Docstoc

Coupled Electromagnetic and Ther

Document Sample
Coupled Electromagnetic and Ther Powered By Docstoc
					Coupled Electromagnetic and Thermal Solution for Electric Machine Design

Xiao HU Zed (Zhangjun) TANG ANSYS, INC.

© 2009 ANSYS, Inc. All rights reserved.

1

ANSYS, Inc. Proprietary

Introduction
• Electric machine design is a multi-physics problem
– – – – Electromagnetic Fluid and thermal Mechanical (Stress, Vibration) Power electronics/control

• Electromagnetic, thermal and mechanical designs are interrelated
– Losses from electromagnetic design affect temperature – Temperature rise will change material properties – Thermal induced mechanical stress

• A design environment that accommodates all physics and their interaction is highly desired
– ANSYS Workbench environment
© 2009 ANSYS, Inc. All rights reserved.

2

ANSYS, Inc. Proprietary

Simulation Driven Product Development - Electric Machine Design Methodology

Much better solution with ANSYS CFD/Mechanical

© 2009 ANSYS, Inc. All rights reserved.

3

ANSYS, Inc. Proprietary

Maxwell2D for Electromagnetic • Majority of Electromagnetic Designs are Done in 2D for Electric Machine
– >80% – Faster – Enough accuracy

• Maxwell2D Transient Solver
– Transient excitation – Transient motion – Motion induced transient effects

• Coupling between Maxwell3D and ANSYS is possible and follows the same design flow
© 2009 ANSYS, Inc. All rights reserved.

4

ANSYS, Inc. Proprietary

Design Flow

Centroids Workbench Mesher

Temperature

Geometry Workbench DM Maxwell UDP

Losses Maxwell

Mapped Losses
ANSYS Mechanical (automated) ANSYS CFD (Scripted)
ANSYS, Inc. Proprietary

© 2009 ANSYS, Inc. All rights reserved.

5

Accurate Loss Coupling • Most Losses are Distributed – Eddy loss (PMs) – Core loss (Stator & Rotor)

• Time Averaged Spatial Losses – Time constants are very different for electrical and thermal
© 2009 ANSYS, Inc. All rights reserved.

6

ANSYS, Inc. Proprietary

Export Thermal Data to ANSYS Mechanical

© 2009 ANSYS, Inc. All rights reserved.

7

ANSYS, Inc. Proprietary

Import Maxwell Loads to ANSYS Mechanical

© 2009 ANSYS, Inc. All rights reserved.

8

ANSYS, Inc. Proprietary

Maxwell 2D – ANSYS Thermal

© 2009 ANSYS, Inc. All rights reserved.

9

ANSYS, Inc. Proprietary

Mechanical Eigenmode analysis of thermal pre-stressed model with Maxwell 3D transient losses

© 2009 ANSYS, Inc. All rights reserved.

10

ANSYS, Inc. Proprietary

1.75 KHz mode results of pre-stressed structural model

Thermal deformation

© 2009 ANSYS, Inc. All rights reserved.

11

ANSYS, Inc. Proprietary

Need for Computational Fluid Dynamics (CFD) • CFD is the science of predicting fluid flow and heat transfer by solving mathematical equations • Electric machine cooling involves fluid flow and heat transfer and thus can benefit from CFD simulation

© 2009 ANSYS, Inc. All rights reserved.

12

ANSYS, Inc. Proprietary

CFD Models for Electric Machine • Conjugate heat transfer with mapped losses from Maxwell – Solids with different properties – Liquid or air for cooling – Air trapped inside electric machine • Multiple Reference Frame (MRF) used to account for rotor rotation – Steady state solution with the impact of rotating rotor

© 2009 ANSYS, Inc. All rights reserved.

13

ANSYS, Inc. Proprietary

Cooling Methods for Electric Machines
• Forced convection liquid cooling – Most effective cooling – Expensive • Forced convection air cooling – Effective cooling – Somewhat expensive • Natural convection air cooling – Not as effective – Cheap

© 2009 ANSYS, Inc. All rights reserved.

14

ANSYS, Inc. Proprietary

Test Cases and Purposes • Three test cases are conducted to see the effectiveness of cooling and different temperature and its gradient distribution
Cooling Method Case 1 Case 2 Case 3 Forced Water Forced Air Natural Air Mesh Size (K) 916 1007 899

© 2009 ANSYS, Inc. All rights reserved.

15

ANSYS, Inc. Proprietary

Geometry/Mesh
• A sector of geometry is used – Periodic boundary • Hex is used in most of the regions – Except for the winding and the fluid region surrounding it, etc. • Forced air cooling has an air domain outside

© 2009 ANSYS, Inc. All rights reserved.

16

ANSYS, Inc. Proprietary

Loss Distribution for All Cases
• • • • Spatial eddy loss distribution for the magnets Spatial core loss distribution for the rotor, stator yoke, and stator teeth Stranded winding copper loss All losses, which are highly non-uniform, are from Maxwell2D

© 2009 ANSYS, Inc. All rights reserved.

17

ANSYS, Inc. Proprietary

Temperature Distribution
• Max temperature are 398K, 517K, and 550k respectively • Forced water cooling is the most effective and natural air cooling is the least. • Forced water cooling gives similar max temperature gradient • Temperature gradient is responsible for thermal stress. • To keep both temperature and its gradient low is the best

Forced water cooling
© 2009 ANSYS, Inc. All rights reserved.

Forced air cooling
18

Natural air cooling
ANSYS, Inc. Proprietary

Summary for Forced Cooling • Forced water cooling is the most effective. • Natural air cooling is the least effective. • Forced water cooling, however, does not necessarily give the least temperature gradient. • Natural air cooling may face challenge of high temperature. • Forced water cooling may face challenge of high temperature gradient.

© 2009 ANSYS, Inc. All rights reserved.

19

ANSYS, Inc. Proprietary

Observations about Natural Convection Cooling
• Natural convection Heat Transfer Coefficient (HTC) is relatively uniform compared with forced convection • Natural convection cooling can be simulated by using a constant HTC instead of a full CFD calculation. • Well accepted industry practice. • Air trapped inside electric machines is not effective in heat transfer and thus can be removed from the calculation. • Air gap kept but modeled by STILL air (details next) • If air domains both inside and outside of the electric machine are removed, the problem becomes purely conductive • No full CFD

© 2009 ANSYS, Inc. All rights reserved.

20

ANSYS, Inc. Proprietary

Ineffectiveness of Trapped Air
• Relatively low velocity and uniform temperature of the trapped air explains its ineffectiveness for heat transfer

Velocity vector of trapped air (note the max velocity is only 2.5 m/s)
© 2009 ANSYS, Inc. All rights reserved.

Temperature distribution of trapped air (note the temperature scale goes from 500K to 530K)
21
ANSYS, Inc. Proprietary

Test Cases Using Natural Convection Air Cooling
• Case 3 is from previous study and is used as a based line case here. • Case 4 contains only solids • Case 5 also contains the air gap between the rotor and stator to improve the accuracy.
• The air gap is treated as if it is solid Cooling Method Case 3 Case 4 Case 5 Natural Air Natural Air Natural Air Trapped Air Yes No No Full CFD Yes No No Air Gap Yes No Yes Mesh Size (K) 899 394 397

© 2009 ANSYS, Inc. All rights reserved.

22

ANSYS, Inc. Proprietary

Temperature Distribution
• Max temperatures are 550K, 566K, and 563k respectively • Trapped air has minimum impact on max temperature as expected • Air gap has an impact on rotor temperature distribution

Air gap

Full CFD (case3)
© 2009 ANSYS, Inc. All rights reserved.

Solid only, no CFD (case4)
23

Solid and air gap, no CFD (case5)
ANSYS, Inc. Proprietary

Comparison

Max Winding Temperature (K) Case 3 Case 4 Case 5 550 566 563

Error 0% 2.9% 2.4%

Max Rotor Temperature (K) 528 508 523

Error 0% 3.8% 0.95%

Performance on 4 CPUs

4 ~ 12 hrs <10 minutes <10 minutes

The full CFD case is assumed to be correct.

© 2009 ANSYS, Inc. All rights reserved.

24

ANSYS, Inc. Proprietary

Summary for Natural Cooling
• Trapped air in general does not have significant impact on temperature distribution except for the air gap between the rotor and stator

• Adding a layer of STILL air in the gap can improve accuracy – This could be the best comprise considering its much quicker solution than a full CFD calculation. • Note that forced cooling still needs CFD due to highly localized heat transfer coefficient

© 2009 ANSYS, Inc. All rights reserved.

25

ANSYS, Inc. Proprietary

Conclusion
• Forced convection cooling is effective and its thermal analysis needs CFD due to highly localized heat transfer coefficient • Natural convection cooling can be effectively simulated without full CFD and thus making the simulation much easier and faster – Trapped air has impact on the solution only in the gap region, which can be modeled using a layer of STILL air. • ANSYS CFD can be used to perform either the full CFD calculation or the simplified conduction calculation • ANSYS Mechanical can be used to perform the pure conduction, thermal stress, free modal, and pre-stress modal analysis

© 2009 ANSYS, Inc. All rights reserved.

26

ANSYS, Inc. Proprietary


				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:11
posted:12/18/2009
language:English
pages:26