< Back to previous page


Magnetic Springs for Improved Performance of Highly Dynamic Drivetrains

Book - Dissertation

Modern production machines feature a number of highly dynamic motion systems. These motion systems require powerful and expensive motors, with high energy consumption, in order to repetitively move at a competitive production rate for millions of cycles. Some of the examples are pick-and-place robots, packaging machines, weaving looms… In all of these systems, technologies for peak power reduction are of interests as a means to reducing their total cost of ownership. The beneficial effects on peak power reduction of compliant actuation have been repeatedly demonstrated in the robotics academic community in the recent years. However, the full potential of the concept has yet to be leveraged in the industrial state of use. The main challenges are that of system robustness, both with respect to mechanical failure and performance under variable operational conditions, but also a lack of a systematic study on the added value of compliant actuation in the context of production machines. This thesis stems from a specific spring technology - magnetic springs. It leverages the benefits of magnetic spring as an enabler of compliant actuation concept for heavy duty, industrial machines. Namely, magnetic springs do not suffer from fatigue failure avoiding lifetime challenges in design, while also opening up a number of new interesting design choices that are not feasible with conventional springs. Starting from the drivetrain architectures with magnetic springs in highly dynamic drivetrains, the basic concept and fundamental principles of how magnetic springs can improve performance of specific subsets of manufacturing machines are explained. Methodological contribution is then made to be able to embody these architectures. Two main methodological building blocks are developed that facilitate the design of such drivetrains and allow for optimized sizing on component and system level. On the component level of the magnetic spring, the presented design methodology consist of a detailed magnetic spring component design using magnetostatics finite element models for a range of radial flux topologies. On system design level using scalable physical component models in the paradigm of 1D system dynamics models, where sizing and selection of different components and solving of the optimal control problem are addressed to achieve a minimum total cost of ownership as a trade-off between energy consumption, maximum speed of operation and actuator density. Additionally, a patented concept is developed and tested to achieve an adaptive magnetic spring. This development extends the benefits of magnetic springs to the production machines with variable loading. Lastly, the developed methodology has been applied to several relevant industrial scenarios in order to validate the concept, the chosen modelling approaches, but also demonstrating and quantifying the added value for each application. Considered cases are: low-power weaving drivetrain, high-power weaving drivetrain, pick-and-place robot and gear meshing torque ripple in wind turbine drivetrain. Overall a systematic study of the generic impact of this technology on the design space of highly dynamic actuators is presented for a relevant range 1W-100kW. These findings, validated by a combination of experimental data and detailed virtual designs, show a systematic increase in speed of operation in the range 20-200% or a reduction in energy consumption by 55-83% or alternatively a potential for motor downsizing of 28-70% for the studied reciprocating motion.
Publication year:2020