Predicting Performance & Cooling: Electric Motor with Multiphysics

Music Our model is asked the following question: "Why is a multi-physics simulation important?" Well, I wanted to talk about that using the example of an electric motor. We will look at two different simulations.

First, a single physics simulation using ANSYS Maxwell, where we make an assumption of the metal temperature, the magnet temperature, and determine the motor performance in terms of its torque.

Then, we will look at a multi-physics simulation, where instead of making an assumption of the metal temperature, we will actually look at how the coolant temperature affects the metal temperature, which in turn affects the motor performance, which in turn, because of heat losses, impacts the coolant temperature.

We will again look at the torque and compare the two. If we compare the two, we see very different results in terms of torque. For instance, when we assume a temperature of the material of 22 degrees, we have a torque of 180 newton per meter.

While, when we actually do the multi-physics simulation, the actual metal temperature is 54 degrees, where the torque drops to 100. So, we have a torque of 154 newton per meter. A clear difference and a clear indication of why multi-physics simulation is needed.

To do such a simulation, we actually coupled Maxwell 2D to a 3D fluent simulation inside the workbench environment. You see two Maxwell 2D simulations because we are going to look at the behavior of two different magnet materials. The geometry is the same for the two models.

You see the first fluent simulation connected to the first Maxwell 2D and the second simulation connected to the second Maxwell 2D simulation. The mesh is identical for the two fluid solutions, and we compare all the results directly inside one environment using CFD post.

Here you see the 3D representation of the motor we will study. We actually only simulate a sector of it. Now you can see the magnets. It's also liquid cooled.

So, here is the pipe which carries the liquid coolant, and the rest of the electric motor is actually made of air, so it's air-cooled as well. Now, we do the initial simulation inside Maxwell. It's a 2D simulation, and we see the temperature of the magnet.

Now, we are looking at an instantaneous result of the 2D unsteady Maxwell simulation and are actually looking at the thermal losses. Of course, the thermal losses from the material, from the structural material, will be a thermal source for the fluid simulation.

Now, what we are actually interested in is not looking at the unsteady thermal losses but what we'll do is we'll time-average them and extract the steady thermal losses and use that in the fluid simulation.

We do this because the frequency of changes of the losses in the electromagnetic material is much higher than the one in the fluid. So, the fluid will never see the oscillation of changes in losses that the material is seeing.

Now, we switch to the fluid computation and see that we just imported the time-averaged thermal losses that we computed inside Maxwell. It was very easy to do. It's automated inside the fluid environment.

We see that the maximum losses are in the area close to the intersection of the stator and the rotor and are actually ready to perform the fluid simulation. Now, we'll go one step further and look at two different materials: SM2 CO27 magnet and NDFEB magnet material.

We'll see if there's any difference in behavior when we use one material versus another, if there's any difference in performance of the electric motor.

It's actually very interesting when we start looking at the results because we can look at the temperature of the magnet for the two materials, and the temperature predicted by the multi-physics simulation only varies by a few degrees. Now, that being said, what does that mean?

Does it mean that the actual performance of the electric motors will be exactly the same for the two materials? No, because the magnetic properties of the two materials can be very different and have very different dependencies upon temperature.

In the same way, I can look at the temperature of the liquid coolant for the two cases. As you can see, the temperature is pretty much identical. So, will there be any difference in the performance of those motors? And what a surprise when we actually compare single to multi-physics simulation.

Look at the single physics simulation. The best performance is actually predicted with the NDFEB material, whereas when we do the multi-physics simulation, where we actually look at the real temperature of the magnet, we see that the SM2CO27 material is actually giving the best performance.

Quite impressive for a result, quite impressive to see how multi-physics simulations are important. Thank you.