Ensuring Electric Machine Efficiency with MagNet and OptiNet

The task of the machine designer--a tireless effort to optimize torque ripple, running torque, efficiency, cost and a whole host of other factors--is now made easier with the latest releases of MagNet and OptiNet from Infolytica.

An IPM (interior permanent magnet) machine is optimized to minimize the torque ripple while maintaining a minimum running torque, ensuring that the back EMF does not exceed the supply voltage, with the additional constraints that the efficiency be at least 80% at both 1800 and 4000 rpm.

The design parameters include the geometry of the permanent magnet and the advance angle of the stator field.

Results

Click image to see full size version

The flexibility and power of OptiNet allows real-world optimizations to be set up easily. Improved designs can be ready in a matter of hours, even with a realistic number of complex constraints and objectives. This optimization problem includes:

-Torque ripple minimization
-Ensuring running torque of 0.475 Nm at 1800 RPM
-Ensuring running torque of 0.350 Nm at 4000 RPM
-Ensuring back EMF does not exceed supply voltage at 1800 and 4000 RPM
-Maintaining efficiency of 80% at both 1800 and 4000 RPM (efficiency calculation includes ohmic, eddy current and core losses)

Click image to see full size version

Efficiency is of prime concern for many modern machine applications. Accurate calculations require MagNet's advanced models for core losses, eddy currents as well as ohmic losses due to current flow in complex winding structures. The tight coupling between our packages allows MagNet and OptiNet to interact seamlessly to solve this challenging problem.

The images at right show the losses from ohmic heating in the windings and in the magnets (due to eddy currents) as well as the core losses.

Click image to see full size version

A combination of OptiNet's built-in functions and the easy but powerful scripting capability of MagNet allows the specification of these constraints. In the case of multiple, potentially conflicting objectives, relative weights can be assigned. For constraints, they can be strictly enforced or somewhat relaxed to allow for a better overall satisfaction of multiple constraint functions.

Click image to see full size version

Despite the number, diversity and complexity of the constraints and the challenges inherent in multi-objective, multi-variable optimization, OptiNet can demonstrate real benefits. In just over 6 hours, an improved solution was found that satisfied all the constraints, and reduced torque ripple by a factor of 40%.