Ferrofluid Cooling for High Power Dense PM Machines
by Guang-Jin Li of University of Sheffield

In this presentation, we explore key findings derived from a recently concluded EPSRC project, focusing on the innovative application of ferrofluid for machine cooling. Ferrofluid, an oil-based liquid containing nano-sized ferrimagnetic particles, was introduced into the end space (end-windings) of permanent magnet (PM) machines. Leveraging the magnetic body force generated by the end-winding leakage flux, the ferrofluid autonomously circulates, eliminating the need for external pumps. This unique circulation mechanism establishes an efficient heat transfer pathway from the end-windings to the housing equipped with a water jacket. Consequently, the incorporation of ferrofluid enhances the heat transfer rate, thereby improving the overall thermal performance of the machine. To simulate the cooling effects of ferrofluid in electrical machines, we developed multiphysics models that account for the intricate coupling between electromagnetic (EM) fields, fluid dynamics, and heat transfer. These models facilitated an in-depth exploration of the impact of various winding structures, such as concentrated double and single layer windings, and distributed overlapping windings, on the cooling efficiency of ferrofluid. Our findings revealed that distinct magnetic fields in the end space, generated by diverse winding structures, result in varying cooling efficiencies. Nevertheless, all machines employing ferrofluid cooling demonstrated a significant enhancement in both electromagnetic and thermal performance. This research sheds light on the promising potential of ferrofluid as a transformative cooling agent, providing valuable insights into the nuanced interplay between winding structures and the resulting cooling efficiency in electrical machines.

You must be a member to download papers. Membership Information...

Improved Understanding of Noise and Vibration Issues in Electric Machines
by Simon Redfearn, David Moule, Rob Holehouse, Riza Jamaluddin of ZF Group / Hexagon, System Dynamics

The presentation presents the work delivered by ZF in collaboration with Hexagon’s System Dynamics to understand the NVH characteristics of a power-steering electric motor(e-Motor). In the project, methods to analyse vibration response of an electric motor for power-steering application was developed. Hexagon provided the NVH simulation solutions, whilst ZF performed the 2D and 3D finite element electromagnetic analysis and provided measured test results of the power-steering motors tested on a newly constructed rig. The project used practical applications from ZF to drive the evolution of the simulation software, to validate simulation and test, enhance the capabilities of the Romax software suite, and prove a new simulation approach for e-Motor NVH. The key electromagnetic and mechanical modelling considerations identified during the study are presented. Finally, the key insights into the dynamic response of the electric motor are discussed outlining how these may have been missed had the correct modelling and simulation considerations not been used.

You must be a member to download papers. Membership Information...

Panel Discussion - Open Floor
by Bilquis Mohamodhosen, Cris Emson, Jonathan Godbehere of Dassault Systemes, Infologic design Ltd and ANSYS

Simulation plays a crucial role in product design from cradle to grave. As software providers, our aim has always been to create and democratize new and more meaningful ways to make simulation as close to reality as possible. Up to this date, many challenges have been taken up to better exchange, synchronise and harmonize data amongst various physics involved in a product design and simulation. Yet, this will never be satisfactory enough until we achieve a perfect virtual twin. Since 'perfect' is unattainable, we can only strive towards it. The intention of this panel discussion is to provide a platform for the simulation community to express their concerns, ideas, opinions and remarks regarding the current evolution of the multiphysics simulation domain, and the role of AI powered solutions in modern engineering. This will be a good opportunity for software providers to listen to those concerns so as to work towards a common goal.

Design optimization of a Synchronous. Reluctance Electric Motor using Deep Learning Technology
by Nicolas Riviere of ANSYS

The automotive landscape has developed rapidly in recent years, due to the electrification trend to support the transition towards a climate-resilient, energy-efficient, and low-carbon economy. Most of passenger EVs today are using PM motors with NdFeB material in their rotor to boost performance across the full operating range and exhibit high efficiency levels out of a limited space envelope. However, since the dramatic surge of rare-earth materials’ price in 2011, car manufacturers are actively looking into viable alternative options. As a PM-free motor, the Synchronous Reluctance (SyncRel) machine is a cheap and attractive candidate, although it cannot outperform PM-based technologies in any way. To get the most out of the SyncRel motor, an efficient optimization strategy is required. As of today, traction motors are mostly optimized through FEA-based parametric optimization procedures, either directly or from surrogate models. Such strategies are not suitable for the optimization of SyncRel motors giving the high variety of possible shapes for the rotor flux barriers. In this study, it is proposed to optimize the rotor of a SyncRel motor leveraging the proprietary deep learning technology of Neural Concept. The optimization workflow uses CAD and CAE data generated from the Ansys Motor-CAD multi-physics integrated tool to train sophisticated neural networks. The calculated predictive models are then used to rapidly optimize the electric motor for maximum output power within stress and torque ripple requirements. This novel optimization strategy has the unprecedented benefits of generating out-ot-the-box design shapes by going beyond the original parameter space, while reducing computation times and preserving a high level of accuracy with respect to the latest FEA-based parametric optimization approaches.

You must be a member to download papers. Membership Information...

Machine Learning Applied to Electromagnetic Device Analysis and Design
by Dr Cris Emson / Prof D Lowther of Infologic Design Ltd / McGill University, Montreal

Engineering design has gone virtual, with devices being simulated in a virtual world. This allows for rapid prototyping, plus the ability to include multi-physics phenomena in a cost effective manner. But the question then arises whether this changes the traditional approaches to design? For example, does the use of a Digital Twin enhance or replace the skills and expertise of a human designer? Can computers be as effective as humans in the design process? This presentation introduces the concepts behind AI and Machine Learning as applied to Electromagnetic Design, demonstrating how these techniques can be useful in aspects of design engineering that benefit from either user expertise, or extensive computer resources.

You must be a member to download papers. Membership Information...

PM Motors in High Performance Applications: Modelling Challenges
by Alex Michaelides of HiSPEED Ltd

We present the development of a high performance PM motor, with a novel continuous winding stator, flooded in oil. The construction greatly enhances continuous power, while peak Torque density is very competitive. The methodology of construction and aspects of automation / manufacturing are presented. Modelling of this PM Technology magnetic performance is presented (to compare with competing architectures) and the challenges faced in modelling this electromechanical arrangement and optimizing its performance are discussed.

Digital twin development for transportation electrification
by Tao Yang of University of Nottingham