3746 IEEE TRANSACTIONS ON MAGNETICS, VOL 31, NO. 6, NOVEMBER 1995 zyx esign Optimisation of Moving-Magnet Actuators for Reciprocating Electro-Mechanical Systems R. E. Clark, D. S. Smith, P.H. Mellor, D. Howe University of Sheffield,Department of Electronic and Electrical Engineering, S 1 3JD, U.K. Abstract zyxwvutsrqpo - A methodology for the design optimisation of linear moving-magnet actuators for use in reciprocating electro- mechanical systems is described. It is analysis based and accounts for the dynamic performance with due account of the system non-linearities. Predictions are validated against measurements on a prototype actuator. I INTRODUCTION Linear actuators are employed in a wide range of reciprocating electro-mechanical actuation systems, ranging from diaphragm air-compressors and Stirling cycle and pulse- tube cryo-coolers to fuel pumps and artificial heart devices. The normal mode of operation is to drive the actuator at the mechanical resonant frequency so that no work is done in accelerating/decelerating the moving mass. Optimum efficiency and displacement are thereby attained, the actuator only supplying power to the mechanical load and system losses. The mechanical resonant frequency is a function of both the inertia and compliance of the system, e.g. due to the elastomeric diaphragm and gas compression in air- compressors, which exhibit a non-linear stiffening spring characteristic. This paper describes a design methodology for moving-magnet actuator prime-movers for such systems, in particular compressors for Stirling cycle cryo-coolers. 11. MOVING-MAGNET LINEAR ACTUATORS Actuators for such applications are required to be efficient and reliable, typically offering 20,000 hours of maintenance free operation, and must also meet demands for functionality and cost, whilst fulfilling minimum volume and loss criteria. Typically, the stroke varies between 4mm and 8mm peak to peak at frequencies between 20Hz and 60Hz. The most suitable actuator technology is the linear moving-magnet type, which has the advantages of a high specific force capability; no flying leads -which are a potential source of unreliability and limit the achievable stroke; good thermal dissipation -as the windings are bonded to a yoke; and bi- directional force production. Of the possible topologies, an axi-symmetric 2-pole configuration, such as that shown in figure 1, comprising a central limb, two slotless windings, an outer shell, and two radially magnetised magnets, is preferred as it is essentially self-shielding and does not require endcaps, thereby allowing a good mechanical interface with the load. 111 DESIGN METHODOLOGY An actuator design must meet the specified performance criteria regarding force and displacement, whilst minimising parameters such as volume and material cost, and maximising Manuscript received February 17, 1995 The authors would like to thank Lucas Advanced Engineenng and Huntleigh Nesbit Evans for the provision of EPSRC/CASE awards Shell zyxwvutsr 0 zyxwvu D Fig. zyxwvut 1 SchemaQC diagram of moving-magnet linear actuator, showing leading design parameters efficiency. The design methodology must, therefore, take account of the electro-mechanikal system and its inherent non-linearities, predominantly the stiffening spring characteristic of the load. However, a full state-variable system simulation would carry a significant time penalty when incorporated into a design optimisation routine, whilst the utility of formal optimisation procedures, such as simulated annealing [I], is limited when both static and dynamic performance considerations are important.. Therefore, the design synthesis presented here uses a parameter scanning optimisation technique [I], based on analytical models of the actuator to predict its static characteristics, coupled to an equivalent circuit representation of the mechanical system, to locate optimum designs. The leading design parameters of the actuator are first identified and then incremented between specified limits, feasible designs being produced for many parameter combinations. Their performance is displayed graphically by plotting one performance factor against any other, to aid the user in identifying a possible optimum design. zyx IV. ANALYTICAL MODEL The first phase of the design is to dimension the magnetic circuit and evaluate the magnetostatic performance parameters. Initially a simple linear model of the magnetic circuit or a non-linear lumped parameter model is used. A critical design consideration is the dimensioning of the central limb and outer shell, for which an iterative technique is employed to ensure that a specified maximum flux density is not exceeded. In this way, maximum leverage is achieved from the component materials. For example, when the scanned parameters include the magnet dimensions and' airgap length, on the first iteration the diameter of the centre limb is set at an arbitrary value and all other dimensions are derived fiom the calculated magnet flux to ensure that maximum flux density constraints are not exceeded. The diameter of the centre limb is then recalculated and the process is repeated until the calculated airgap flux density on 0018-9464/95$04.00 zyxwvut 0 1995 IEEE