Small Magnet Design Change Could Deliver Big Energy Savings for AI Data Centers

Highlights

• A chamfered ferrite magnet geometry achieved 93.04% modeled efficiency and torque ripple below 1% in a 3-kW outer-rotor PMSM cooling fan motor.
• The optimized prototype met the IE5 efficiency class under IEC 60034-30-2, validated experimentally with less than 0.5% efficiency deviation from simulation.
• Ferrite-based designs avoid rare earth supply chain dependence, though researchers caution they are not yet ready to replace rare earth magnets in high-performance applications.
• With AI data centers consuming vast electricity, even incremental fan motor efficiency gains translate to significant energy savings at scale across thousands of cooling units.

Researchers led by Chaewon Jo and Professor Ju Lee of Hanyang University, working with collaborators from the Korea Electrotechnology Research Institute (KERI) and the University of Science and Technology (UST), report that a relatively simple redesign of permanent magnet geometry can improve the efficiency of electric motors used in cooling fans for AI data centers. Rather than developing a new motor or magnetic material, the team compared four permanent magnet shapes in ferrite-based motors and found that a chamfered magnet delivered the best overall performance while meeting the IE5 efficiency class under the applicable IEC standard. The study demonstrates that careful engineering optimization—not just new materials—can produce meaningful energy savings in one of the world's fastest-growing electricity markets.

How the Study Worked

The researchers designed four versions of an outer-rotor permanent magnet synchronous motor (PMSM) using ferrite magnets and optimized each geometry using a multi-objective genetic algorithm (MOGA) coupled with finite element analysis (FEA). The best-performing design was then manufactured and experimentally validated under both no-load and variable-load operating conditions.

What They Found

The chamfered magnet design achieved the highest modeled efficiency (93.04%) while maintaining torque ripple below 1%. Experimental testing closely matched computer simulations, with only 1.4% error in back electromotive force (BEMF) and less than 0.5% efficiency deviation. At the representative operating point defined under IEC 60034-30-2, the prototype satisfied the IE5 efficiency class for variable-speed AC motors.

Why It Matters

AI data centers require enormous amounts of electricity, and cooling systems account for a substantial share of that demand. Even small efficiency improvements in fan motors can translate into meaningful energy savings when deployed across thousands of cooling units.

Perhaps more importantly for the critical minerals and rare earth elements sector, the study demonstrates the growing role of materials optimization. Ferrite magnets are significantly less expensive and avoid dependence on rare earth supply chains, although they generally deliver lower magnetic performance. This research suggests that thoughtful motor design can narrow that performance gap in specific applications—but importantly, it does not suggest ferrite motors are ready to replace rare earth permanent magnet motors in high-performance applications such as electric vehicles, robotics, aerospace, or defense, where power density remains paramount.

Study Limitations

The study evaluated only a 3-kW EC fan motor designed for HVAC applications and compared four magnet geometries under controlled laboratory conditions. The findings should therefore be viewed as application-specific rather than universally applicable across the broader permanent magnet motor industry.

Citation: Jo, C., Kim, S., Paul, S., Han, P., Chun, Y., & Lee, J. Comparative Study of Permanent Magnet Shapes in Outer Rotor PMSM for EC Fan in Data Center. IEEE Access (accepted author manuscript, 2026).

Register today: REEx Marketplace™ (opens in a new tab)