Project 01 · Formula SAEIn progress
Motor Design (In Progress)
Custom spoke-type permanent magnet motors for a 4WD FSAE car, with 16kW continuous front hub motors and 33kW continuous rear inboard motors. Sized from vehicle-level torque and speed requirements and validated in Ansys Maxwell, where the first front motor design delivered 11.4Nm at 25A RMS against a 10.9Nm continuous target.
- Front motors
- 16kW cont. / 26 peak
- Rear motors
- 33kW cont. / 53 peak
- Front torque target
- 10.9N·m cont.
- Front torque (sim)
- 11.4N·m @ 25 A RMS
- Status
- Design phase
Goals & requirements
Approach
Most student designs fail because the scope is too large or they never get finished, not because of the technical ability of the team. With that in mind, I set out to design a functional motor rather than a perfect one. I made a lot of assumptions and placed constraints on the design that may not be optimal, but they let me prioritize progress over perfection. The competition makes the requirements fairly clear: an 80kW combined power limit and a maximum 600V DC bus. On top of that, I added some team constraints. The motors had to be compatible with our existing gearboxes, the front and rear motors needed to be as similar as possible, and manufacturing should be simplified wherever possible (no carbon wrapped rotors!). I also sized the motors for full power capability rather than our current pack, so the motors won't be the limiting factor in the future.
Front and rear power split
Putting a motor at each wheel gives us full torque vectoring capability. The main drawback is unsprung mass, since the motor and gearbox sit on the wheel side of the suspension. Because the front wheels need to steer and the driver sits between them, it's very difficult to get the front motors onto the sprung side of the car. The rear motors, however, can fairly easily be placed inboard. By shifting power (and therefore mass) to the rear inboard motors, we can drastically reduce the car's unsprung mass. This also puts more power at the rear tires, which have the most traction under acceleration. I chose a ⅔ rear to ⅓ front split based on the team's understanding of vehicle dynamics, what other teams have done, and advice from FSAE alumni. Optimizing this split would need a validated vehicle model that accounts for unsprung mass, which we don't have yet, so it was kept out of scope. This gave a front motor sized at 16kW continuous / 26kW peak and a rear motor at 33kW continuous / 53kW peak.
Electromagnetic design
Rotor configuration
I looked at three rotor configurations. Surface permanent magnet (SPM) rotors allow a very small air gap and suit high speed applications. Interior permanent magnet (IPM) rotors let you tune the motor so its best efficiency point lines up with your most common operating point. Spoke rotors direct magnet flux tangentially instead of radially, which gives extremely high flux density and torque density. Efficiency matters less for competition scoring than raw performance, so the IPM's main advantage didn't buy us much. The rear motor has high torque and low speed requirements, which made it a natural fit for a spoke rotor. The front motor would suit an SPM, but I designed it as a spoke too, to keep the design problem simpler. The tradeoff is higher back EMF, which limits top speed, but the front motor only needs to reach roughly 20,000RPM, which is achievable with this layout.
Pole and slot selection
Since I designed around an off-the-shelf inverter and our existing gearboxes, pole selection was fairly simple. Top speed is set by the maximum electrical frequency the inverter can produce (n = 120f / p). I took our simulated maximum speed at competition (~120km/h), worked it back through the gear ratios to a motor RPM, and calculated the maximum synchronous speed for a range of pole pair counts. That landed on 5 pole pairs for the front motor (24,000RPM max) and 10 pole pairs for the rear motor (9,000RPM max). A nice side effect is that both motors run at the same electrical frequency at their respective speed targets (~1167Hz at 14,000RPM front and 7,000RPM rear). The front gearbox is 14.33:1 and the rear is 4.24:1, so the ratios don't line up perfectly, but reaching top speed is a rare edge case and the small difference isn't a concern. For the front motor I chose a 10 pole, 12 slot combination with concentrated windings. It's a very torque-dense combination, but it has a lot of subharmonics, which add to rotor heating. Our events don't have many extended high power draws, and several other FSAE teams that build their own motors run this combination without issues, so I was comfortable moving forward with it.
Sizing and windings
From the vehicle requirements I built a spreadsheet for each motor to calculate torque, speed, inverter voltages and currents, air gap flux density, and windings. The front motor needs 10.9Nm continuous and 17.5Nm peak, with a flux weakening point at 14,000RPM. It has a 96mm OD, a 60mm stack length, a 0.6mm air gap, and an estimated average air gap flux density of 0.93T. Choosing the number of turns is a balancing act. More turns means more flux and more torque, but it also means more back EMF at a given speed, which pushes the flux weakening point lower. I calculated the turns needed to meet the torque requirement and the turns needed to hit our bus voltage at the flux weakening point. These came out to ~16 and ~13 turns. I wasn't willing to sacrifice peak torque, so I started my simulations at 16 turns. That way I wasn't guessing at the windings, and if the simulation agreed with the hand calcs I could be much more confident in both.
Electromagnetic simulation
I'm validating the design in Ansys Maxwell 2D. I started with a simple back EMF study: no current applied, just spinning the rotor. This let me learn the motor and flux layout and get comfortable with the software. The first results looked nice, but they weren't real. The tutorial I followed didn't set the correct magnet and coil directions for my layout, so I went back to the textbook winding layout, drew it out on paper, and corrected the coil polarities and magnet orientations. Once the flux linkage plots came out as clean sine waves, I knew the model was set up correctly. From there I reduced the stator size based on my current and copper area calculations.
Driving the motor
Next, I fed the motor current. I tried coupling Maxwell to an inverter model in a co-simulation, but I had trouble getting the load to draw more current, so I've set that aside for now. Instead I drove the model directly with sinusoidal phase currents. At 25A RMS and 14,000RPM (the flux weakening point), the front motor produces about 11.4Nm with fairly low torque ripple. That beats the 10.9Nm continuous target, and it's a good sign that the hand calculations and the simulation agree.
Thermal & mechanical
Mechanical design considerations
The mechanical side is where much of the risk is. Getting the air gap as small as possible improves performance, but any contact between the rotor and stator destroys the motor. That means building a thorough tolerance stack for the air gap at MMC/LMC, full load deflection, and peak temperature. I've chosen angular contact bearings in a back-to-back configuration. They are preloaded and very rigid, which tightens the shaft's radial position and lets me shrink the air gap stackup. The bearings will be ceramic hybrids. Circulating currents in a motor try to pass through the bearings, and arcing between steel rolling elements and races eventually causes bearing failure. The output end will be fixed to reduce moments on the shaft, and the inner end will float on an O-ring and wave spring to allow for thermal expansion.
Manufacturing & test
Not started yet. See the progress log below for what's next.
Progress log
-
Next steps
This project is still in progress. The next steps are:
- Finalize the slot/pole combination for the rear motor
- Optimize the front motor geometry
- Run modal and rotor dynamics analysis
- Design the cooling and housing (metal 3D printed housings are a strong option, since a motor's power rating is really a measure of how well it can be cooled)
- Work out how to retain the laminations on the rotor and stator
- Complete the full air gap tolerance stack
I'll keep updating this page as the design progresses.