An analytical Network for Switched Reluctance Machines with Highly Saturated Regions Q. Yu, C. Laudensack, D.Gerling Universiteat der Bundeswehr Muenchen, Germany Qiang.Yu@unibw.de Abstract- In this paper an analytical network for flux linkage and flux density distribution of a switched reluctance machine (SRM) is developed based on the magnetic equivalent circuit (MEC). Due to the positional dependent reluctance in the air gap, all paths are separately modeled. Instead of using empirical equations, the effective air gap reluctance is analyzed from machine geometry. Due to high local saturation in the teeth, the machine is discretized into adjustable elements. In addition, the slot leakage and flux distribution in the end part are taken. Finally, the simulation is discussed and compared with PC-SRD and finite element method (FEM). Keywords- SRM, magnetic curves, high saturation, magnetic path, machine discretization. I. INTRODUCTION The MEC method is often used in electrical machines [1]. The magnetic curves are prerequisite for analytical design of SRMs. However, the precise calculation of such curves poses great challenge by complex air gap path and extreme high local saturation [2]. Very often the air gap reluctance is analytically solved by empirical estimations and saturation distribution is analyzed by FEM [3-4]. Although it is impossible to get exactly the flux linkage, it is demonstrated in this paper how to simulate the machine by making use of its special characteristics. First, different phases are assumed to be independent. Next, the model is divided into 3 types according to the degree of overlapping, namely the unaligned and non-overlapping positions (Type 1), the partial overlapping positions (Type 2), and the total overlapping including aligned positions (Type 3). Type 3 is very simple as almost flux goes through the air gap vertically and not discussed here. II. THE CALCULATING PRINCIPLE Consider Path i at aligned position. From stator teeth it goes through the air gap vertically, and then split in half to each side of rotor yoke (Fig.1). The network is obtained by investigating one quarter of the machine. For the air gap, stator and rotor teeth, the effective width is divided in half so that the value of reluctance is doubled. There are two of such elements connected in series, so the coefficient is 4 times. The inductivity is Li = ( 4 NTP ) 2 4 R AG + 4 RST + 4 RRT + RSY + RRY (1) where NTP is the number of turns per pole. However, not all paths can be modeled by (1) because the air gap reluctance is time dependent. Consider the circuit nearby the unaligned position (Fig.2), similarly there are 2 loops in parallel. Each loop revolves around one quarter of the machine and goes through the air gap twice. The two air gap reluctances are represented by RAG1 and RAG2. They depend on rotor position and are not necessarily the same. Fig. 2. The Magnetic path and network of the second kind The equivalent network in Fig.2 can be further divided into 2 circuits that their inductivities are connected in parallel, according to the difference of the air gap reluctances. Meanwhile, the reluctance of yoke and teeth of each sub-path is estimated as half of the original (Fig.3). The model is developed under a 3-phase 12/8 SRM and the path is shown in Fig.1. The rotor position θ = 0° is defined as aligned position while θ = 22.5° as unaligned position. The machine is divided and each part is lumped by a reluctance block, namely the air gap RAG, stator teeth RST, rotor teeth RRT, rotor yoke RRY, and stator yoke RSY. Fig. 3. The equivalent network of the second kind So each path is converted and characterized by the air gap that it goes through only once. With symmetry the inductivity of Path i is alternatively written as Fig. 1. The Magnetic path and network of the first kind Li = ( 4 NTP ) 2 2 R AG + 2 RST + 2 RRT + 0.5 RSY + 0.5 RRY ( 2) Given rotor position, there are plenty of paths going through the air gap with different reluctances, and each should be modelled by (1) or (2). In SRMs, only the flux tubes vertically going through the air gap, here called the main path, tend to pass the overlapping area of teeth and then splits in the yoke as the first kind while the rest paths are from (2). The separate modelling for different paths is simple, but decisive. III. There are 4 paths altogether and Path 1 goes vertically through the air gap (Fig.5). With rotor moving towards left, the point D at end of the teeth passes W, the crossing point of OC and DE, and WD(θ) becomes negative. When teeth are approaching, Path 2 becomes very small. The boundary of Path 2 and 3 tagged with X is estimated by the shortest way principle. Path 3 and 4 are not symmetric with curvature. The width of Path 4 can be expressed with non-overlapping length on the rotor tooth. THE AIR GAP RELUCTANCE The air gap reluctance is decisive as permeability of air is very small against even highly saturated steel. The reluctance is calculated based on geometry only. For each type, the following principles apply. (1) Each path is recognized by straight lines or arcs. (2) The shortest way principle: the flux chooses the way in the air gap through the minimum length, through flux leakage perpetuates. (3) The geometrical curvature is taken into account. A. Type 1 IV. DISCRETIZING THE MACHINE Because of local saturation, it is not sufficient to present each part of the machine by a single reluctance. As part of the network, each part is further discretized. A. Type 1 In Fig.6, the stator teeth are divided into 3 areas, namely top, mid and end according to the degree of saturation. The mid and end are sorted by slot leakage tagged with LK, and the top subjects to local saturation, depending on rotor position. Similarly, the rotor is divided. Any path that going through each area is expressed by a reluctance block, except for Path 1 and 5 at the top of teeth. The circuits of totally 7 paths are also shown. Only Path 1 and 5 go through top, mid and end of both rotor and stator teeth. Next, the size of each block is initialized without using Ampere’s Law. For top reluctance of each path, the width and length are initially set as the width of air gap reluctance. Meanwhile, the lengths of end reluctances are defined as the same as that of the leakage, so lengths of the mid are known. Fig. 4. Modeling of flux paths in the air gap for Type 1 This type is defined from unaligned position to where teeth begin to overlap. All the air gap flux can be collectively represented by 7 paths (Fig.4). As rotor moves counterclockwise, Path 1 becomes the shortest whereas Path 5, 6 and 7 are gradually losing their potential. The lengths OD and OF are equal to outer radius of rotor teeth. The points A, I, L are not necessarily on the same line. AI is vertical to flux Path 1. AL is in parallel with side of teeth DN whereas DJ is in parallel with AI. C and F are respectively mid point of the width of teeth AB and DE. Giving rotor position, all the geometries of the air gap reluctances are reduced to the widths of Path 1 and 5 only. B. Type 2 Fig. 6. The network for Type 1 Fig. 5. Modeling of flux paths in the air gap for Type 2 For rotor teeth, the flux path is a little complicated as it does not go vertically down. It is assumed that the flux enters rotor teeth at an angle of π / 4 measured from horizontal axis, and the actual reluctances in the mid for Path 3 and 4 can be seen as an isosceles triangle linked with width of its air gap reluctance. The width of mid, end and yoke reluctances are initially evenly divided. B. Type 2 Fig. 7. The network for Type 2 In Type 2, 4 paths are modelled. With increasing degree of overlapping, the flux in rotor teeth begins to go vertically down. In analogy, the reluctance at top of teeth in Path 1 is modeled by a triangle, indicating degree of saturation. The size of each reluctance is initialized in a similar way. Note that when teeth are at the beginning of overlapping, the size of rotor top reluctance in Path 4 may become very large and even exceed the height of teeth because it comes from the size of air gap expressed by (DE-AY) in Fig.5. In this case the length is set lower than width as this reluctance is not dimensional sensitive. When teeth are far from overlapping and so very large the air gap length, the top tends to be unsaturated. AA1 and DD1 are initially estimated as 1/15 of the width of the tooth. However, the saturated region may expand as teeth are approaching, depending on phase current. The area can be iteratively updated through AA1 and DD1 by prescribing a flux density, but at the cost of computing time. The expected convergence varies with current and position, and is difficult to obtain without resort to finite element analysis. As Path 1 is not as decisive as it is in other types, usually the size is estimated as constant, but less accurate when extremely high current is used. Here the size AA1 and DD1 is assumed to change proportionally from unaligned position as 1/15, to commence of overlapping as 1/5 of width of teeth. The widths of mid, end reluctances of teeth and the yoke are also managed. A convergence is prescribed, based on the fact that the flux density of any part (the mid, for example) is circumferentially identical, as is shown in Fig.8. Type 2 is modeled in a similar way. The highly saturated region is expressed by an isosceles triangle with one side equal to the length of overlapping. However when the aligned position is approaching, the flux density on top of teeth makes less outstanding and height of the region is not necessarily equal to the length. Finally the triangle region reduces to a block. VI. OTHER FACTORS A. The slot leakage V. MANAGING THE SIZE OF ELEMENTS The size of elements indicating the geometry of path is subject to effective air gap length and degree of saturation. Here the iterative updating method is used and each type is discussed. For Type 1, all reluctances in the air gap and top of teeth can be regulated by the size of Path 1 in the air gap. The flux distribution at the edge of teeth is shown in Fig.8. Path 1 chooses the shortest length centered by AD into rotor tooth. Fig. 9. The flux leakage paths Fig. 8. Detailed paths at highly saturated region for Type 1 (left); Regulating the size of mid, end reluctances (right) The sizes of the highly saturated region and the air gap are geometrically related. It is assumed that the flux within the triangle travels at an angle of π / 4 measured from horizontal axis. The highly saturated region can be collectively seen as an isosceles triangle and the width of air gap reluctance depends only on the side AA1 or DD1. The main flux leakage happens at the root of stator teeth. The flux flow in the slot is shown in Fig.9, and there are in general 2 categories. Path 1 revolves around bottom while Path 2 passes across the slot. As the circumferential distance between teeth is very large, Path 2 is in this case negligible. Path 1 can be lumped by a reluctance block and the flux goes through in parallel with that in the stator teeth. The effective length is the path in the air and estimated as an arc with radius of 1/4 height of stator teeth hst as lmag _ lk = (1 / 4 )i hst i(π / 2 ) (3) B. The machine end part Type 1 is taken as an example and the flow chart is shown in Fig.11. The program is characterized by 2 loops and each is regulated by convergence from a prescribed value. The inner loop donates to material permeability from B-H curve while the outer to consistency of flux density in end, mid and yoke part of the machine by adjusting the dimensions for each reluctance. Tab.1 Local flux density at highly saturated region under 140A I = Finite Element Method II= Analytical Method Angle 22.5° 17.5° 12.5° 7.5° 2.5° 0° Fig. 10. Flux lines at the end part, upper: aligned position, middle: partial overlapping position, lower: non-overlapping position [5]. The paths at the end part of each type are quite similar. They are characterized by vertical lines through the air gap and smooth arcs elsewhere (Fig.10). The main difference between aligned and partial overlapping is the height of the path measured from X direction, which depends on the degree of overlapping. At non-overlapping angles, the equivalent length in parallel with the air gap lmin no longer equals to the air gap length δ, but relates with position. VII. ST Mid I 0.90 1.05 1.38 1.72 1.84 1.94 II 0.92 1.10 1.43 1.61 1.69 1.79 ST_max I 0.72 1.55 1.98 1.70 1.52 1.61 II 0.74 1.69 1.87 1.53 1.34 - RT Mid I 0.41 0.76 1.18 1.35 1.45 1.62 II 0.36 0.88 1.22 1.46 1.54 1.63 RT__max I 0.64 1.44 1.88 1.58 1.52 1.45 II 0.72 1.59 1.86 1.43 1.34 - The machine is first simulated with constant single phase current. The flux density at different parts is obtained and compared with FEM. Tab.1 shows that the density by both methods fit each other well. The flux density at the highly saturated area (Stator Teeth and Rotor Teeth) is tagged with ST_max and RT_max. The flux linkage at unaligned and aligned and positions are compared with PC-SRD. It is shown in Fig.12 that both methods fit each other well and some divergence occurs under large current. RESULTS AND DISCUSSION With geometry and material input, the main program is characterized by recognizing different model types from a rotor position. Fig. 12. Flux linkage at aligned and unaligned positions Next, the magnetic curves at all positions are calculated. There are altogether 23 curves with positional difference of 1 degree (Fig.13). Starting from unaligned position, the rotor angle from 22.5° to 16.5° is simulated by Type 1 while 0.5° belongs to Type 3. In general the flux linkages at different types connect to each other well. The FEM in 3 dimensions is used to compare. High accuracy is achieved by such a simple analytical network. VIII. Fig. 11. Flow chart for Type 1 CONCLUSION An analytical network is developed to calculate magnetic curves and flux density for SRM, which is characterized by Fig. 13. The magnetic curves, left: analytical, right: FEM time dependent air gap and high saturation. The model is divided into 3 types and different flux paths are modeled separately. The air gap reluctance is obtained from machine geometry and empirical equations are avoided. The machine is discretized according to the degree of saturation and all elements are adjustable. The sizing oriented method is simple and time saving, and high accuracy is achieved. Institute for Electrical Drives and Actuators, Universitaet der Bundeswehr Muenchen, Germany, 2004. BIOGRAPHIES Qiang Yu was born in China, 1982. He graduated from the Northwestern Polytechnical University in Xi’an, China in 2008 and is now with the Institutes of Electrical Drives, Universitaet der Bundeswehr, Muenchen, Germany, as a Ph.D. student. His technical interest includes field analysis and switched reluctance machines. REFERENCES [1] [2] [3] [4] [5] P. Rafajdus, I. Zrak and V. Hrabovcova, “Analysis of the Switched Reluctance Motor (SRM) Parameters”, J. Electrical Engineering, Vol. 55, No.7-8, 2004, Page(s): 195-200. A. Arthur, “Analytical Computing the Flux Linked by a Switched Reluctance Motor Phase When the Stator and Rotor Poles Overlap”, IEEE Trans. on Magnetics, Vol. 36, 2000, Page (s): 1996-2003. G.J. Li, X. Ojeda, S. Hlioui, et al, “Comparative Study of Switched Reluctance Motors Performances for Two Current Distributions and Excitation Modes”, IEEE Conf. Industrial Electronics, IECON, Porto, Portugal, Nov. 2009, Page(s): 4047-4052. R. Krishnan, “Switched Reluctance Motor Drives”, tutorial book, CRC Press, ISBN: 0-8493-0838-0, 2001, Page(s): 30-77. D. Gerling, A. Schramm“Calculation of the Magnetic Field in the End Winding Region Caused on Axial Leakage of the Air Gap Field and the End Winding Leakage Inductance of an Unsaturated Reluctance Machine at Arbitrary Rotor Position”, Technical Report, TB 05-2004, Christian Laudensack was born in Germany, 1981. He graduated in 2007 at the University of Federal Defense, Munich, in aerospace engineering. His employment experience included the German Federal Armed Force, the IAB GmbH, Ottobrunn, the Systemzentrum für Luftfahrzeugtechnik, Erding, and the Institute for Electrical Drives and Actuators at Universitaet der Bundeswehr, Muenchen, Germany. His special fields of interest include switched reluctance machines. Dieter Gerling, born in 1961, got his diploma and Ph.D. degrees in Electrical Engineering from the Technical University of Aachen, Germany in 1986 and 1992, respectively. From 1986 to 1999 he was with Philips Research Laboratories in Aachen, Germany as Research Scientist and later as Senior Scientist. In 1999 Dr. Gerling joined Robert Bosch GmbH in Bühl, Germany as Director. Since 2001 he is Full Professor and Head of the Institute of Electrical Drives at Universitaet der Bundeswehr, Muenchen, Germany.