Modeling Considerations for Optimizing EV Motors

Figure 1. An electric motor modeled in the COMSOL Multiphysics® software. This model shows the magnetic flux density in the laminated iron and the current density in the stator hairpin conductors. (Image: COMSOL)

Optimizing EV motors before they hit the market is crucial. As the demand for clean energy vehicles grows, customers need assurance that their chosen vehicle is reliable and fulfills sustainability promises. The best way to accomplish this is to use multiphysics modeling and simulation during the optimization process (Figure 1). For the engineer working on the optimization, there are several modeling aspects that are important to consider when it comes to EV motors. We will dive into these considerations here.

Investigating Torque and Iron Thickness

When starting the optimization process for an EV motor model, particularly a compact permanent magnet (PM) motor, it is essential to find a balance between compactness and efficiency, as these factors conflict. Achieving efficiency goals while limiting the size of machine components is challenging.

Efficiency equates to less energy consumption, and finding a way to limit this involves examining the motors’ current loading, which represents the amount of current applied to the stator winding. In a PM motor, the magnetic fields generated by the rotor magnets and the stator current rotate in synchronization. The interaction between these magnetic fields generates the net torque that converts the stator winding currents into mechanical power. A maximum amount of mechanical torque needs to be exerted on the rotor so that it will rotate with as much force as the engineer wants, and in the direction they want. All of this considered, the engineer will want to determine whether the current loading is maintainable or too high, as having a too-high current loading will cause the core material to go into heavy saturation, which will result in a decline of mechanical power.

The best way to examine the current loading and torque is by testing the model setup. To better visualize the topics discussed here, consider a 10-pole PM machine with 12 slots ( Figure 2).

Figure 2. Example model of a 10-pole PM motor with 12 slots. (Image: COMSOL)

Because the torque relies on the synchronization of the magnetic fields produced by the rotor and the stator, the optimum angle offset between the rotor and stator fields must be determined. The offset angle can be found by either rotating the rotor with an angular span corresponding to an electrical period or by cycling the stator current through an electrical period, with the rotor at rest.

Once the optimum angle offset is determined, the engineer can begin investigating ways to improve efficiency and save material use. There are of course many ways one can do this, but here, we will discuss one method: examining the dependency between the stator iron material, current loading, and torque output. This can be found by analyzing the thickness of the iron core and its impact on efficiency.

The engineer should simulate different variants of iron thickness, focusing on the effects that each level of thickness has on the rotor torque. By doing this, an iron thickness value that maintains optimal torque and stays within size (compactness), weight, and pricing restrictions or goals can be determined.

For the example motor seen in Figure 2, it is found that the optimal iron thickness is 2 mm; using a thickness under this value will negatively impact the torque. For example, as seen in Figure 3, a thickness of 1 mm resulted in lower torque. Going thicker than 2 mm, however, is not the optimal solution, as it requires an increase in material and therefore increases in weight and cost without a substantial increase in torque.

Figure 3. A plot showing variations of the example motor’s rotor torque and iron thickness. (Image: COMSOL)

Determining Losses, Temperature Rise, and Cooling Solutions

Once the optimal iron thickness is determined, it is time to calculate the iron and copper losses. An increase in speed (and the consequent losses) can make it difficult to keep a motor cool, which will cause a motor to reach its thermal limit. When this spike in motor temperature occurs, the motor flux density can decrease, having a domino effect on the torque and overall efficiency. For these reasons, one should look at iron and copper losses first, and then at temperature rise and cooling solutions.

Iron losses occur due to time-varying magnetic flux density and they include hysteresis and eddy current losses. Copper losses, on the other hand, happen because of conduction current flow, and they include ohmic losses within the stator coil. Both types of losses can be calculated in the COMSOL Multiphysics® software. The software offers a built-in feature for loss calculation as well as features for creating plots of the losses and their important characteristics, such as how they vary as functions of both speed and current amplitude.

From there, it is useful to compute the losses in combination with heat transfer phenomena using simulation software. This makes it possible to visualize how much the motor will rise in temperature and how different forms of cooling, such as forced or natural air convection or forced water cooling, can help subdue it. Obtaining a visualization of these conditions will help in determining the needed insulation class. For example, assume that for the model shown here, forced air convection is the preferred method. When forced convection with 1 m/s flow velocity is applied to the model, the results indicate that the most beneficial insulation class in this case would be 130 (B), which means that 130 °C would be the maximum hot spot temperature for the design (Figure 4).

Figure 4. The temperature results when forced air convection with 1 m/s flow velocity is applied. (Image: COMSOL)

Analyzing an Efficiency Map

The data collected thus far can be combined and accounted for in an efficiency map (Figure 5), which shows the efficiency as a function of speed and torque. This makes it possible to first estimate the overall energy used by a motor during a complete drive cycle and then to deduce the range of the vehicle after a single charge for a given battery size. Given that an EV motor’s marketability depends on how well it utilizes electrical power, it is safe to say that taking the time to create an efficiency map is worthwhile.

Figure 5. An efficiency map from a parametric simulation. The numbers labeled in the map (65, 75, and 85) refer to the color scale shown on the right. (Image: COMSOL)

Efficiency maps use color and contour to help visualize the efficiency. In Figure 5, the blue, yellow, and green regions have low efficiency, and the vehicle altogether consumes less power at the low speed or low torque points as compared to the red region, where both speed and torque are high.

If satisfied with the energy conservation results that the efficiency map shows, the engineer will have optimized the motor effectively and can begin the next stage of development.

Improving Motor Noise Levels

In addition to optimizing the motor’s functionality and sustainability, a crucial aspect must be taken into account: motor acoustics. EVs are significantly quieter than their conventional counterparts, which initially raised safety concerns due to the difficulty in alerting nearby cyclists and pedestrians. As a result, it has become standard practice for EV designers to incorporate a warning sound. Multiphysics modeling and simulation can be used to manage this sound as well as the overall acoustics of the vehicle, enabling engineers to ensure that it produces ideal sound before spending time and money to make a prototype.

To determine the acoustic behavior, there are three analyses the engineer should perform. First up is a transient analysis to determine the electromagnetic forces in the time domain for a given rotation speed. A time-dependent analysis is needed for electric machines since time-harmonic analyses do not account for motion.

Next, the Fourier transform can be used to convert the time-domain forces into frequency-domain harmonics. This allows for an efficient computation of the vibrations and noise emitted from motor housing.

The third step is to perform a vibro-acoustic analysis of each harmonic at different speeds of rotation. These harmonics can be plotted using a Campbell diagram (Figure 6), which shows loudness as a function of speed and frequency for a set of harmonics. With this, the engineer can find the frequency and intensity of the emitted noise (also known as the sound pressure level).

Figure 6. A Campbell diagram for analyzing motor frequency. This plot focuses on a given microphone position. Each line represents a harmonic, and on the color scale, red represents louder noise. (Image: COMSOL)

With COMSOL Multiphysics, the engineer can export the noise plots to WAV files and hear how the motor will sound. Depending on the results, they can make any adjustments they wish to the rotor slot sizes, position, and shapes until the motor produces ideal noise.

The optimization of an EV motor involves a careful examination of numerous factors, such as torque, iron thickness, and temperature rise. Multiphysics modeling and simulation make it easy for engineers to see how the different physics phenomena within an electric motor interact and optimize their design accordingly.

COMSOL Multiphysics is a registered trademark of COMSOL AB.

This article was written by Beth Beaudry of COMSOL (Burlington, MA). For more information, visit here .