I. INTRODUCTION
The calculation of terminal characteristics like forces and torques developed in electromechanical devices, and their variation with changes in position or excitation, is often what the designer is, ultimately, interested in [1]. This paper addresses problems related to the calculation of terminal characteristics and local field values of a cylindrical magnetic actuator. Analytical and numerical field solutions are employed. The selected design parameters are flux, flux densities, inductance, force and stored energy. It is made an investigation of the behavior of these variants under changes in the position of the actuators plunger. The plunger is free to slide vertically through a circular hole in the outer magnetic casing, the air gap between the two being very small. The device is used to produce large forces through a relatively small displacement [2-3]. When the coil current is zero, a lower bearing supports the plunger and the air gap is maximum. When a DC current sufficient to raise the related mass excites the coil, the plunger is lifted. The displacement is constrained by an upper bearing, and for this situation the air gap is minimum. Fig. 1 shows a sketch of the actuator.
The various solutions are obtained by considering two physical assumptions: The magnetic material is infinitely permeable; The magnetic material behavior is described by a nonlinear magnetization characteristic. The analytical approach is based on simple magnetic equivalent circuits. Under the assumption of infinitely permeable path, design parameters such as flux, stored magnetic energy and force can be expressed in terms of the main air-gap length:
(g ) =
3 10 4 Weber g + 0. 01 r
(1)
where g is the main air-gap and r=2.54x10 -2 meters is the radius of the plunger;
W (g ) =
1 ( NI ) = 1500 Joule 2
(2)
CALCULATION OF DESIGN PARAMETERS OF AN AXYSIMMETRIC MAGNETIC ACTUATOR USING ANALYTICAL AND FINITE ELEMENT SOLUTIONS
f (g ) =
(1000 g + 0. 254 )2
11430
Newton
(3)
TABLE II DESIGN PARAMETERS Analytical approach Flux Density [Wb/m2] 1,3493E+00 7,0677E-01 4,7878E-01 3,6200E-01 2,9102E-01 Flux Density [Wb/m2] 1,3285E+00 6,8864E-01 4,6311E-01 3,4779E-01 2,7775E-01 Stored Energy of gap [J] 4,1022E+00 2,1488E+00 1,4556E+00 1,1006E+00 8,8478E-01 Stored Energy of gap [J] 4,1264E+00 2,1765E+00 1,4813E+00 1,1242E+00 9,0656E-01
The above equations lead to design tables or curves as presented in section III. Under the second assumption, it is possible to built up the flux-magnetomotive force (m.m.f.) characteristic for the whole magnetic circuit. For a given air-gap length and an assumed value for the magnetic flux, the m.m.f. related to the magnetic path is read from the nonlinear characteristic; the m.m.f. related to the gaps is computed. The numerical approach uses a free downloadable software based on the finite element method [4]. In order to investigate the comparative behavior of design variants a unique finite element model has been carefully set. Material re-identification technique has been used to reduce discretization error from comparative evaluations [5]. The actuator shown in Fig. 1 is essentially axisymmetric. As in the analytical solutions, the circuit is driven by a source of magnetomotive force. Table I resumes the main features of the finite element model.
Numerical approach Gap (cm) 0,254 0,508 0,762 1,016 1,270 Flux [Wb] 2,6927E-03 1,3958E-03 9,3864E-04 7,0492E-04 5,6295E-04
Analytical approach Gap (cm) Inductance [H] 0,254 0,508 0,762 1,016 9,1159E-01 4,7750E-01 3,2347E-01 2,4457E-01 1,9662E-01
TABLE I FEATURES OF THE FINITE ELEMENT MODEL Potential solution Magnetic vector potential Mesh Automatically generated, using 1st. order elements Dirichlet along the vertical symmetry line +NI=3000 Ampre-turns set in right coil region r = 44000 for infinitely permeable case
1,270
Numerical approach Gap (cm) Inductance [H] 0,254 8,9756E-01 4,6525E-01 3,1288E-01 2,3497E-01 1,8765E-01
Boundary conditions
Sources
Relative permeability
III. RESULTS
Results from analytical and numerical solutions under the assumption of infinitely permeable magnetic material (first assumption) are summarized in Table II.
Once the work focuses on a computer-based model, discussion of results is based on the level of agreement of values produced by the two approaches: analytical and numerical. Presented results show an excellent agreement, for all variants and air-gap lengths. Indeed, graphical plots almost coincide. Fig. 2 presents the flux lines for minimum and maximum air-gap lengths. It should be noted a considerable increase of leakage flux crossing the coil window for maximum airgap length.
CALCULATION OF DESIGN PARAMETERS OF AN AXYSIMMETRIC MAGNETIC ACTUATOR USING ANALYTICAL AND FINITE ELEMENT SOLUTIONS
B. Energy and co-energy Numerical simulation allows the designer to verify important theoretical features related to these state variables: once stored energy and co-energy of different regions are separately computable, it is possible to observe their distribution in the device. For the case of minimum air-gap length, Fig. 4 shows the relative amount of stored energy in each region: air-gap, outer box, coil and steel path.
82%
Under the assumption of nonlinear behavior of the magnetic path, it is possible to combine analytical and numerical procedures in order to determine the required magnetomotive force of different regions to produce a given value of flux. Fig. 3 shows the curves of the flux as a function of the magnetomotive force associated to the magnetic path and to the air-gap region, considering maximum and minimum air-gap lengths.
8%
2,50E-03
values for total stored energy and co-energy are very close and their sum is equal to the surface integral of A.J in the coil region.
C. Force calculations
2,00E-03
1,50E-03
1,00E-03
5,00E-04
The lifting force of the magnetic plunger has been computed from numerical field solutions using the methods of virtual work and Maxwell stress tensor. Results have been compared to analytical calculations given by (3). The virtual work method has been used in its classical form, by evaluating the stored energy in the main air-gap region for the device in two consecutive positions. The implementation of this method requires a judicious choice of the positional displacement. On theoretical basis, smaller displacements should lead to more accurate force calculations [6]. Different positional displacements representing 25,0%, 10,0% and 5,0% of the total plunger displacement have been used to compute stored energies at the two ends of each subinterval and perform the approximate derivative. Results presented in Table III show
Flux (Wb)
CALCULATION OF DESIGN PARAMETERS OF AN AXYSIMMETRIC MAGNETIC ACTUATOR USING ANALYTICAL AND FINITE ELEMENT SOLUTIONS
that, for this particular problem, better accuracy is associated to the use of smaller positional displacements. To predict forces by the Maxwell stress tensor method, straight lines placed at the center of the air-gap have been used as integration contours. Mesh model has been carefully built to avoid integration contours crossing any nodes.
TABLE III ERROR IN FORCE PREDICTIONS Maxwell Virtual Virtual Virtual Stress Work Work Work Tensor dg=0,254cm dg=0,1016cm dg=0,0508cm 4,371% 5,530% 6,525% 7,579% 8,680% 25,742% 6,315% 3,249% 2,339% 2,007% 2,711% 0,869% 0,662% 0,631% 1,141% 0,588% 0,674% 0,724% 0,375% 0,344%
Results associated to the virtual work method are related to the smaller positional displacement. Comparison of results show that an excellent level of agreement between numerical and analytical approaches has been obtained.
CONCLUSIONS
Terminal characteristics and local field values of an axysimmetric magnetic actuator have been calculated using analytical and numerical approaches. Results from the two approaches have been compared and an excellent level of agreement has been obtained for all design parameters. Special attention has been given to a very difficult task: the calculation of forces from numerical field solutions. Earlier reported recommendations on problem modeling have been considered to produce an economical, precise and flexible numerical model of the device. Presented results stresses on how finite element solutions can help the designer to observe and quantify important features like energy distribution and leakage flux, something difficult to realize on experimental basis.
Computed forces from the two methods, along with analytical calculations are also shown in Fig. 5.
REFERENCES
1600
[1] E.S. Hamdi, A.F. Licario-Nogueira, P.P Silvester, Torque computation by mean and difference potentials, IEE Proceedings-A, Vol. 140, No. 2, March 1993. [2] A.E. Fitzgerald, C. Kingsley Jr., A. Kusko, Mquinas Eltricas, McGraw-Hill do Brasil Ltda., 1975, pp. 120-2. [3] G.R. Slemon, Equipamentos magneteltricos: transdutores, transformadores e mquinas, Vol. I, Editora da Universidade de So Paulo, 1974, pp.101-2. [4] Finite Element Method Magnetics, Version 3.2, Users Manual, 2002. [5] D.A. Lowther and P.P Silvester, Computer-aided design in magnetics, Springer-Verlag, 1985. [6] A.F. Licario-Nogueira, Computation of cogging torques in permanent-magnet machines using the finite element method, PhD thesis, University of Wales College of Cardiff, 1993.
1200
Foce (N)
800
600
400
200
Position (cm)