Magnetic Core Selection and Engineering Design Guide for 2MW Three-Phase Solid-State Transformer (20kV/800V, 70kHz, DAB Topology)

Under the operating conditions of 2MW high power and 70kHz high frequency, magnetic components become the core bottleneck for the power density, efficiency, and thermal reliability of Solid-State Transformers (SST). The magnetic core material, operating point, structure, and thermal dissipation design must be highly matched to the high-frequency square wave excitation characteristics of the Dual Active Bridge (DAB). Based on current engineering practices in high-frequency power magnetic components, this article provides a complete and implementable magnetic core selection and design guideline.

I. Material System Comparison: Determining the Optimal Solution for 70kHz/2MW Scenarios

For DAB applications requiring 20kV high-voltage isolation, 2MW power transmission, and a 70kHz switching frequency, the magnetic core material must simultaneously satisfy: high saturation flux density, low high-frequency core loss, high-temperature stability, and suitability for high-voltage insulation. The applicability of three mainstream materials is assessed below.

1. Nanocrystalline Alloy – Preferred Solution 
Nanocrystalline materials (such as the FINEMET / VITROPERM series) are currently the benchmark materials for high-power isolated DAB converters operating at the 70kHz level.

• High Saturation Flux Density (Bs≈1.2T): Significantly higher than high-temperature ferrites, allowing for a smaller core cross-sectional area at the same power level, thereby substantially increasing power density.
  
• Low Loss in the 20kHz–80kHz Range: Within this frequency band, the volume-specific loss is significantly better than amorphous alloys; high Curie temperature and minimal temperature drift make it suitable for long-term full-power operation.
  
• Low Magnetostriction: Results in lower high-frequency noise, making it suitable for modular SST cabinet applications.
  
• Disadvantages: Higher cost, difficulty in large-scale monolithic molding, material brittleness limiting machinability; the loss advantage diminishes at frequencies >100kHz.
  
• Conclusion: In 2MW/70kHz SST applications pursuing high power density, high efficiency, and high reliability, nanocrystalline is the comprehensively optimal solution.
  

2. High-Frequency Low-Loss Ferrite – Only as a Low-Cost/Low Power Density Alternative
Ferrites offer extremely low losses at very high frequencies (>100kHz), low cost, and mature processing. However, in the specific context of 70kHz, 2MW, and 20kV isolation:

• Low Saturation Flux Density (only about 0.4T at high temperatures above 100°C): This leads to a significantly enlarged core volume, contradicting the SST miniaturization goal.
  
• Difficulty balancing temperature rise and core size at high power levels.
  
• Conclusion: Only suitable for applications with no volume constraints, high cost sensitivity, or lower power levels. Not recommended for this project.
  

3. Amorphous Alloy and Silicon Steel – Not Recommended

• Silicon Steel: Eddy current losses increase dramatically at 70kHz, making it completely unsuitable.
  
• Amorphous Alloy: While Bs≈1.5T, losses at 70kHz are higher than nanocrystalline. It exhibits high magnetostriction, high noise, and difficult-to-control temperature rise at high frequencies. It is more suitable for power frequency / medium frequency applications < 10kHz.
  

II. Key Design Parameters: Core Elements Determining Temperature Rise and Lifespan

1. Operating Flux Density Bmax Must Be Strictly Derated
Under high-frequency square wave excitation, core loss exhibits a strongly non-linear relationship with flux density (B). The saturation flux density of 1.2T cannot be used as the design basis.
Engineering Recommended Range for 70kHz / 2MW DAB: Bmax (Peak) = 0.25T – 0.35T

• Below 0.25T: Core becomes oversized, reducing power density.
  
• Above 0.35T: Core loss increases rapidly, leading to uncontrolled hot spot temperatures and significantly increased long-term reliability risks.
  
• Recommended Process: iGSE improved loss calculation → Loss density distribution → Thermal simulation (liquid cooling/oil cooling) → Iterative determination of optimal Bmax.
  

2. Core Structure: Suitable for High Voltage, High Current, Easy Winding, and Good Heat Dissipation
Recommended Structure:

• C-Core / Cut Core (Block Core)
  
• Advantages:
  
• Suitable for winding on both the 20kV high-voltage side and the low-voltage high-current side.
  
• Controllable leakage inductance, which can be directly utilized as the resonant inductance for the DAB.
  
• More friendly design for assembly, insulation, encapsulation, and cooling channels.
  

3. Air Gap Strategy: Eliminate Hot Spots from Concentrated Air Gaps
DAB converters typically do not require large concentrated air gaps; leakage inductance can be achieved through winding arrangement and magnetic circuit design.
Recommendation:

• No Air Gap / Distributed Small Air Gaps
  
• Concentrated Air Gaps are Strictly Prohibited:
  
• Concentrated air gaps lead to fringing flux, additional winding losses, and localized eddy current hot spots, which are极易过热 at high frequencies and high power.
  
• Advanced Approach: Integrated magnetics can be employed: precisely set the leakage inductance through symmetrical and interleaved winding arrangements, achieving no air gap, low loss, and low noise.
  

4. High-Voltage Insulation and High-Frequency Stress Design
20kV isolation combined with a 70kHz high-frequency square wave presents an extremely high risk of Partial Discharge (PD).
Key Measures:

• Use high-withstand voltage skeletons like PEEK on the high-voltage side.
  
• Employ multi-layer insulated wires or triple-insulated wires.
  
• Utilize vacuum encapsulation (epoxy/gel) to eliminate air bubbles.
  
• Design creepage distances and clearances according to enhanced insulation standards for high-frequency high voltage.
  

III. Final Selection Decision Table

IV. Key Engineering Recommendations for 2MW/70kHz DAB

1. Do Not Force High Bmax: Nanocrystalline's Bs=1.2T does not mean it can operate at 1.2T. At 70kHz, a slight increase in B causes an exponential rise in loss. A 0.5% loss translates to 10kW of heat.
   
2. High-Frequency Loss Requires Both Simulation and Experimental Validation: Nanocrystals exhibit non-linear effects like inter-laminar eddy currents and dynamic losses at high frequencies, causing deviations between theoretical calculations and actual results. Mandatory use of FEM electromagnetic and thermal coupled simulation, followed by prototype temperature verification.
   
3. Thermal Management is the Primary Constraint: Even if core loss in a 2MW system is controlled to 0.3%, it still amounts to 6kW. Liquid cooling or oil cooling must be adopted to strictly control hot spot temperatures and prevent core performance degradation and insulation aging.
   
4. High-Voltage High-Frequency Insulation Equals Safety Design: The voltage stress of a 70kHz square wave is far higher than at power frequency. Partial discharge can rapidly destroy insulation. Insulation design must adhere to certification-level standards for high-frequency high voltage.
   

Conclusion

The optimal magnetic core solution for a 2MW/20kV/800V/70kHz DAB solid-state transformer is: Nanocrystalline Alloy + Bmax≈0.3T (0.25–0.35T) + C-Core/Ring Core Structure + No/Distributed Air Gaps + Reinforced High-Voltage Insulation + Liquid Cooling/Oil Cooling. This approach simultaneously achieves high power density, high efficiency, low noise, and high reliability, representing the most mature and rigorous engineering solution currently available.