Elie Libbos with adviser A. Banerjee

Elie Libbos Figure 17: Optimal pole color map based on (a) MTPA and (b) ML strategies: Low pole counts are the choice at low torque levels while high are better in high torque regions. A variable-pole IM mixes all pole operation to guarantee optimal operation over the whole torque/speed range

(a)

Elie Libbos Fig17b Figure 17: Optimal pole color map based on (a) MTPA and (b) ML strategies: Low pole counts are the choice at low torque levels while high are better in high torque regions. A variable-pole IM mixes all pole operation to guarantee optimal operation over the whole torque/speed range

(b)
Figure 17: Optimal pole color map based on (a) MTPA and (b) ML strategies: Low pole counts are the choice at low torque levels while high are better in high torque regions. A variable-pole IM mixes all pole operation to guarantee optimal operation over the whole torque/speed range

High power density, high efficiency, and inexpensive drivetrain operating over a wide speed range are critical for future electric vehicles (EV). The induction machines (IM) offers a cost-effective and reliable alternative to permanent magnet-based solutions. A multi-phase drive (more than three phases) can reconfigure the machine to various pole counts to extend its speed range and implement strategies such as maximum torque per ampere (MTPA) and minimum losses (ML). Figure 17 shows a color map of the pole count for wide torque/speed range using the MTPA and ML optimization strategies. At low torques, lower pole counts are the choice since they minimize the no-load magnetizing current. At higher torques, higher pole counts are preferred as their high magnetizing current helps produce the required torque with minimal current and losses. From zero to full torque, the pole gradually shifts from 2 to 8 for the simulated machine. Each pole loses its attractiveness after its base speed because of flux weakening due to the dc bus voltage limit. Figure 17 (a) and (b) are similar in shape, yet the intersection points between variable poles occur at different torque levels. In other words, there is a region where a lower pole count minimizes the stator, yet the higher pole count minimizes power losses.

Figure 18 (a) shows our experimental setup. Two 9-leg GaN-based inverters are used to drive a 36-slot toroidally wound IM with access to each of its individual slots. This setup allows us to reconfigure the IM to various pole counts (2-, 4-, 6- and 8-pole). A control PCB that includes a cyclone FPGA was built to filter noisy measurements, deal with 18 current measurements and change the pole count of the machine. Figure 18 (b) shows our experimental torque-per-ampere plot which indicates again that the pole must be changed at points A (2 to 4) and B (4 to 6) to minimize the stator current. This project is supported in part by the Grainger Center for Electric Machinery and Electromechanics.

Elie Libbos Figure 18: (a) Experimental setup with the 36-slot toroidally wound IM driven by two 9-phase GaN-based inverter modules. (b) Experimental torque per ampere for all torque levels at 900 rpm. A gradual shift from a 2- to 4- to 6- pole is done to maximize torque per ampere

Figure 18: (b) Experimental torque per ampere for all torque levels at 900 rpm. A gradual shift from a 2- to 4- to 6- pole is done to maximize torque per ampere

Elie Libbos Figure 18: (a) Experimental setup with the 36-slot toroidally wound IM driven by two 9-phase GaN-based inverter modules. (b) Experimental torque per ampere for all torque levels at 900 rpm. A gradual shift from a 2- to 4- to 6- pole is done to maximize torque per ampere

Figure 18: (a) Experimental setup with the 36-slot toroidally wound IM driven by two 9-phase GaN-based inverter modules.