysteretic, Preisach model described by 4 parameters of a typical cycle and
- Isotropic hysteretic, Preisach model identified by N triplets.
- LS model and
- Modified Bertotti model.
- Isotropic splines for mechanically cut lamination steel.
To help users in these tasks preceding the creation of a material, Flux provides a unified Material Identification tool based on the Altair Compose environment.
- How to run the Material Identification tool.
- Using the Material Identification tool.
- Specificities of the identification modules.
How to run the Material Identification tool
In Flux Supervisor, at the bottom left part of the window, the button Material Identification allows the user to launch the provided tool dedicated to the parameter identification of a magnetic material model.
Please remark that the Material Identification module requires the Altair ComposeTM environment to be executed. Consequently, the user must install both Altair FluxTM and Altair ComposeTM on his computer to perform a material identification with this tool. Both programs are available for download at Altair One.
- click on the Supervisor's Options button and then select Coupled software under Access paths.
- then set the Compose environment script path to the file Compose.bat in the Compose installation folder found in Windows systems. The path to this file should be similar to Compose_Installation_Folder\Compose\ in the case of a standard installation. Alternatively, in Linux systems, set that path to file Compose, which is available in a path similar to Compose_Installation_Folder\scripts.
Using the Material Identification tool
For most of the materials models in the Flux Material Identification tool, the identification consists of a 3-step procedure. The general workflow is described below:
- Step 1: After choosing a specific type of B(H) magnetic property or an iron losses model, the Flux Material Identification tool will ask the user for a file containing magnetic measurements representing the behavior of the material subjected to identification. The input data required by the Flux Material Identification tool is given by a .csv (in Windows and Linux) or a .xlsx (Windows only) file containing magnetic measurements. The file content and format depend on the specific kind of model being identified, as detailed in the next section.
- Step 2: Once the file is correctly loaded, the identification algorithm launches automatically and finds the best model parameter set fitting the selected model. The results are displayed automatically in the graph window, as shown in Figure 2 below. The red lines represent the reconstructed behavior provided by the identified model, while the blue lines correspond to the source measurement file. Depending on the model, the user may consider adjusting the parameters iteratively with the sliders on the left side of the panel.
- Step 3: Another feature of this panel is the possibility to export the pyFlux command containing the identified model parameters of the material under identification. This export is achieved by clicking on the button Save pyFlux. The pyFlux command will be directly printed in the Compose console and may be copied and pasted in Flux's console, leading to an automatic creation of the material in a Flux project. This action also creates a python file containing the pyFlux command in a directory directly chooses by the user. With the help of this file, the material may be alternatively created in a Flux project by clicking on: Project > Command file > Run a python file.
Specificities of the identification models
In the previous section, a general workflow for the utilization of the Material Identification tool was presented. However, since the modules integrating this tool are all different, specific remarks apply for each identification case.
For the Non-linear magnet described by HcB, HcJ and Br module model, the input file must contain the magnet's B(H) curve (2nd and 3rd quadrants) and be provided either in .csv or a .xlsx formats. Two columns are required in the file: the first for the magnetic field H in amperes per meter and the second for the magnetic flux density B in teslas. An example input file is available here.
In the case of the Isotropic analytic saturation + knee adjustment model, the input file containing B(H) data has the same file format described in the previous paragraph, but represents instead the anhysteretic curve or the first magnetization curve of the material. An example input file is available here.
For the hysteretic B(H) properties Isotropic hysteretic, Jiles-Atherton model; Isotropic hysteretic, Preisach model described by 4 parameters of a typical cycle; and the Isotropic hysteretic, Preisach model identified by N triplets, the required file must describe a complete B(H) hysteresis loop . The formats of the .csv or .xlsx file remain similar to the others mentioned above. An example file is provided here.
In the case of the LS model, the Material Identification tool will launch a dedicated tool called MILS, with its specific identification workflow. To perform an LS model identification with MILS, please refer to this documentation chapter.
In the case of the Modified Bertotti model, the input data required by the identification tool consists of a set of files relating the peak magnetic flux density Bmax (in teslas) in the material to the specific iron losses (in W/kg). The input file also contains additional information such as frequency f (in Hz), material density ρ (in kg/m3), electrical conductivity σ (in S/m) and lamination thickness d (in m). An example input file is provided here. The goal of this identification tool is to find coefficients (k1, k2, k3) that best fit the input data. By default, the exponents (α1, α2, α3) are respectively forced to the set of values (2 , 2 , 1.5), but they may be manually adjusted with the help of sliders once a first identification is completed. It is also possible to select an option to force the classical losses coefficient k2 to remain equal to a suggested theoretical value during the data fitting procedure. That value is valid for low frequencies (which are equivalent to large skin depths), for a classical losses exponent α2= 2 and is given by (2π2σd2)/12. In the lower left corner of the GUI, the user has also the possibility to set the Bmax upper limit (in teslas) used to plot the estimated model and thus extend it beyond the values contained in the measurement files.
For a multi-frequency approach, the user can select several files corresponding to different values of frequency at Step 1 of the identification.
The Isotropic splines for mechanically cut lamination steel module is dedicated to users willing to account for the degradation of the B(H) magnetic property of electrical steel sheets subjected do mechanical cutting (e.g., punching) in Flux simulations. Its inputs are two .csv files containing the width in milimiters and (B,H) magnetic measurements for two differently cut samples of an electrical steel sheet: a narrow strip and a wide strip. From these, the built-in method identifies two equivalent materials with B(H) properties of type Isotropic spline saturation. The first material represents the B(H) property of the zone degraded by the cutting, and the second represents the B(H) property of the remaining intact parts. The depth of the degraded zone is also evaluated, and all these results may be easily introduced in a Flux project through a PyFlux file generated with the tool.