User's manual for finite-element GoFER in PC-BDC 9.04
1-July-14 Copyright 2014 Computational Dynamics Ltd, "CD-adapco" Contents 1. Using finite elements with PC-BDC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2. Single load point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3. Bgap distribution (open circuit) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 4. Btooth waveform (open circuit) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 5. Cogging torque . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 6. Torque versus gamma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 8. i-psi calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 9. i-psi Polygon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 10. Slot permeance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 11. Dealing with errors and program failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 12. Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 13. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 Note : This manual is a reference manual for the PC-BDC GoFER. Tutorial examples can be found with step-by-step instructions in the SPEED Tutorials and in Electric Machine Design Using SPEED and Motor-CAD, available from Motor-Design Ltd. PC-BDC GoFER 9.04 Using finite elements with PC-BDC Page 1 Figure 1 Organization of finite-element processes with PC-BDC, PC-FEA, and the GDF Editor 1. Using finite elements with PC-BDC "GoFER" means "Go to finite-elements and return". The GoFER in PC-BDC automatically sets up standard finite-element calculations (Table 1), for all types of synchronous machines in PC-BDC, including synchronous reluctance machines and brushless permanent-magnet machines. The GoFER alleviates the tedious task of setting-up finite-element calculations, especially for standard calculations. With the increasing use of PC-FEA as an embedded finite-element solver in SPEED, the GoFER is an important method of setting up and analysing results in more detail. See p. 3. With PC-FEA, the GoFER also ensures the return of data to PC-BDC, for (i) display; (ii) comparison of PC-BDC and PC-FEA results; and (iii) adjustment of PC-BDC's parameters, if necessary. PC-BDC has special utilities to receive the finite-element data and to make these functions efficient and easy to use. They are accessed via Tools | FE Results, and the comparison tools are labelled MatchFE. Operation of the GoFER is by a short series of keystrokes [F11], as shown in Fig. 1. Many finite-element calculations can be completed with the basic keystroke sequence. However, at each stage it is possible to pause and view the progress of the calculation. The main stages are (1) creation of sub-domains and regions, with boundary node coordinates, material properties, and excitation levels (PC-BDC and GDF Editor) [F11, F11] (2) meshing (PC-FEA) [run] or [F9] (3) solution (PC-FEA) [continue] or [F8] (4) post-processing (PC-FEA) (5) read results in MatchFE window in PC-BDC [Tools | FE Results] The process is open so that parameters can be edited at several different levels, the highest level being that of the original PC-BDC parameters, and the lowest level being that of individual node coordinates in the finite-element mesh. In addition, the GoFER itself has several options for setting up the finite-element calculation. These options are the main concern of this manual. Page 2 Using finite elements with PC-BDC PC-BDC GoFER 9.04 Function Result Single load point A snapshot of the electromagnetic field at a particular rotor position and a particular load. Bgap distribution (open circuit) The distribution of airgap flux-density on open-circuit. Btooth waveform (open circuit) The waveform of the average flux-density in one stator tooth as the rotor rotates through an electrical cycle, on open-circuit. [Cogging torque] 1 The variation of cogging torque caused by the interaction of the magnetized rotor with the slotted stator, on open-circuit. Torque versus gamma The average electromagnetic torque as a function of the current control phase angle gamma, based on the fundamental component of airgap flux-density and the fundamental component of the stator ampere-conductor distribution. i-psi calculation The variation of the phase flux-linkage as the rotor rotates through one electrical cycle, with the ampere-conductor distribution correctly phased according to the rotor position. This is used to calculate the electromagnetic torque from the energy-conversion loop (iR loop). i-psi Polygon A set of current /flux-linkage pairs at the vertices of a polygon of the energy-conversion diagram; used for computing the torque of motors with square-wave drive. Slot permeance A flux-plot of the leakage flux produced by current in one stator slot. In line-start machines with cages, this can also be used with rotor slots. In both cases a value is calculated for the slot-permeance coefficient at the specified current-density. Table 1 GoFER functions in PC-BDC Rotor position Ampere-conductor distribution Single load-point Fixed Fixed Bgap distribution (open circuit) Fixed Btooth waveform (open circuit) Rotates through 180Eelec [Cogging torque] Rotates through 1 slot-pitch Torque versus gamma Fixed Rotates through specified range i-psi calculation Rotates through 360Eelec Changes in synchronism with rotor position i-psi Polygon A series of fixed positions corresponding to the vertices of the polygon. See Fig. 48, p. 44. A set of fixed distributions corresponding to the vertices of the polygon. See Fig. 49, p. 44. Slot permeance Not relevant User-defined current-density in the slot Table 2 Constraints applying to GoFER functions in PC-BDC 1 The square brackets indicate that the results should be approached with special care, because they may depend on the choice of mesh parameters, the No. of airgap layers, the No. of rotor steps, and other factors. PC-BDC GoFER 9.04 Using finite elements with PC-BDC Page 3 Some definitions S d-axis The reference axis of the rotor, i.e. the centre of a South magnetic pole. Solution domain The extent of the finite-element mesh, usually expressed as a fraction of the motor cross-section: for example, "two half poles" or "whole motor". The solution domain is reflected in the PC-FEA script as AngFE. Distribution The graph of flux-density that would be obtained by tracing a locus with a moveable probe while the rotor and stator and the ampere-conductor distribution remain fixed: for example, the open-circuit airgap flux-density distribution B(r,2) at r = mid gap with 2 swept through a range of 180Eelec. The idea of a "distribution" is spatial variation. Waveform The graph of flux-density (or any other parameter) that would be obtained by a fixed probe while the rotor rotates through a range of angles: for example, the tooth flux-density B(r, 2, >) at r = mid tooth, 2 = tooth centre- line, with > swept through a range of 180Eelec. The idea of a "waveform" is variation as a function of time, as in an oscilloscope trace. B gap distribution The distribution of airgap flux-density, generally understood to mean the open-circuit value. Table 3 Definitions Note Most of the GoFER settings are persistent: i.e., the program remembers them from the last time the OK button was pressed. They are not saved in or associated with a particular PC-BDC datafile. Using the GoFER with other finite-element programs The PC-BDC GoFER can sometimes be used with certain other finite-element programs. Consult SPEED for details. The embedded solver PC-BDC can also operate with an embedded finite-element solver using PC-FEA. The embedded solver is used independently of the GoFER, and is described in the PC-BDC User's manual and tutorials. However, many of the settings are the same as in the GoFER, so it is a good idea to use the GoFER to check the mesh and the solution before using the embedded solver. The GoFER also makes the results available in much more detail. For example, in most cases the embedded solver is limited to the return of flux-linkage values to PC-BDC, but the GoFER produces a wealth of detail including flux-plots and information about other field quantities. Page 4 Single load-point analysis PC-BDC GoFER 9.04 Figure 2 Single load point GoFER, page 1 2. Single load point The Single load point GoFER produces a finite-element solution at a fixed rotor angle with a fixed pattern of stator ampere-conductors. It returns the airgap flux-density distribution and the torque calculated by Maxwell stress. Single load point GoFER Files produced by PC-BDC file Test1.bd4. Test1.bgp The file containing the B gap data vs. azimuth from the S d-axis, compatible with MatchFE Test1.tq The torque values calculated by Maxwell stress within the two middle airgap layers, together with their average Test1.btby Flux-densities in the individual teeth and yoke-sections Test1.lst PC-FEA script listing Test1.rep PC-FEA report file Test1.et Element table Table 4 Files produced by Single load point GoFER Note : the filename is derived from the .bd4 filename used with PC-BDC. It should contain only alphanumeric characters, but the underscore character _ is permitted. PC-BDC GoFER 9.04 Single load-point analysis Page 5 Figure 3 Single load point GoFER options, page 2 Figure 4 Solution domain ' Half motor, Start ' q-axis Single load point GoFER options (2) Geometry X-axis the entire solution can be oriented so that the x-axis coincides with a stator slot centre-line (as in the OUTLINE EDITOR in PC-BDC); or a stator tooth centre-line. The choice is a matter of preference but there may be slight differences in the precision of the finite-element computations between the two options. The choice of x-axis has no effect on the relative angle between the rotor and stator. Solution domain Solution domain determines the area covered by the finite-element solution. For example, Fig. 4 shows a domain which is just half the motor. The angular extent is the same for the rotor and the stator, and appears in the finite-element script as AngFE. The finite-element program also needs to be told the rotor position and where to start the solution domain. The rotor position is defined by Rotor Angle. In Fig. 4 it is set to Ax1, the axis of phase 1, which happens to be at 60E in this example. The start is specified as q-axis, hence the designation Half motor, Start ' q-axis. The Default solution domain is a minimum domain derived from the highest common factor between the number of slots and the number of poles, with the aim of minimizing the solution time. Page 6 Single load-point analysis PC-BDC GoFER 9.04 The algorithm used to determine the Default solution domain does not take account of the winding distribution. Therefore the user must check that the winding is correctly represented. This will normally be the case if the winding shares the same symmetry as the pole/slot combinations, as for example with integral slots/pole and WdgType ' ConcEqual. If WdgType ' Custom or FracSlot, the GoFER will display a warning to the effect that the winding might not have this symmetry. In general the safest option is to set Solution domain ' Whole motor, although this maximizes the solution time. For motors with integral numbers of slots/pole and symmetric windings, all Solution domains are valid, but the simplest is 1 pole. If the number of slots/pole is fractional, 1 pole is not a valid Solution domain, and a warning message may be displayed. However, the user should not rely on the GoFER to check the validity of all possible combinations automatically, but exercise his or her technical judgement in every individual case. Rotor Angle This is the physical angle of the S d-axis (i.e., the rotor reference axis) relative to the x-axis; see Fig. 4. Option Position of S d-axis (rotor reference axis) 0 Aligned with the x-axis RotorAng At the angle RotorAng as set in the OUTLINE EDITOR or Ted Ax1 (no skew) Aligned with the axis of phase 1, computed with no skew User [m.deg] User-defined angle in Emech. Table 5 Rotor angle options Node tolerance If two nodes are within a distance * of each other, they will be treated as coincident. * is the "node tolerance". If * = 0, coincident nodes will be treated as separate and either the GDF Editor will fail or PC- FEA will crash. If two nodes are extremely close, setting * > 0 will merge them, so avoiding extremely short line segments and triangles with poor aspect ratio. Using too large a value for Node tolerance will lead to meshing errors. (See also Epsilon, p. 55). Include shaft If this box is not checked, the finite-element solution will not include the shaft, and the boundary conditions will be set so that no flux enters the region occupied by the shaft. Airgap layers The number of layers of elements in the airgap can be selected as 2,4, 6 or 8. The recommended number is 4. When the rotor surface is profiled, the number of layers in the airgap is set to 4. Include can This option is self-explanatory, and only applies if the motor has one or more cans. Although the Single load point GoFER does not compute eddy-currents in the can(s), the mesh is certainly affected by the presence or absence or a can. Note that the can(s) can be made magnetic by specifying a suitable bh_code in the Solving section of the PC-FEA script. PC-BDC GoFER 9.04 Single load-point analysis Page 7 Figure 5 Single load point GoFER options, page 3 Single load point GoFER options (3) Load point scale factors This panel is for setting the stator current in all phases of the winding. The GoFER uses the CONDUCTOR LOCATION VECTOR from PC-BDC to determine the number of ampere-conductors in each slot from the phase currents and the winding layout. To view the winding layout, use the WINDING EDITOR in PC-BDC. Open-circuit Zero current Th0, gamma, and other options shown in Fig. 5. The current in each phase can be defined by ISP and Th0 or gamma, exactly as it is in PC-BDC, but Fig. 5 shows additional orientations available when Drive = Sine or AC Volt. User values The current is assigned individually in each phase by a scale factor Xi1, Xi2, etc., which multiplies the displayed value of ISP. Table 6 Load point scale factors Th0 (or gamma) Th0 is the phase advance angle for squarewave drives, exactly as in PC-BDC. [Eelec]. gamma is the phase advance angle for sinewave drives. A positive value means that the current leads the EMF. Zero means that the current is "in the q-axis". "Negative d-axis" means that the ampere-conductor distribution is demagnetizing, i.e. aligned with the magnet but opposing it; (see 10). ISP The set-point current, i.e. the peak value of the line current. For delta-connected windings the peak phase current is 1//3 times the peak line current. Use template editor values If this is checked, the GoFER will retrieve the TEMPLATE EDITOR values from the PC-BDC datafile. Page 8 Single load-point analysis PC-BDC GoFER 9.04 Options Show distribution instructs PC-FEA to display the B gap distribution using its own charting facility. Create file. If this is checked, PC-FEA writes the B gap distribution file for transferring back to PC-BDC. If it is unchecked, the file is not written. RadBGap is the radius at which the B gap distribution is sampled from the finite-element solution. If the finite-element model covers an arc of less than 360E RadBgap should be set no larger than the "Mid gap" value, otherwise the B gap distribution is being specified partly outside the solution domain. In electrical machine theory the so-called BLV method of calculating EMF requires the B gap distribution at the stator bore. What is generally meant is an equivalent smooth cylindrical surface whose radius is determined from the rotor radius together with the effective airgap, computed from the actual airgap modified by Carter's coefficient. The classical B gap distribution is bereft of slot- modulation, because the effects of slotting are incorporated in an average sense through Carter's coefficient. This equivalent smooth cylindrical surface does not exist in the finite-element solution. The finite-element B gap distribution will be very uneven at the actual stator bore, being highly modulated by slotting. It does not correspond to the notional B gap distribution of classical machine theory, and in fact there is no rigorous way of obtaining the correctB gap distribution by finite-element analysis. The best that can be done is to use the B gap distribution at a radius near the centre of the airgap, and perhaps to use the fundamental component together with a correction for the fact that it is not exactly at the stator bore. This is one of the reasons why the BLV method is not recommended in conjunction with finite-element computation when the number of slots/pole is small. In such cases it is better to use the Btooth GoFER). If RadBgap is very close to the stator bore, Solution domain should be set to Whole motor, otherwise MEASURE CURVE COMMAND errors will be encountered. [Calculate torque] instructs PC-FEA to calculate the instantaneous torque by the Maxwell stress method. This is done at two radii: for a four-layer mesh in the airgap these radii correspond to the centres of the two middle layers. The two values are averaged but are also reported separately. Note that for reliable results it may be necessary to refine the mesh. The square brackets are a reminder of the sensitivity of this calculation to the users skill in refining the mesh. Calculate Bt, By (sections) instructs PC-FEA to calculate and display the flux-density values in individual teeth and yoke-sections, for comparison with the values computed by PC-BDC. Element table When this is checked, an element table file is produced for the single finite-element solution. PC-BDC GoFER 9.04 Single load-point analysis Page 9 Figure 6 Single load point GoFER options, page 4 (Permanent-magnet machines) Single load point GoFER options (4) GDF Parameters and settings Fig. 6 shows the fourth page of options for permanent-magnet machines, and Fig. ? for wound-field machines. For permanent-magnet machines, Br is the remanence of the magnet that will be used in the finite- element analysis. PC-BDC may be using an effective value such as BrT (which includes the effect of temperature), or BrTEff (which includes the effect of Shim). The effective values may or may not be appropriate in the finite-element calculation, so there is a set of options to filter them out. For wound-field machines, the equivalent excitation parameter is the field current If. Even when PC- BDC is using an effective value (modified by XIf), the XIf factor should be filtered out so it is not used in the finite-element calculation. Page 10 Single load-point analysis PC-BDC GoFER 9.04 For permanent-magnet machines the options are as follows: Br = Br (database) uses the value in the magnet database for 20EC Br = BrT uses the DESIGN SHEET value BrT, which has been adjusted by XBrT. This value is calculated at the operating temperature T_r, which also appears in the DESIGN SHEET. After a Static design, T_r = T_Mag. Br = BrT/XBrT uses the unadjusted value BrT/XBrT. Br = BrTEff uses the effective value BrTEff, which includes the effect of the Shim; if CalcBB = EffMag, it also includes the effect of Bracing. Br = BrTEff/XBrT uses BrTEff/XBrT, which retains the effect of the Shim and any Bracing bridges (if CalcBB = EffMag), but excludes the adjustment by XBrT. This causes the finite-element program to use the equivalent magnet which incorporates the effect of the shim and any bracing bridges without modelling them as separate regions. (Note that the shim can be modelled as a separate region for certain rotor types with PhysShim = true; in these cases the recommended setting is BrT/XBrT). Table 7 GDF Magnetisation options Note that the relative recoil permeability of the magnet is also adjusted along with BrT. This controls the format of the GDF file produced by the GoFER. For use with PC-FEA, this should be set to Auto, but for compatibility with 3rd-party FE links it may be necessary to use a particular version. More details about the GDF versions are published in the GDF File Format manual. Warn on GDF file overwrite If this box is checked, a warning message will be displayed if the GoFER attempts to overwrite an existing GDF file. Save design file If this box is checked, the .bd4 file must be saved before the GoFER can execute. This not only saves any design changes, but it also registers a definite path for the .bd4 data, which is required later when the MatchFE functions are executed. PC-BDC GoFER 9.04 Single load-point analysis Page 11 Faces The main purpose of the Faces option is to provide a means of saving changes to the boundary node distribution which you might make in the GDF Editor, so that PC-BDC does not over-write them next time you run the GoFER. In a GDF file a sub-domain is defined by a list of connected lines and arcs called faces. Each face has properties, such as how many divisions it is split into, the distribution of nodes, and the style (i.e. whether it is internal to a region, whether it has a fixed potential, periodic boundary condition, etc). When a SPEED program is used to create a GDF file, it uses an internal set of conditions for every face which the user can then alter in the GDF Editor. These alterations are saved with the GDF file, but if the user goes back to the SPEED program to run another analysis, the internal settings will be reapplied. A Faces file is essentially a file containing a list of face definitions for a given problem, which can be used to override the internal settings, either when the GDF file is created or from within the GDF Editor. Face information is associated with a face by means of the face name. This is formed from the names of the two end co-ordinates which are shown in the reference figures in the SPEED program manuals: for example, AB. Co-ordinate (or point) names are added to the GDF file only if the GDF version is 1.2 or higher. To distinguish between rotor and stator co-ordinates, both of which are called A Z, the convention used by SPEED software is that the name is prefixed with R to specify rotor co-ordinates and S to specify stator co-ordinates. If a co-ordinate does not have a corresponding name in the reference figures, it is omitted from the list of faces and information about it cannot be saved. When a Faces file is saved, information about all faces with valid names is written to a file. When a Faces file is loaded, a comparison is made between the face names in the current GDF file and the face names in the Faces file; where these match, the face information is updated. Faces files are simple text files, so they can be edited using a program such as NOTEPAD, for example, to include only stator information, or only specific faces of the model. Although Faces files can be generated from within the GDF Editor (or by hand) SPEED programs can use them when the GDF file is first created, provided that Use previous or User file is selected. Note that if the GDF version is less than 1.2 the Faces options will be disabled. There are three options: Faces Action Defaults The GDF file will be created with the internal settings, as in all previous versions. A Faces file can still be used to update the model from within the GDF Editor. Use previous 2 The GoFER will attempt to override the internal settings from a Faces file with the same name as the current .bd4 file with a .faces extension. For this mode to function the .faces file must obviously exist so it must first be created from within the GDF Editor. If the .faces file does not exist then no warning will be displayed but PC-BDC will continue to the GDF Editor. If you then save any changes you make to the faces in the GDF Editor the default file name will match the one expected in the GoFER and so the next run will automatically use the changed faces information. User file The GoFER will attempt to override the internal settings from a Faces file with the given name. This option is intended for use when you have specific changes you want to make, but you have stored them in a different folder or under a different name to prevent them from being accidentally overwritten. Table 8 Faces file options Require simulation if checked, a warning message appears if the GoFER is run before any simulation results are available. This is needed when Shim or Bracing affect the effective magnet properties. 2 You must save the GDF file after making changes (or use Region|Save Faces information) in the GDF Editor. Also in the GDF Editor Options menu, GDF version must not be less than v1.2 and Save Faces information must both be checked. Page 12 Single load-point analysis PC-BDC GoFER 9.04 Figure 7 Setting up the link to another finite-element program Run Create files only The .gdf and other files will be created but nothing else will happen. Edit using GDF editor The .gdf file will be displayed in the GDF editor, where it can be edited before being passed to the finite-element program. Edit using other editor The .gdf file will be passed to another specified editor (or program). Direct to FEA The .gdf file will be passed straight to the finite-element program without stopping. Other FEA links... The .gdf file will be passed to the finite- element program, optionally stopping at the GDF Editor on the way. Table 9 Run options PC-BDC GoFER 9.04 Single load-point analysis Page 13 Figure 8 Registering the executable finite-element program so that the GoFER knows where to find it Figure 9 GoFER options showing other FEA links Other FEA links... The links to other finite-element programs operate in much the same way as the link to PC-FEA, but the path to the executable finite-element program must be registered. If this is not done automatically at installation time, it can be set by selecting Other FEA links... and clicking the Add button. Fig. 8 shows the data that is required. If Default linkis checked, the identified finite-element program will be used as the default, and its name will appear with an asterisk in the list in Fig. 9. Page 14 Single load-point analysis PC-BDC GoFER 9.04 Figure 10 PC-FEA options : Meshing and Solving Single load point GoFER options (4) PC-FEA meshing Meshing MaxNvt (nodes) This can be used to limit the number of nodes which the mesh generator produces, in case the computer has a memory limitation. The maximum number of nodes in PC-FEA 3 is 32,000; but in PC-FEA 5 it depends on the available memory. Only in rare instances should this parameter need to be changed. Do not use an unreasonably low number, as it could prevent the generation of a suitable mesh. MaxAngle (recommended 120E) The maximum angle permitted in any triangular element during meshing. For a well-conditioned solution the triangles should all be approximately equilateral, but sometimes it is necessary to change MaxAngle up to even 130E to assist in cases of difficult mesh generation. Occasionally a lower limit (e.g. 90E) helps. "Include virtual lines and arcs" The PC-FEA mesh generator can be set to add new boundary nodes when necessary to improve the mesh. By default, these new nodes are placed on straight lines between existing nodes, but by using "virtual arcs" the mesh generator can be instructed to place a given set of boundary nodes on an arc to improve the geometric modelling accuracy. In some cases with motors having large numbers of slots the "virtual arcs" cannot be created with enough precision and the meshing fails. In such cases uncheck this option to prevent their inclusion in the .fea file. Do optimisation Mesh optimisation is generally required but it also increases the number of nodes and elements, which increases the size of the problem and slows down the solution. Pause If this is checked, PC-FEA pauses after the mesh optimization so that the mesh can be inspected. If it is not checked, PC-FEA continues with the solution without stopping. Table 10 Meshing options PC-BDC GoFER 9.04 Single load-point analysis Page 15 Figure 11 The BH button Solving Solver tolerance Normally this is 10 !7 . A smaller value makes the solution more precise, but it takes longer to solve. A larger value (e.g., 0@0001) gives a faster solution but less precision. With larger values, the precision of the solution is not uniform and there is a risk of serious local inaccuracy even though the overall solution may appear to be correct. Max. iterations Usually the solver completes in 4 or 5 iterations, but some problems may require more iterations especially if the tolerance is very small. The default is 99, but if the number of iterations exceeds 10 or so, it could be that the steel data is not sufficiently smooth (see WinSPEED manual), causing some numerical instability in the solver. Table 11 Solving options The BH button The BH button (Figs. 10 and 11) specifies the method of interpolation of the B/H curve. It is preferable to use B interpolation. In the steel database this curve is defined by a series of points and the data must be smooth all the way down to zero B, otherwise there is a risk of slow convergence. The WinSPEED manual explains how to prepare the steel data to achieve best performance and accuracy. With H interpolation there is a risk of losing detail at low field values if the range of H is very high. This will be the case if the steel data extends up to extreme values of B, for example, above 2@3 T. The B/H data in the .BH file is obtained from the steel database data by cubic-spline interpolation in the GoFER. PC-FEA normally interpolates linearly between these points, although cubic-spline interpolation can be selected by means of the the Spline parameter of the Region command in the PC-FEA script. With linear interpolation, a larger number of points (typically 100 or more) is required. Pre-convert to SD1 format This checkbox appears only if the steel database used in PC-BDC is in .SD2 format. If you are using a .SD2 steel database format, the GoFER can prepare the .BH file directly from the steel data without pre-converting it to .SD1 format. Software releases prior to the date of this manual would automatically pre-convert to .SD1 format, and in so doing they risked losing important detail in the B/H curve. The check-box makes it possible to revert to the old method, but it is better to leave this box unchecked and use .SD2 steel database formats whenever possible. Page 16 Bgap Distribution PC-BDC GoFER 9.04 Figure 12 Bgap distribution GoFER, page 1 3. Bgap distribution (open circuit) The Bgap GoFER produces an open-circuit finite-element calculation, with fixed rotor angle, and returns a .bgp text file for comparison with PC-BDC's airgap flux-density distribution in MatchFE. This GoFER is intended to check or adjust PC-BDC's calculation of the airgap flux-density distribution which is used to calculate the EMF waveform when EMFCalc = BLV, BLV/Unbal, HBMethod, KFR, and FullRing. Normally the Bgap GoFER will be used when the slots/pole is 3 or more; see Tutorial B08. Bgap distribution GoFER Files produced by PC-BDC file Test1.bd4. Test1.bgp The file containing the Bgap data vs. azimuth from the S d-axis, compatible with MatchFE and EMFCalc = ExtBgap. Test1.lst PC-FEA script listing Test1.rep PC-FEA report file Table 12 Files produced by Bgap distribution GoFER PC-BDC GoFER 9.04 Bgap Distribution Page 17 Figure 13 Bgap distribution GoFER options, page 2 Figure 14 Bgap distribution GoFER options, page 3 Bgap distribution GoFER options (2) All the options and settings on this page are the same as for the Single load point GoFER, p. 4, except that Solution domain is restricted to Start = d-axis so that the B gap distribution returned from PC-FEA to PC- BDC is plotted in the same location as the PC-BDCs internal B gap distribution in the MatchFE window, no matter where the rotor was positioned during the FE analysis: changing the rotor position only moves the relative locations of the slot-modulation notches. Bgap distribution GoFER options (3) The options on this page are the same as for the Single load point GoFER, p. 8. Page 18 Bgap Distribution PC-BDC GoFER 9.04 Figure 15 Bgap distribution GoFER options, page 4 Figure 16 PC-FEA options : Meshing and Solving Bgap distribution GoFER options (4) The options on this page are the same as for the Single load point GoFER, p. 9. Bgap distribution GoFER options (5) The options on this page are the same as for the Single load point GoFER, p. 14. PC-BDC GoFER 9.04 Bgap Distribution Page 19 Figure 17 MatchFE example with B gap distribution Example result After running PC-FEA, use Tools | FE Results | MatchFE: Bgap or [Ctrl+Shift+G] to view the result. Press [F] to include the fundamental components. The slot modulation dips will generally be correctly aligned only when the FE calculation has been computed with the rotor at the angle RotorAng. The adjustment parameters appearing in the edit box at the top left in Fig. 17 are not necessarily the best ones to use in the "calibration" or adjustment of PC-BDC. Others such as u_Lkg or bBsat may be more effective for certain types of rotor (RotType). It is often a good idea to make a private table of adjustment factors using the Custom Editor in PC-BDC. Page 20 Btooth waveform PC-BDC GoFER 9.04 Figure 18 Btooth waveform GoFER, page 1 4. Btooth waveform (open circuit) The Btooth waveform GoFER produces an open-circuit finite-element calculation, in which the rotor position is stepped through a sufficient angle to determine the waveform of tooth flux-density as the rotor rotates through 180 Eelec. PC-BDC can construct the entire phase EMF from this waveform. PC- FEA returns a .tfw text file for comparison with PC-BDC's B tooth waveform in MatchFE. This GoFER is mainly intended to check PC-BDC's calculation of the tooth flux waveform and EMF waveform when EMFCalc = ToothFlux. Btooth waveform GoFER Files produced by PC-BDC file Test1.bd4. Test1.tfw File containing Btooth waveform data vs. rotor position, compatible with MatchFE and EMFCalc = ExtTFW. The B-values are averaged across the tooth width. Test1.scvp Similar to Test1.tfw, but the average B tooth is obtained from the difference of vector potentials at two adjacent slot centre-lines, divided by the tooth width. Test1.savp Similar to Test1.tfw, but the average B tooth is obtained from the difference of vector potentials averaged over two adjacent slots, divided by the tooth width. Test1.bth Columns containing rotor angle, B r , B 2 and B in the tooth as the rotor rotates Test1.yoke Columns containing the rotor angle and B in the yoke Test1.lst PC-FEA script listing Test1.rep PC-FEA report file Table 13 Files produced by Btooth waveform GoFER PC-BDC GoFER 9.04 Btooth waveform Page 21 Figure 19 Btooth waveform GoFER options, page 2 Btooth waveform GoFER options (2) All the options and settings on this page are the same as for the Single Load Point GoFER, p. 4, except that Rotor Angle is disabled, as the rotor will be rotated automatically through 180E elec. Page 22 Btooth waveform PC-BDC GoFER 9.04 Figure 20 Btooth waveform GoFER options, page 3 Btooth waveform GoFER options (3) Create file Must be checked, if the Btooth waveform file is to be written. RadBtooth Radius at which tooth flux is measured. RadBtooth is calculated from the slot depth and other parameters, but it can be replaced by a user-defined value. This may be necessary in cases where the slot-sides are parallel, or where the fillet radii may affect the effective width of the tooth. See WstCalc in the PC-BDC reference manual. ByokeAngle Angle at which yoke flux is measured, normally 90Eelec away from the tooth centre-line. RotSteps The number of steps of rotor rotation. The default is 45, and the "sweep" is 180Eelec., so the points are normally spaced at intervals of 4Eelec. For a smoother result, increase RotSteps, for example to 180E. (Refining the mesh in PC-FEA also helps to produce a smoother result). Create vector potential file When this is checked, the GoFER produces the additional files *.scvp and *.savp defined in Table 13. These files have the same format as the *.tfw format and they can be imported into the MatchFE window in PC-BDC. Their purpose is to provide an alternative method of calculating the flux in one tooth-pitch in cases where the slot flux may be significant. Otherwise they merely serve to check the *.tfw file. Generally there is very little difference between the *.tfw, *.scvp and *.savp files. PC-BDC GoFER 9.04 Btooth waveform Page 23 Figure 21 Btooth waveform GoFER options, page 4 Figure 22 PC-FEA options : Meshing and Solving Btooth waveform GoFER options (4) The options on this page are the same as on p. 9. Btooth waveform GoFER options (5) The options on this page are the same as on p. 14. Page 24 Btooth waveform PC-BDC GoFER 9.04 Figure 23 Example of MatchFE / Btooth Example result After running PC-FEA, use Tools | FE Results | MatchFE: Btooth or [Ctrl+Shift+T] to view the result. Press [F] to include the fundamental components. The .tfw file produced by PC-FEA can be used in PC-BDC in the MatchFE function as shown in Fig. 23. When Create vector potential file is checked, the *.scvp and *.savp files can be used in the same way. The .tfw file can also be used in PC-BDC with EMFCalc = ExtTFW, to replace the EMF waveform calculated by the internal magnetic equivalent circuit. The *.savp and *.scvp files can be used in the same way. Plot can display differentiated versions of the graph, corresponding to the EMF in a single-turn search- coil wound around the tooth, or the EMF in the entire phase winding, or the line-line EMF. PC-BDC GoFER 9.04 Cogging torque Page 25 Figure 24 Cogging torque GoFER, page 1 5. Cogging torque The Cogging torque GoFER obtains the variation of cogging torque caused by the interaction of the magnetized rotor with the slotted stator on open-circuit. The data can be returned to PC-BDC for use in the Tools | FE Results | Skew cogging function, which determines the effect of skew on the cogging torque waveform. The Cogging torque GoFER is displayed in square brackets to indicate that particular care is sometimes required in the setup, to get a satisfactory result. With this GoFER it is less safe to rely on the default settings, which are generally preset for demonstration and fast calculation. For example, the number of steps (RotSteps) may need to be increased well above the default value, and mesh refinement is often needed as well. Cogging torque GoFER Files produced by PC-BDC file Test1.bd4. Test1.cog The file containing the cogging torque data, compatible with Tools | FE Results | Skew Cogging in PC-BDC. The results include TqMx1 and TqMx2 (T 1 and T 2 on p. 61); TqMaxwell, and the torques determined from coenergy, TqCoE; Maxwell surface integral, TqMSI; and Coulombs virtual work method, TqVW. Test1.w Columns containing energy, coenergy and vector potential as the rotor rotates. Test1.lst PC-FEA script listing Test1.rep PC-FEA report file Table 14 Files produced by Cogging torque GoFER Page 26 Cogging torque PC-BDC GoFER 9.04 Figure 25 Cogging torque GoFER options, page 2 Cogging torque GoFER options (2) The options on this page are the same as for the Single load point GoFER, p. 5. PC-BDC GoFER 9.04 Cogging torque Page 27 Figure 26 Cogging torque GoFER options, page 3 Cogging torque GoFER options (3) Cogging torque The calculation of cogging torque is susceptible to numerical discretization errors. Particularly in motors where the cogging torque is very small, extreme precision is required to get a good result. To improve the likelihood of getting a reliable result, PC-FEA provides four different methods of torque calculation, which can be compared graphically without requiring tedious adjustments to the mesh. The theory of these methods is summarized in [5] and [6]. The achievement of a smooth waveform requires a large number of rotor steps, a tight convergence tolerance, and a fine mesh in all cases. Coenergy The system coenergy is evaluated at each rotor step, and the differences are divided by )2 to give the instantaneous torque. Maxwell The classical Maxwell stress is evaluated at two radii and averaged. Virtual work The local coenergy derivative is evaluated in the airgap region (Coulomb). MSI Maxwell Surface Integral method (Arkkio) applied in the airgap. Table 15 Cogging torque options Remove DC instructs PC-FEA to remove any residual DC component from the cogging torque waveform. Show waveform instructs PC-FEA to plot the cogging torque as a function of rotor angle. RotSteps is the number of steps by which the rotor is rotated. Energy/Avg. A If the Create file box is checked, PC-FEA calculates a file of energy and coenergy vs. rotor position, using the magnet regions only, according to the algorithm described in [1]. The file also includes the vector potential at the point defined by RadA and ThetaA. This point can be defined to be at the centre of the magnet. Page 28 Cogging torque PC-BDC GoFER 9.04 Figure 27 Cogging torque GoFER options, page 4 Figure 28 PC-FEA options : Meshing and Solving Cogging torque GoFER options (4) The options on this page are the same as on p. 9. The options on this page are the same as on p. 14. PC-BDC GoFER 9.04 Cogging torque Page 29 Figure 29 Example of cogging torque calculation Figure 30 Tools|FEResults|SkewCogging in PC-BDC Example of cogging torque When all the Cogging torque boxes are checked in Fig. 26, PC-FEA plots the cogging torque from the coenergy method and the Maxwell stress method, vs. rotor position; see Fig. 29. In PC-BDC the cogging torque (computed from the coenergy method) can be further processed to estimate the effect of skew, using Tools | FEResults | SkewCogging; see Fig. 30. Page 30 Cogging torque PC-BDC GoFER 9.04 Figure 31 Harmonic analysis in Tools|FEResults|SkewCogging Figure 32 Table of harmonics in Tools|FEResults|SkewCogging The harmonic analyses of the skewed and unskewed cogging torque waveforms are shown in Figs. 31 and 32. PC-BDC GoFER 9.04 Torque vs. gamma Page 31 Figure 33 Torque-versus-gamma GoFER 6. Torque versus gamma The Torque versus gamma (T-gamma) GoFER obtains the average electromagnetic torque as a function of the current control phase angle gamma for sinewave drives, based on the interaction between the fundamental components of (i) the airgap flux-density and (ii) the stator ampere-conductor distribution. This method is applicable to motors with sinewave drive and it is generally assumed that the windings will be more or less sinusoidally distributed, with a reasonably sinusoidal distribution of airgap flux- density on open-circuit. The computation assumes that the fundamental component of stator flux- linkage is due entirely to the airgap component of the flux: i.e., it neglects the leakage components even though these are included in the field solution (with the exception of the leakage fields at the ends of the machine). The method also neglects the effect of permeance harmonics, even though the torque computed by Maxwell stress includes the cogging torque, (but only at one rotor position). These factors give rise to uncertainty in the torque calculation, but it can be checked by the i-psi GoFER, which is more rigorous but slower. The speed of calculation is maintained very fast by this method, because it uses a fixed rotor position. The torque can be calculated for one fixed value of peak line current and one fixed value of the phase angle gamma; or over a range of currents and phase angles. Page 32 Torque vs. gamma PC-BDC GoFER 9.04 Torque versus gamma GoFER Files produced by PC-BDC file Test1.bd4. Test1.tgamma Torque calculated by fundamental B and Maxwell stress methods for each current and gamma angle. Test1.psi psid (R d ) and psiq (R q ) for each current and gamma angle. Test1.vdq V d and V q calculated from the phasor diagram for each current and gamma angle. Test1.xd X d for each current and gamma angle. Test1.xq X q for each current and gamma angle. Test1.eq Open circuit E q , followed by E q for each current and gamma angle (computed using frozen permeabilities with the current set to zero). Test1_total.psid psid (R d ) including the effects of end-turn inductance (calculated by PC-BDC and passed via the GoFER) Test1_total.psiq psiq (R q ) including the effects of end-turn inductance (calculated by PC-BDC and passed via the GoFER) Test1.lst PC-FEA script listing Test1.rep PC-FEA report file Table 16 Files produced by Torque versus gamma GoFER PC-BDC GoFER 9.04 Torque vs. gamma Page 33 Figure 34 Torque versus gamma GoFER options, page 2 Torque versus gamma GoFER options (2) The options on this page are the same as for the Single load point GoFER, p. 5. Note that the rotor position is fixed such that its reference axis (the S d-axis), is aligned with Ax1, the axis of phase 1. Page 34 Torque vs. gamma PC-BDC GoFER 9.04 Figure 35 Torque versus gamma GoFER options, page 3 Torque versus gamma GoFER options (3) ISP is the peak line current. If the motor is connected in delta (or open-delta) the peak phase current will be ISP//3. Count is the number of evenly distributed levels of current for which solutions will be obtained. The first value will be the one entered in the min box, and the last one will be the one entered in the max box. Similarly the values of gamma start at the min box value and finish with the max box value. Default sets the counts, min and max values to a set of defaults. Show graph instructs PC-FEA to plot a graph of "T-gamma" curves, as in Fig. 38. Maxwell stress instructs PC-FEA to calculate the torque by Maxwell stress as well as from the fundamental flux-density and ampere-conductor distributions, for comparison. Cartesian switches the definition of phasor current to cartesian components I d + jI q . Since gamma is measured from the q-axis, I d + jI q = ISP e j(gamma + 90E) . I d and I q are peak values, not RMS. PC-BDC GoFER 9.04 Torque vs. gamma Page 35 Figure 36 Torque versus gamma GoFER options, page 4 Torque versus gamma GoFER options (4) The options on this page are the same as for the Single load point GoFER, p. 9. Page 36 Torque vs. gamma PC-BDC GoFER 9.04 Figure 37 PC-FEA options : Meshing and Solving Figure 38 Example of a Torque versus gamma curve for one current; gamma varies from 80E to 80E Torque versus gamma GoFER options (5) The options on this page are the same as on p. 14. Fig. 38 shows an example of a T-gamma curve displayed in PC-FEA. The two methods of torque calculation give different results as anticipated on page 31. PC-BDC GoFER 9.04 i-psi calculation Page 37 Figure 39 i-psi calculation GoFER, page 1 8. i-psi calculation The i-psi GoFER is designed to calculate the average electromagnetic torque from the energy-conversion loop, especially when Drive = Sine. The driving current waveform can be either the ideal sinewave or squarewave, but the i-psi Polygon GoFER may be more convenient for 3-phase square-wave drives; see page 44. The current waveform can be the ideal Static design waveform or the Dynamic design waveform including the effects of chopping in the inverter. When Drive = AC Volt, a sinewave current is used, with the amplitude and phase calculated in Static design. The i-psi GoFER is valuable for calculations with highly saturable machines, especially in cases where the synchronous reactances X d and X q may vary strongly with current, as for example in wound-field machines or PM machines with embedded magnets. The result is valid even when the windings are not sine-distributed. It also takes into account any variation in the level of saturation as the rotor rotates. In such cases the i-psi GoFER is the only rigorous method for calculating the torque. The i-psi GoFER obtains the variation of the phase flux-linkage R as the rotor rotates through one electrical cycle, with the ampere-conductor distribution correctly phased according to the rotor position. Its name comes from the symbols i for current [A] and R for flux-linkage [V-s]. As the rotor rotates, the current varies in synchronism according to its ideal sinusoidal or square waveform. In each phase the point (i, R) traces out a closed loop once per electrical cycle. The area enclosed by the i-psi loop is equal to the average energy conversion per cycle in each phase, and if this is accumulated for all the phases and divided by 2B, the result is the average electromagnetic torque; see Fig. 45. The torque computed from the i-R loop includes all the electromagnetic torque produced by the phase currents. This includes both the "magnet alignment" torque and the reluctance torque, but not the cogging torque. However, it should be noted that the main function of the i-R loop calculation is to determine the average electromagnetic torque. Since the average cogging torque over one cycle is zero, it has no effect on the required result. Since the flux-linkage R is the total flux-linkage obtained by integrating the vector potential along the entire winding, there is no reliance on the fundamental components or the "airgap" component of flux. For these reasons the i-psi calculation is more rigorous than the Torque versus gamma calculation; but it is considerably slower. The i-psi GoFER can also compute iron loss on an element-by-element basis as the rotor rotates. If the i-psi GoFER is run with zero current, it can be used to compute the phase EMF. Page 38 i-psi calculation PC-BDC GoFER 9.04 i-psi GoFER Files produced by PC-BDC file Test1.bd4. Test1.ltq (PC-FEA > 5 only) torque calculated from areas of i-psi loops for each current and angle. Test1.wfe Iron loss parameters. Test1.psidq Values of Psid (R d ) and Psiq (R q ) averaged over the cycle. Test1_Ix_Gy .ipsi 3 i-psi (i,R) information at each rotor position, for the x'th current and y'th value of gamma (or Th0). Test1_Ix_Gy.a Average vector potential A in each slot for each rotor position, for the given current and gamma (or Th0). Test1_Ix_Gy.frc Current in phase 1, together with force on a single tooth for each rotor position, for the given current and gamma (or Th0). Test1_Ix_Gy.trq Current in phase 1, together with torque calculated from Maxwell stress for each rotor position, for the given current and gamma (or Th0). Test1_Ix_Gy.lm Self and mutual inductances at each rotor position, for the given current and gamma (or Th0), calculated with the magnet "switched off" and the permeabilities of all elements frozen at their open-circuit values. Discontinued in Version 8. Test1_Ix_Gy.bload On-load waveforms of tooth and yoke flux-densities. Test1_Ix_Gy.et "Element table" data for each rotor position, from which iron loss and other quantities can be computed. This file can be large. To disable the writing of it, uncheck Element Table. Test1_Ix_Gy.phim Flux pulsation file used with FPLoss and NPHx. Test1.lst PC-FEA script listing. Test1.rep PC-FEA report file. Table 17 Files produced by i-psi GoFER 3 In the _Ix_Gy naming convention, x is the current index and y is the gamma (or Th0) index. PC-BDC GoFER 9.04 i-psi calculation Page 39 Figure 40 i-psi calculation GoFER options, page 2 i-psi GoFER options (2) The options on this page are the same as for the Single load point GoFER, p. 5. Note that the rotor starting position is fixed such that its reference axis is aligned with Ax1, the axis of phase 1. Page 40 i-psi calculation PC-BDC GoFER 9.04 Figure 41 i-psi calculation GoFER options, page 3 Figure 42 User dialog specifying the coordinates of a point at which the flux-density waveform will be recorded. i-psi GoFER options (3) Normally the i-psi calculation is performed with the ideal current waveform corresponding to Static design in PC-BDC. In this case the calculation can be performed for a range of evenly distributed set- point currents ISP and phase advance angles gamma. The numbers of calculations in these ranges are set with count values, and the ranges are set with min and max. Use TED values sets ISP and gamma to the template-editor values in the PC-BDC datafile. RotSteps is the number of steps taken through a rotation of 360Eelec. The default value is 45, but if the Dynamic design current waveform is used, many more steps may be required. The Transient solver typically requires at least 180. The same is true of accurate EMF calculations showing slot-ripple. If Dynamic design current is checked, only one i-psi calculation is performed, corresponding to the actual Dynamic design current obtained from PC-BDC. If DC current is checked, the current remains fixed. Use coil-side model is used only when S_Slot = GolfTee or MISlot, and Throw = 1. In this case each coil- side is defined as a separate region. In all other cases the combined ampere-conductors of all coil-sides are uniformly distributed over the slot area. Show graphs instructs PC-FEA to plot a graph of the i-R loop, and other graphs. Maxwell stress instructs PC-FEA to include the Maxwell stress calculation. As elsewhere, the square brackets indicate the requirement for care in processing these results. Mesh refinement may be necessary to achieve a satisfactory result. Force on tooth instructs PC-FEA to produce a file of the force on a single tooth. (See Socks in the template editor). Bt and By instructs PC-FEA to produce files of tooth and yoke flux-densities. User instructs PC-FEA to produce a file of the flux-density at a single point; see Fig. 42. Start = gamma + 180 causes the origin of the flux-linkage and current waveforms to be consistent with the phasor diagram, which is referred to Eq1 sin Tt If this is not checked, the origin is the same as that of PC-BDCNs time-waveforms, which are referred to ISP sin Tt. See Fig. 76 on page 58. PC-BDC GoFER 9.04 i-psi calculation Page 41 (1) (2) Element table instructs PC-FEA to produce a table containing the magnetic flux-density and field strength and other parameters for every element in the solution. This raw data can be processed for various purposes; for example, in PC-BDC Tools | FE Results | Elements Table computes the iron loss using a modified Steinmetz equation on an element-by-element basis. Incremental inductance instructs PC-FEA to compute a field solution with frozen permeabilities at each RotStep. This data is used to compute the incremental phase inductances in the *.lm file. It is important not to confuse these with the actual inductances. PhiM causes PC-FEA to write a file of the flux through one magnet block, in order to calculate its variation as the rotor rotates; see Tutorial B18. Transient (FEA 5.5) executes the transient solver in PC-FEA 5.5. See Tutorial B23. Cycles is the number of electrical cycles, each of RotSteps steps, through which the calculation will be performed. Cycles must be specified when Element Table and Transient are both checked. The Transient solver generally requires at least 3 cycles. If Element Table is not checked, Cycles is not required, and only 1 cycle will actually be calculated. If Element Table is checked, the rotation of the rotor must return a point on the rotor to a position that is magnetically equivalent to the starting position. Such a position can be defined, for example, as when the magnet centre-line is aligned with a slot centre-line. This can be expressed by the equation or where n and Cycles are both integers. For example, in a 3-phase motor with 12 slots and 8 poles, we have Pole-pairs ' 4 and Slots ' 12, so we get the required result with Cycles ' 1 and n ' 3. For a 3-phase 12/10 motor with Slots ' 12 and Pole-pairs ' 5, we need Cycles ' 5 with n ' 12. Provided that eqn. (2) is satisfied, only the value of Cycles need be specified; the value of n is then implicit. The condition expressed by eqn. (2) is not sufficient to guarantee that the rotor has returned to a position that is magnetically equivalent with respect to the winding. Because windings can be laid out with complete freedom, there is no general mathematical constraint that can provide this guarantee, and it is left to the user to check it. The GoFER will display a warning if Cycles is less than the value given by eqn. (2). A safe value is to set Cycles ' Pole-pairs (which implies that n ' Slots). Skewed Skewed is a special variant of the i-psi GoFER that performs multiple finite-element computations corresponding to a series of slices which are phase-shifted to reproduce the effect of skew. The resulting i-R loop data can be imported into PC-BDC (Tools | FE Results | MatchFE : i-psi loop) and compared with PC-BDCs internal calculation. The GoFER options pages allow the setting of the number of slices. The phase-shift can be effected either by phase-shifting the current waveform or by physically rotating one slice relative to the next. Page 42 i-psi calculation PC-BDC GoFER 9.04 Figure 43 i-psi calculation GoFER options, page 4 Figure 44 i-psi calculation GoFER options, page 5 : Meshing and Solving i-psi GoFER options (4) The options on this page are the same as for the Single load point GoFER, p. 9. The options on this page are the same as on p. 14, except for the check-box Use individual solution files (.mes). When this is checked each combination of rotor position, ISP and gamma values results in a separate .mes file; otherwise all solutions are combined in a single .mes file. In a large calculation the single .mes file can slow the post-processor in PC-FEA. PC-BDC GoFER 9.04 i-psi calculation Page 43 Figure 45 Example of i-psi calculation Figure 46 MatchFE i-psi : current waveforms Figure 47 MatchFE i-psi : flux-linkage waveform Example of i-psi calculation Fig. 45 shows the i-psi loop created with a standard [Alt+7] example in PC-BDC with high-energy magnets and sinewave drive. The PC-BDC calculation is from a Static design although Dynamic design could equally be used. The current waveform (Fig. 46) is forced to be sinusoidal, but because of the magnet flux-distribution and the winding layout, the flux-linkage waveform (Fig. 47) is not. As a result, the energy-conversion loop does not have a perfectly elliptical shape. Saliency, saturation, and permeance harmonics can distort the shape of the loop even further. The average electromagnetic torque can be checked from the area W of the ellipse, T = W 3 phases Pole-pairs /2B. In the example, the internally calculated value from PC-BDC is close to the finite-element value, but differences can arise especially in IPM motors where saturation may reduce the value of X q by a significant factor sometimes as much as 5:1. Adjustment factors such as XBrT, XCq, XCd, XL, XBrT, XLdiff and others can be used to modify the PC-BDC values to account for saturation. In general it is best to make adjustments with the flux-linkage waveform displayed, because the currents always match. For sinewound motors it may be helpful to use the Show fundamental option. The MatchFE function in PC-BDC can also display the voltage obtained from the flux-linkage waveform, and it displays suggested values for Xq and Eq1; (these refer to only the first file of finite-element data). The results should be compared with those obtained with the embedded finite-element solver; see Tutorials B14 and B19. Include end-effects applies the end-winding inductance from PC-BDC. Page 44 i-psi Polygon PC-BDC GoFER 9.04 Figure 48 i-psi diagram for 3-phase square-wave motor drive Figure 49 i-psi Polygon GoFER options, page 2 9. i-psi Polygon The i-psi Polygon GoFER is intended for 3-phase wye-connected square-wave drives. It is similar to the i-psi calculation GoFER, in that it uses the current and flux-linkage waveforms plotted as a closed loop in the i-R plane to determine the average electromagnetic torque. Fig. 48(a) shows the i and R waveforms and Fig. 48 (b) shows the i-R loop. The main options for the i-psi Polygon GoFER are similar to those of the i-psi calculation GoFER, but Fig. 49 shows the special options page for selecting which of the points A,B,C,... etc. are to be calculated. The more points selected, the slower will be the calculation. PC-BDC GoFER 9.04 Slot permeance Page 45 Figure 50 Slot permeance GoFER, page 1 10. Slot permeance The Slot permeance GoFER produces the slot permeance coefficient for a stator or rotor slot, for a given value of current-density in the slot. Slot permeance GoFER Files produced by PC-BDC file Test1.bd4. Test1.pcslot The file containing the slot-permeance coefficient and related parameters Test1.lst PC-FEA script listing Test1.rep PC-FEA report file Table 18 Files produced by Slot permeance GoFER Page 46 Slot permeance PC-BDC GoFER 9.04 Figure 51 Slot permeance GoFER options, page 2 Figure 52 Slot permeance GoFER options , page 3 Slot permeance GoFER options (2) Fig. 51 shows the setup options. Solution domain refers to one stator slot or one rotor slot. The boundary condition at the airgap is usually Dirichlet, meaning that no flux will cross the airgap; this is the normal assumption in analytical formulas for slot-permeance coefficient. The Neumann boundary condition allows flux to cross the gap. The Air-gap length can be assigned to the value in the *.bd4 datafile, or given an independent value. The depth of the solution domain can also be set in such a way as to avoid having an unnecessarily deep solution region in the back iron. The average current density in the slot can be specified in steps between Min J and Max J. Slot permeance GoFER options (3) The options on this page are the same as for the Single load point GoFER, p. 9. PC-BDC GoFER 9.04 Slot permeance Page 47 Figure 53 Slot permeance GoFER options, page 4 Figure 54Slot permeance GoFER, GDF editor Slot permeance GoFER options (4) The options on this page are the same as for the Single load point GoFER, Fig. 10 p. 14. Slot permeance GoFER GDF Editor The GDF editor functions in the normal way, as with all the other GoFERs. Press [F11] to continue to PC-FEA. Page 48 Slot permeance PC-BDC GoFER 9.04 Figure 55 Slot permeance GoFER, PC-FEA script Figure 56 Slot permeance GoFER, PC-FEA mesh Slot permeance GoFER PC-FEA script The PC-FEA script functions in the normal way. Press Run or [F9] to continue. Slot permeance GoFER PC-FEA mesh The PC-FEA mesh functions in the normal way. Press Continue or [F8]. PC-BDC GoFER 9.04 Slot permeance Page 49 Figure 57 Slot permeance GoFER, PC-FEA flux-plot Slot permeance GoFER PC-FEA flux-plot The solution appears at first as a flux-plot in PC-FEA, Fig. 57. In View listing the text file of results can be displayed. It has the filename extension *.pcslot. It contains a table of values, one row for each level of current-density specified in the GoFER options. The following parameters are given: J(A/mm^2) The specified current-density. Current(A) The total current in the slot. Energy(J) The magnetic stored energy (see PC-FEA manual). EnergyJA(J) The magnetic stored energy calculated from the integral of J@dA (see PC-FEA manual). CoEnergy(J) The magnetic coenergy (see PC-FEA manual). FluxLnkg The flux-linkage of the conductor in the slot, evaluated as 2 EnergyJA/Current. Page 50 Slot permeance PC-BDC GoFER 9.04 BridgeFlux The flux in any bridge (closed rotor slots only), evaluated from the difference between A1 and A2. A1 The vector potential at the slot centre-line, at the radius Rad1. A2 The vector potential at the slot centre-line, at the radius Rad1 ! R_Bridge. PCSlot(old) The slot-permeance coefficient evaluated as 2*$Energy/[Mu0]/(Jslot*condarea*Jslot*condarea), where [Mu0] is the constant : 0 , Jslot is the current-density, and condarea is the conductor area.
PCSlot The slot-permeance coefficient evaluated as FluxLnkg/Current/[Mu0]. This is the preferred interpretation of the slot-permeance coefficient. PC-BDC GoFER 9.04 Errors and program failures Page 51 Figure 58 Mesh angle error in PC-FEA's mesh generator Figure 60 (a) Poor aspect ratio (b) Good aspect ratio 11. Dealing with errors and program failures 11.1 Meshing errors arising from a poor initial distribution of nodes Most meshing errors arise from a poor initial distribution of nodes at the sub-domain boundaries. The most usual solution is to adjust the boundary node distribution in the GDF Editor, but many adjustments to the mesh generator can be made in PC-FEA itself, either in the script or in the program interface; see the PC-FEA Manual for details. The initial distribution of boundary nodes is hard-coded in PC-BDC and this distribution is first seen in the GDF Editor even before PC-FEA is run. The distribution is set up to give good results with "standard" motor geometries, so the probability of failure is increased in cases where the geometry is far from the standard one. Of course the whole purpose of finite-element analysis is to analyze any geometry, but because the SPEED software always starts from standard templates and does not use adaptive meshing, adjustments may be necessary when the geometry is far from the "standard". A typical error arising in PC-FEA's mesh generator is shown in Fig. 58 4 , and the failed attempt at meshing is shown in Fig. 62. In this instance the error message is fairly friendly because it states the problem in terms that are reasonably clear, and makes one or two suggestions as to what to do next. In this example Do optimization and Include virtual lines and arcs (p. 14) were both switched off, but a second attempt at meshing was made in PC-FEA by pressing [OK] on the Optim page of the mesh generator. This attempt failed with the less friendly error message in Fig. 59. The boundary node distribution that caused the problem is shown in Fig. 61, where two adjacent "faces" have widely different node distribution densities. Fig. 60 shows the problem in more detail. The triangular elements near the corner will initially be generated with poor aspect ratio as in Fig. 60(a). For reference, Fig. 60(b) shows the much improved aspect ratio that results from a better initial node distribution. Figure 59 Wrong triangle error in PC-FEA 4 In previous versions the Mesh18 error was entitled "TRIDIM". A similar error was entitled "KVT". Page 52 Errors and program failures PC-BDC GoFER 9.04 Figure 61 Bad node distribution Figure 62 Mesh failure, with "skinny" triangles, holes in the mesh, and leakage PC-BDC GoFER 9.04 Errors and program failures Page 53 Figure 64 Solution of the mesh in Fig. 63 Figure 65 Mesh optimized by PC-FEA Figure 66 Mesh with improved initial node distribution Fig. 63 shows the effect of setting MaxAngle = 120E, resulting in a successful mesh in spite of the poor initial node distribution. MaxAngle can be edited in the PC-FEA script directly, or it can be pre-set in the GoFER options, Fig. 10. PC-FEA can actually solve this mesh, in spite of the "lousy mesh" in Fig. 63, and the flux-plot arising from the solution has all the appearance of an acceptable solution, Fig. 64. But it is bad practice to content oneself with "apparently acceptable results" and no reliance should be placed on the solution of such a poor mesh as Fig. 63. In this example the poor mesh gives a stark visual warning of the weakness in the modelling. Much better results are obtained with Do optimization and Include virtual lines and arcs both enabled. Even with no adjustment to MaxAngle, an improved mesh is achieved as shown in Fig. 65. Even this mesh has a few "skinny" triangles near the corner between the two faces marked in Fig. 61, and what is really needed is to go back to the GDF Editor and change the "33 divisions" to something more like 5 or 6, and the "1 division" to something more like 812. Fig. 66 shows the effect with 8 and 12 divisions respectively. Many further refinements are possible. The solution of the mesh in Fig. 66 has significant differences of local flux-density compared with the solution shown in Fig. 64. Figure 63 Successful meshing, but with poor aspect ratios Page 54 Errors and program failures PC-BDC GoFER 9.04 Figure 67 Impossible geometry Figure 68 The result of trying to analyze Fig. 67 with PC-FEA Figure 69 Use Task Manager to close PC-FEA Figure 70 Impossible geometry: overlapping sub-domains 11.2 Meshing errors arising from impossible geometry It is possible for PC-BDC to draw motor cross-sections that are physically impossible to make. Any CAD program can do the same. Fig. 67 shows an example. Although PC-BDC contains many checks on the "validity" of input data, it is virtually impossible to trap all possible instances of "impossible geometry". It is actually undesirable in the template-based outline editor to have too many checks, because they would prevent the engineer from seeing what is wrong. To an engineer, a drawing such as Fig. 67 is worth any number of computer-generated error messages such as Fig. 68. The operating system may try to continue even after the error message in Fig. 68 is cleared. It is replaced by the one in Fig. 69, which is hard to remove other than by using a system utility to close PC-FEA. Every Windows user will be familiar with these error messages. They can arise with PC-BDC as well as PC-FEA. In the case in Fig. 67, PC-BDC blithely produces a result which has all the appearance of being credible. It is one of the cases that escapes the validity-checking. But it should never have been tried in the first place, should it? Although all this is obvious, there are subtle cases where the geometry might appear acceptable at first sight, but the "impossible topology" is in fine details that can be seen only under a high magnification (zoom); see Fig. 70. In such cases the problem might not show itself until PC-FEA is run. PC-BDC GoFER 9.04 Errors and program failures Page 55 Figure 71 Epsilon 11.3 Precision problems Precision problems can arise with very small airgaps or very large motors, as a result of necessary tolerances which are used to "capture" boundary nodes and associate them with particular sub-domains. Occasionally the Include virtual lines and arcs option can itself have problems if the translation from line segments to arcs defined by centre, radius and angles is not sufficiently precise. When the mesh is being optimised new boundary node points (.bnd points) can then be added along the wrong arcs, causing the mesh to fail. Such cases are more likely to arise with very large motors. The solution is to uncheck Include virtual lines and arcs or to delete the corresponding lines in the PC-FEA script. The Wrong PC condition arises from meshing problems and incorrect boundary conditions, including periodic boundary conditions. It will arise, for example, in trying to solve a 4-pole, 15-slot motor with a solution domain of ony 1 pole, which on the stator corresponds to 3 slots. Convergence problems can sometimes be relieved by increasing the Solver Tolerance (p. 15). 11.4 Errors connected with rotating meshes Wrong list of boundary edges; this mesh can't be rotated 5 means that PC-FEA cannot find a suitable set of nodes around the airgap to create the rotating mesh. For problems requiring multi-position solutions (Btooth, Cogging, i-psi), PC-FEA starts with separate rotor and stator meshes and has to identify the nodes at the opposing surfaces, (which are nominally at RadGapI and RadGapO in the PC- FEA script). It includes nodes that are within a band defined by RadGapI Epsilon for the rotor, and within a band defined by RadGapO Epsilon for the stator. The parameter Epsilon in the [Rotate] section of the PC_FEA.INI file (and in the PC-FEA Options | Settings menu) can be used to adjust this band width. Depending on the Node Tolerance, if Epsilon is too small, some nodes will be missed; if it is too big, extra nodes will be included. As a rough estimate Epsilon should be smaller than the distance between the rotating rotor surface and the nearest "non-surface" node i.e. all the surface points should lie in a band of radius Rad1Epsilon to Rad1 Epsilon, as in Fig. 71. The Wrong list of boundary edges problem is more likely to arise if Epsilon is large. This in turn can happen if the GoFER has previously been working on very large motors where the node tolerance has been increased, and then returns to work on a small motor with the same value. 5 In previous versions this error message was entitled "Ordonedges". Page 56 Errors and program failures PC-BDC GoFER 9.04 11.5 Errors connected with material properties Most errors arising from material properties are a result of irregularities in the BH data : for example, if the data is non-monotonic, or if it does not extend to a sufficiently high value of B, or does not start from zero; or if the spline interpolation introduces spurious irregularities. If the guidelines for preparation of BH data in the WinSPEED manual are followed, these errors will be avoided. Particular care is needed with cobalt-iron alloys which have very square BH curves. PC-BDC GoFER 9.04 Theory Page 57 Figure 74 Phase 1 amp-conductors when the current is in the q-axis: I = jI q and gamma = 0E Figure 72 Rotor reference axis aligned with the axis of phase 1, Ax1 Figure 73 Phase 1 amp-conductors: the current is in the d-axis: I = I d j0. Figure 75 Phasor diagram showing the definition of gamma (() 12. Theory 12.1 Reference axes for the rotor position, the current waveform, and the phasor diagram The rotor reference axis is a S d-axis, i.e. the centre of a south magnetic pole at the rotor surface. When this is aligned with the axis of phase 1, the open-circuit flux-linkage in phase 1 has its maximum positive value, +R pk . The phasor diagram also takes its reference from this position. As shown in Fig. 72 for a 4-pole motor, Q ' Q Md j0. Since the phasor diagram always uses r.m.s. values and assumes sinusoidal time-variation, R pk ' Q Md //2. The q-axis leads the d-axis by 90Eelec. In PC-BDC and PC-FEA, a "Go" conductor has a positive value in the conductor location vector, and points out of the paper, i.e., in the positive z direction; a "return" conductor points into the paper. The single coil representing phase 1 in Fig. 72 is wound such that positive current produces a radially inward MMF aiding the magnet, Fig. 73. This condition arises when the phasor current is in the d-axis, as defined by the phasor diagram in Fig. 73, so I = I d + j0. The axis of the coil in Figs. 72!73 is at the angle Ax1, measured from the x-axis. In Fig. 74, the current is "in the q-axis", I = 0 + jI q . The q-axis leads the d-axis, both being fixed to the rotor. Therefore in Fig. 74 the rotor position is 90E ahead of (earlier than) its position in Fig. 73, and the q-axis is aligned with Ax1. Note the direction of torque in Fig. 74. In Fig. 73, the torque is zero. In Figs. 7274 the instantaneous d- and q-axis currents i d and i q are denoted by lower-case letters, while the phasor values are denoted by upper-case boldface letters. The r.m.s. components of the phasor values are denoted by upper-case italics, I d , I q . The relationships between the instantaneous values and the phasor values are developed as follows.
Page 58 Theory PC-BDC GoFER 9.04 (3) (6) (8) (7) Figure 76 Instantaneous current, magnet flux-linkage, and EMF (4) (5) When Drive ' Sine, the phasor current I is defined in terms of its peak value i pk (ISP) and the phase advance angle ( (gamma), as defined in Fig. 75. In phasor terms, where I, I d and I q are r.m.s. values and I ' ISP//2. By convention, the instantaneous current i 1 in phase 1 is the projection of the phasor I on the x-axis as it rotates in the positive (anti-clockwise) direction. Likewise the instantaneous magnet flux-linkage R 1 is the projection of the phasor Q. If the time-origin t ' 0 is the instant when the rotor is aligned with phase 1 and R 1 ' R pk (as in Fig. 72), then and it follows that This choice of time-origin is used in the i-psi GoFER, which uses eqn. (5) to assign the currents. But it would cause the current waveform to appear with a strange-looking phase angle, especially if ( is non- zero. In any case the magnet flux-linkage R 1 is not observable under loaded conditions, making it unsuitable as a reference waveform for viewing the waveforms of current, EMF, etc. In PC-BDC the time-origin is taken at a positive-going zero of i 1 : thus It follows from Fig. 75 that the magnet flux-linkage R 1 lags behind i 1 by the angle ( 90E, so that These relationships are illustrated in Fig. 76. The expressions for i 2 , and i 3 are similar, but with arguments of Tt 120E and Tt 120E respectively; for R 2 and R 3 the arguments are Tt ( 120E. Whereas PC-BDCs time-origin is a zero of i 1 , the time-origin of the phasor diagram and PC-FEA is a positive peak of R 1 . It therefore appears that the waveforms of phase currents and phase flux-linkages in PC-FEA are phase-retarded by (( 180)E relative to those used by PC-BDC. This angle is shown explicitly in Fig. 76. It means that waveforms of i 1 , i 2 , i 3 , R 1 , R 2 and R 3 computed in PC-FEA must be phase- advanced by (( 180)E before comparing them with PC-BDCs waveforms. As the rotor rotates, the flux-linkage in each phase varies and generates an EMF. If the variation of R is sinusoidal, then for phase 1 (using PC-BDCs time-origin) we can write When ( ' 0, the EMF e 1 is in phase with the current i 1 . With phase advance, ( is normally positive, making i 1 lead e 1 by (. Note that eqn. (8) uses the sink sign convention for induced EMF. PC-BDC GoFER 9.04 Theory Page 59 (9) (10) (14) (11) (12) (13) Figure 77 Reference angles for dq transformation 12.2 Instantaneous dq components The relationship between instantaneous d,q components i d and i q and the instantaneous phase currents i 1 , i 2 and i 3 is the dq-transformation: where 2 is the angle between the rotor d-axis and the axis of phase 1; (see Fig. 77). The inverse is With the phasor diagram time-origin corresponding to Fig. 75, as used in the PC-FEA i-psi GoFER, The phase currents in the PC-FEA i-psi GoFER are obtained from eqn. (5) along with the corresponding equations for i 2 and i 3 . The corresponding dq currents are obtained by substituting these currents in eqns. (9), with 2 given by eqn. (11). The result is These relationships are consistent with the phasor diagram, in which The flux-linkages R 1 , R 2 and R 3 computed by PC-FEA can be transformed into R d and R q in the same way, using R 1 , R 2 , R 3 in place of i 1 , i 2 , i 3 in eqn. (9), with 2 given by eqn. (11). But for comparison with the internal waveforms of PC-BDC, all the waveforms i 1 , i 2 , i 3 , R 1 , R 2 and R 3 must be advanced in phase by the angle ( 180E, as shown in Fig. 76. This phase advance is applied by the i-psi GoFER. On the other hand, if the phase currents are expressed relative to the PC-BDC time-origin, as is the case with the embedded solver, the phasor diagram appears as in Fig. 78. This is obtained by phase- advancing Fig. 75 (i.e. rotating it clockwise)by the angle ( 180E. As we have already seen, the maximum value of R 1 occurs ( 180 Eelec after the positive-going zero of i 1 , so the rotor position is given by Page 60 Theory PC-BDC GoFER 9.04 (19) (16) (17) (18) (15) Figure 78 Phasor diagram with PC-BDCs time-origin corresponding to eqns. (6), that is, at a positive-going zero of i 1 If eqns. (6) and (14) are substituted into eqns. (9), we get exactly the same as before. This shows that the values of i d and i q are not sensitive to the time-origin, provided of course that the correct value of 2 is used in the dq transformation, eqns. (9). The same is true for the flux-linkages. The phase flux-linkages R 1 , R 2 and R 3 computed by PC-FEA can be transformed into R d and R q in the same way, again using eqn. (9) but with 2 given by eqn. (14). For comparison with the internal waveforms of PC-BDC, there is now no need to phase-advance the waveforms of i 1 , i 2 , i 3 , or of R 1 , R 2 and R 3 . This is the process that is required by the embedded solver. Internally, when working in dq axes with Drive ' Sine, in its Dynamic design calculation PC-BDC uses eqns. (6) for the currents, but instead of eqn. (14). Consequently all the internal dq variables i d , i q , R d and R q in PC-BDC are inverted; but not the ordinary phase variables i a , i b , i c , R a , R b , R c . When using the embedded solver, 2 is passed directly to PC-FEA, which in effect is being told that the rotor is 180Eelec out of phase. To correct this, PC-BDC must pass negative values of i a , i b and i c to the embedded solver. It then inverts the values of R a , R b and R c returned by PC-FEA. as well as those of R d and R q . A useful instantaneous check on the values in the PC-FEA script is at the instant t ' 0. With the i-psi GoFER, 2 ' 0 and then the rotor position in PC-FEA will be such as to align a S d-axis with the axis of phase 1, and since the time-origin is according to the phasor-diagram, If ( ' 0 this becomes With the embedded solver, the currents are prescribed by PC-BDC, regardless of the rotor position, as which is the opposite of the values in the i-psi GoFER. However, the embedded solver passes the negatives of the phase currents to PC-FEA, so the values will be the same as in eqn. (18). PC-BDC GoFER 9.04 Theory Page 61 (20) (21) (22) In the finite-element calculation, the ampere-conductors in each slot are obtained from the CLV (conductor location vector C) and the instantaneous phase currents i 1 , i 2 and i 3 , according to a relationship of the form where AC[j] is the ampere-conductors in slot j, and m is the number of phases. 12.3 Cogging torque In PC-FEA, four approaches are used in computing the cogging torque component. Two of them are summarized here; the others can be found in [5] and [6]. 1. The Maxwell stress tensor is calculated on a surface in the air-gap containing the rotor, for each rotor angular position: where L stk is the stack length and R is a radius within the airgap. This integral is evaluated twice: T 1 (rotor-side integral) is computed at R = Rad1 + 3 Gap/8, between angles TqAngMin and TqAngMax, and T 2 (stator side integral) at R = Rad1 + 5 Gap/8, between angles 0 and AngFE. Then the average Maxwell torque is taken as (T 1 + T 2 )/2. The calculation can be seen in the Computations section of the PC-FEA script. Note that PC-FEA uses four layers in the air-gap. In some cases, it is recommended to try different values for the radii, for example, Rad1 + Gap/2 for both of them. The angle limits should not be modified, as they are set to agree with the solution domain. PC-FEA uses first-order triangular elements, so the vector potential A is more accurate than the distributions of B and H, which are obtained by differentiating the trial functions for A. While A is described by a linear function over each element, B and H are piece-wise constant over each element. The Maxwell stress should not be evaluated in a layer adjacent to a boundary between materials of different permeabilities. 2. The virtual work method calculates the torque from the rate of change of the magnetic co-energy with rotor position: In PC-FEA, this method is programmed according to the theory in [1]. The co-energy method can be identified as TqCoEnergy in the PC-FEA script file, Computation Section. For both methods described above, the user-selected solution domain is reflected in the value of AngFE, while RotSteps is the number of steps of rotor rotation. The accuracy is essentially independent of these variables, provided RotSteps is not too small, but the computation time increases if RotSteps is increased or if a larger solution domain is used. Page 62 References PC-BDC GoFER 9.04 13. References 1. SPEEDs Electric Machines 2. PC-BDC Reference Manual 3. PC-FEA Reference Manual 4. WinSPEED Manual 5. Popescu M, D.M Ionel, TJE Miller, S.J. Dellinger and M.I. McGilp Improved finite element computations of torque in brushless permanent magnet motors, IEE Proc-Electr. Power Appl., Vol. 152, No 2, March/April 2005, pp. 271-276. 6. Ionel D M, Popescu M, McGilp MI, Miller TJE and Dellinger SJ Assessment of torque components in brushless permanent-magnet machines through numerical analysis of the electromagnetic field, IEEE Transactions on Industry Applications, Vol. 41, No. 5, September/October 2005, pp. 1149-1158. 7. SPEED Tutorial B14 : Interior Permanent-Magnet Motor (IPM): Calculating torque and saturation using the i-psi GoFER. 8. SPEED Tutorial B18 : Magnet losses in PC-BDC/PC-FEA. PC-BDC GoFER 9.04 Index Page 63 Index Accuracy of torque calculation . . . . . . . . . . . . . . . . 8, 25 Airgap layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Areva T&D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Ax1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Bgap distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 8 BH button . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 BLV method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Br . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 BrTEff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 BrTEff/XBrT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 ByokeAngle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Calculate Bt, By (sections) . . . . . . . . . . . . . . . . . . . . . 8 Calculate torque . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Carter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Conformal transformation . . . . . . . . . . . . . . . . . . . . . 8 Create file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8, 22 DC current checkbox . . . . . . . . . . . . . . . . . . . . . . . . . 40 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Do optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Element table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38, 41 in Single Load Point GoFER . . . . . . . . . . . . 8 Embedded solver . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 3 EMF waveform . . . . . . . . . . . . . . . . . . . . . . . . . . . 16, 20 EMFCalc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16, 20 End-winding inductance . . . . . . . . . . . . . . . . . . . . . . 43 Epsilon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Faces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Files produced by Bgap GoFER . . . . . . . . . . 16, 45 produced by Btooth GoFER . . . . . . . . . . . . 20 produced by Cogging Torque GoFER . . . 25 produced by i-psi GoFER . . . . . . . . . . . . . . 38 produced by Single Load Point GoFER . . 4, 7, 10 produced by Torque/gamma GoFER . . . . 32 gamma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 GDF Auto . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 GDF v 1.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 GoFER Bgap distribution . . . . . . . . . . . . . . . . . . . . . 16 Btooth waveform . . . . . . . . . . . . . . . . . . . . . 20 Cogging torque . . . . . . . . . . . . . . . . . . . . . . . 25 definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 i-psi calculation . . . . . . . . . . . . . . . . . . . . . . 37 operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Single load point . . . . . . . . . . . . . . . . . . . . . . 4 Torque versus gamma . . . . . . . . . . . . . . . . 31 IES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Include shaft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Include virtual lines and arcs . . . . . . . . . . . . . . . . . 14 Infolytica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Interpolation of BH data . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 ISP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 J-MAG Studio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Layers of airgap elements . . . . . . . . . . . . . . . . . . . . . 6 Load point scale factors . . . . . . . . . . . . . . . . . . . . . . . . 7 MagNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Magneto . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 MatchFE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 16, 20 Max. angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Max. iterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Max. nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Maxwell stress . . . . . . . . . 4, 8, 25, 29, 32, 34, 38, 40, 61 Mesh refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . 8, 25 Node tolerance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Nodes max. number of . . . . . . . . . . . . . . . . . . . . . . 14 Other finite-element programs . . . . . . . . . . . . . . . . . 3 Pause . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 PC-FEA meshing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 PC-SRD adjustment . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 using finite elements with . . . . . . . . . . . . . . 1 PhiM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Pivot failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Pre-convert to SD1 format . . . . . . . . . . . . . . . . . . . . . 15 Profiled rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 RadBGap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 RadBtooth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Reference axis of rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Remanence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Rotor reference axis . . . . . . . . . . . . . . . . . . . . . . . . . 3 Rotor Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 RotorAng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 RotSteps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 S d-axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Save design file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 SD2 format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Show distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Skew cogging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Skewed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 SLIM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Slot modulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Slot permeance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Solution domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 5 Solver tolerance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Temperature of magnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Th0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Torque accuracy of . . . . . . . . . . . . . . . . . . . . . . . . 8, 25 Transient solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Use template editor values . . . . . . . . . . . . . . . . . . . . . 7 User-defined angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Vector potential file in Btooth GoFER . . . . . . . . . . . . . . . . . . 22, 24 Warn on GDF file overwrite . . . . . . . . . . . . . . . . 10, 13 Waveform definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3