The Importance of Multiphysics when Simulating Electric Motors

Phenomena:

  • Electromagnetic Motor Simulations

Applications:

  • Automotive
  • Aerospace
  • Defense
  • Transportation
  • Manufacturing
  • Consumer Goods

Main Software:

  • ANSYS Workbench
  • ANSYS Maxwell
  • ANSYS Fluent
  • ANSYS Mechanical (vibro-acoustics)

Tools:

  • DesignXplorer

Analysis Type: FEA & CFD

Computing Power:

  • Maxwell: 8 cores max for 3D solver, 1 core for 2D solver
  • Fluent: 16 cores

Mesh:

  • ANSYS Maxwell internal tetrahedral mesher
  • ANSYS Meshing for ANSYS Fluent

Findings:

  • Without multiphysics the wrong magnet could be chosen
  • Difference in torque results can be as high as 17%
  • Single physics good to eliminate designs not finalize them 

Electrical Motor Design Requires Multiphysics

When designing an electric motor, simulation specialists need to ensure performance at the customer’s specified operational temperatures. However, assuming your motor will remain in this optimal temperature range could lead a design team astray.

“You have to take into account the temperature dependency of the permanent magnets,” said Olivier Roll, senior application engineer at ANSYS. “This temperature dependency will have an impact on the performance of the magnet and therefore affect the resulting torque.”

Based on the current, the motor will experience energy losses. If the coolant system is unable to shed that heat, the motor’s temperature will increase. If the motor overheats it could demagnetize the magnets, reducing the torque output.

Using ANSYS Maxwell and ANSYS Fluent, simulation experts can model the electromagnetics and CFD of the motor design. These simulations can then be linked using ANSYS Workbench for a multiphysics simulation.

Section design of the electric motor includes the magnets, rotor, stator and cooling system.

Tips to Simplify Electro Motor Simulations

Simulation of the electromagnetic performance of the motor. Only ANSYS mechanical results are depicted (ANSYS Fluent not shown but included in the simulation).

The simulation expert should test the electromagnetic and CFD simulations separately before linking them together. This will help with bug fixing the system later.

However, the Maxwell electromagnetic simulation could serve an important pass/fail test.  

“It is worth checking if the initial designs work at a given homogeneous temperature using Maxwell,” recommended Roll. “This is a useful way to quickly eliminate designs that will never meet your requirements. Multiphysics simulations can be computationally expensive. It is recommended to save the best designs for these more in-depth analyses. Don’t spend weeks on a design that won’t work.”

Another way to simplify the simulation is to use a 2D electromagnetic model for long motors. These longer motors will typically have a near uniform result in one direction. Therefore, if accurate results are possible by eliminating the analysis in that direction, then it will significantly reduce the compute time. For shorter motors, however, users will still need to use a 3D simulation.

Though electro motor simulations can often be simplified into a 2D simulation, this isn’t the case for CFD. Fluid dynamics often has many 3D effects that need to be assessed. Extruding and mapping a 2D set of results as input to a 3D simulation has been simplified with Workbench.

Setting up the Electro Motor Multiphysics Simulations

ANSYS Fluent results depicts the temperature in the motor and the coolant. ANSYS Maxwell heat loss feeds into Fluent Simulation.

When users are pre-processing a model to be simulated using ANSYS Maxwell and Fluent, they will need to mesh the geometry with both ANSYS Mesher and an internal meshing tool within Maxwell. Though these tools are optimized for their specific task, it does lead to some results mapping challenges with multiphysics.

“Maxwell mesh is composed of tetrahedrons,” said Roll. “For CFD, on the other hand, multiple elements can be used.”

He added: “This sets ANSYS apart from other multiphysics simulation tools that use one mesh. A single mesh makes data transfer faster. However, you can lose data in the solution or make the computational times needlessly large. A mesh scaling factor can be seen when modeling a mechanical and electromagnetic simulation if you wish to also assess vibro-acoustics. ANSYS believes in meshing for good results and then let Workbench handle the transfer.”

As Roll suggests, Workbench solves the communication challenges between two solvers as they work on their individual mesh. “For global efficiency, ANSYS has created a link between 2D and 3D Maxwell simulations to a 3D ANSYS Fluent simulation,” he said. “We don’t want to force people to use a 3D Maxwell simulation for a simpler mapping of the data exchange. The idea is to automate the mapping between the FEA and CFD meshes.”

Simulations Replace “Best Guess” with Evidence

Single physics and multiphysics results show a 17% difference in predicted torque results.

“We do simulations not to have shiny results to impress the marketing teams,” joked Roll. “We want to have physical information. Something that represents truth.”

“That is why simulations are based on material properties, linear and non-linear dependencies,” he added, “They show us when our best guesses might be wrong.”

Roll warns that without a multiphysics analysis of the motor, users will ignore the temperature affects. This can have as large as a 17 percent difference to the real results.

In this particular design, the least optimal permanent magnet (NdFeB) would be chosen over the true optimal (Sm2Co27) if multiphysics is ignored.

For example, some take for granted that NdFeB magnets are more powerful than Sm2Co27 magnets. However, S2C027 are more stable than NdFeB magnets with respect to temperature.

Using ANSYS Workbench, users can link together various other tools to exchange data. This opens the door not only for multiphysics but design space exploration, optimizations and third party tool interactions. These optimization tools can determine the optimal magnetic material for the design based on the performance, temperature and cost trade-offs.

As it turns out, Roll found that for this particular model, the NdFeB magnets would appear to perform better if temperature was ignored. However, once temperature effects are taken into consideration, results show that the more temperature stable Sm2Co27 is the truly optimal choice.

For additional analysis, analysts may need to look into the noise and vibration of the electric motor.

To speed up optimizations, Roll suggests that users narrow down the design space parameters. But they must be careful doing so or else they’ll risk another assumption similar to the temperature dependence.

By connecting DesignXplorer to the simulation in ANSYS Workbench, simulation experts can perform a 6-sigma analysis. This test will determine the most crucial parameters to an optimization. DesignXplorer can then be used to perform the optimization or design space exploration.

When users are designing electromagnetics for consumer goods, like computer fans and phone vibrators, they will often need to ensure that the motor runs quiet. This will require adding an ANSYS mechanical analysis to the system within Workbench. To learn more about ANSYS Mechanical’s vibro-acoustics simulations, follow this link.

Otherwise, to learn more about simulating electro motors in general, follow this link.

 

ANSYS has sponsored this post. They have no editorial input to this post - all opinions are mine. Shawn Wasserman