Motor Design Improvements with Hardware-In-Loop Simulation

Motors & Generators with MotorSolve

Model-based design is a very common idea. It simply consists of using a block diagram of a system to describe it's behavior, and then to design relative to that description. In particular design specifications are evaluated through system simulation. The big advantages are that it allows for a better integration of components so problems are identified early and also allows the design of a component to be tuned to the system as a whole.

Prius 2004 IPM Traction Motor



The design procedure consists of: 1) creating a system model, 2) Creating models of the different components which can be modified. 3) Creating behavioral models of the various components and evaluating, by simulation, the entire system. 4) Modifying the behavioral models of components and performing a new system simulation.

View larger image


One concept in model-based design is the idea of Multi-resolution modeling. The idea is to represent the component of a system at different levels of complexity or accuracy. The simplest model of a PMSM is its D and Q-axis inductances and the flux linking the stator windngs and the rotor magnets (3 numbers). The next level is to represent the inductance and back-EMF as a function of position, so that they can be non-sinusoidal. The data now consists of a few waveforms, so quite a bit more data. Finally, an even more complete model can take into account the nonlinear behaviour of the magnetic materials. These are now surfaces instead of waveforms, a lot more data. Simulation time also increases with the complexity of the behavioral model. The simplest can take from a few seconds to a few minutes. The linearized model can take 5 times longer and the nonlinear model about 100 to 500 times longer than the minimal model.

View larger image


Changes to a component's design can be rapidly evaluated in a system context. The objective is to determine how a change to one of the components impacts on the performance. For this example we will examine the effects on the rotor design change between the 2003 and 2004 Toyota Prius HEV. The two designs differ on the magnet placement, the magnet is flat in the Prius 2003 design and V-shaped in the 2004 design. The performance measure we want to evaluate is the the time to accelerate from 0 to 60 mph. This requires that the system as a whole has to be evaluated.

View larger image

MotorSolve MODEL

The two versions of the Prius motor were modeled in MotorSolve. Both linear and nonlinear behavioral models based on the MotorSolve models were then exported to an Opal-RT data file.

View larger image


The files containing the behavioral models are loaded into the Opal-RT FPGA based HIL simulator and the system simulations are performed.

View larger image


The results for the 2 designs with linear and nonlinear behavioral models of the motor show that for quick analysis the linear model gives good overall trends. The detail at any speed is better for the full nonlinear models. Comparison of the two rotor designs also show that the newer design has better acceleration than the initial design.

View larger image