Iron Loss Calculations in Laminated Structures Using MagNet
Machine designers today wish to design more efficient electromagnetic and, electromechanical devices. Traditionally the designer would use the Loss Curves provided by the material supplier at the appropriate frequency. These curves are based on Epstein frame measurements of a small sample at various frequencies of a sinusoidal magnetic field.
Most modern core loss calculation approaches apply Epstein frame data neglecting a number of significant factors such as drive types, differences between flux density evolution in various parts of the model etc. These approximations have various shortcomings and many other approaches have been attempted to accurately account for iron losses. Ultimately, FEA based numerical schemes that take into account the histories of flux density in various parts of the machine as well as drive type are likely to be best. In MagNet, Infolytica's FEA based multi-dimensional electromagnetic simulation software, some novel approaches for computing iron losses have been implemented. The new techniques allow the user to separate the loss sources at the component level. Using the example of an interior permanent magnet motor, this novel approach is introduced and demonstrated below.
Calculating Iron Losses with Simulation Software
Traditionally, analysis software has used the manufacturer’s loss data in a very simple fashion. The method is to simply calculate the magnetic flux density at many points throughout the structure. This field is considered to be the peak flux density and the loss is simply read off of the Loss as a function of flux density curve for the appropriate frequency.
This method is fine for magnetostatic or harmonic analysis, since the field displayed can be assumed to be the peak field. It is not the case in general. For example, in a motor, the time-varying flux density field can be induced by motion such as in the rotor of a motor or by time-varying fields produced by currents in a motor stator for example. The frequencies in both cases are not the same for the rotor and stator. For time domain analysis a more sophisticated method of estimating the losses is required.
)
|
| Click image to see full size version |
Calculating Iron Losses in MagNet
The method used to calculate iron losses in MagNet for magneto-static and time-harmonic analysis is the traditional method. For the time domain solutions, a new method is used to get much better estimates of the losses. This method is still based on the Epstein Frame loss measurements but uses the Steinmetz equation augmented with an eddy current term to split the total iron loss into 2 components, hysteresis with anomalous losses and eddy current losses. The Loss curve data from the manufacturer at several frequencies is used to get the coefficients for the Steinmetz and eddy current loss terms. This way both loss quantities can be calculated individually. The new loss page for the material is shown in the figure to the right.
)
|
| Click image to see full size version |
Some Results
The loss calculations were performed on a BLDC model shown in the figure on the right. For the purposes of the analysis, only one quarter of the motor was modeled. This reduces the solution time required and also the post-processing time as well.
)
|
| Click image to see full size version |
)
|
| Click image to see full size version |
)
|
| Click image to see full size version |
)
|
| Click image to see full size version |
)
|
| Click image to see full size version |
)
|
| Click image to see full size version |
Conclusions
Infolytica has increased the capability of its software to get better estimates of iron losses in laminated steel devices. We have shown the example of a BLDC motor where the iron losses were split into hysteresis and eddy current loss components. It was shown that the flux density shows higher frequencies seen in the rotor. From the results it can be seen that the losses are higher in the rotor, in particular, the eddy current losses are much higher in the rotor than in the stator.