Reluctance Motors Synchrel Design & Optimisation A Switched Reluctance Alternative Incorporating Novel Features The End Result 1
Existing Design Procedure Electromagnetic Design A Switched Reluctance solution to the (H)EV target, developed by Jaguar Land Rover and addressing common problems with the architecture. Design and priorities As both Synchronous and Switched Reluctance topologies had been identified as effective solutions to the problem, JLR undertook design of a Switched Reluctance machine alongside Ricardo s Synchronous Reluctance solution. As it was assumed this machine would have no difficulty meeting the target torque and weight targets, a greater focus was put on noise and vibration, and torque ripple reduction strategies. Targets Torque @307V Torque @340V Power @307V Power @340V 2
Existing Design Procedure Stress and Thermal Design Deployment of stress and thermal tools and their effects on cost and manufacture of the designed machine. Stress and Thermal approach At each stage of the design iteration, the model underwent stress and thermal analysis not only to confirm the design is mechanically sound but to predict machine lifetime and high-end performance capability. Early changes to the design frequently have the greatest impact on possible manufacture techniques so a forward-thinking approach can reduce production costs considerably. Conservative designs were in this way identified early, which allowed for a stronger approach in electromagnetic analysis, resulting in a more efficient machine. 3
Existing Design Procedure Iterative Process Between Tools The initial design process used on this project involved an iterative approach between the electromagnetic, stress and thermal tools. design th0 thc Bspd Trq weight power Irms fill Jrms 4 18 38.5 4200 237 42.1 104.2 300 0.5 17.64 4.01 16.8 38 4400 227 46 104.6 300.5 0.338 17.68 4.02 16.8 38 4200 238 48.5 105 300 0.338 17.65 4.03 17.5 38.5 3900 252 50.7 103.2 299 0.338 17.59 4.04 17.3 38.3 4000 248 49.6 103.8 300.5 0.338 17.68 4.05 18 38.5 4100 236 39 101.3 299.6 0.27 32.73 4.06 18 39 4250 233 43.5 104 300.5 0.433 17.68 4.07 18 39 4200 236 41.9 103.7 301.8 0.334 23.08 4.08 16 37.5 3900 263 47.1 107.3 298.4 0.387 17.56 4.09 17.5 40 3000 242 48.9 75.3 223 0.484 13.11 4.1 18 39 3300 215 44.1 74.3 213 0.484 12.56 4.11 15 38 3250 224 44.3 76.3 221 0.491 13.71 4.12 15 38.5 3600 224 53.3 84.4 243 0.376 12.4 4.13 15 38.5 3400 242 53.4 86.1 246.9 0.391 12.6 4.14 16.5 39.5 3000 265 53.8 83.1 240.3 0.451 12.25 4.15 16 38.5 3400 231 48.4 82.1 234.7 0.451 11.96 4.15 15 38.5 3750 264 48.4 103.4 286 0.451 14.6 4.16 16 39.5 3300 260 46.3 90 254.6 0.511 14.98 4.17 16 39 3300 253 56.2 87.5 243.1 0.495 9.78 4.18 16.5 39.3 3600 252 45.5 95.2 271.7 0.46 15.98 4.19 18 39 3400 238 45.6 84.8 243.4 0.499 14.32 4.2 16 39 3600 236 42.6 88.9 250.6 0.511 14.7 4.21 16 38.5 3300 267 48 92.3 252.6 0.5 13.8 4.22 15 38.5 3400 250 55.7 88.9 254 0.45 13 4.23 19.5 39.5 3300 226 48.1 78.2 226.6 0.485 13.3 4.24 15 38.5 3750 267 49.1 104.9 291 0.476 14.8 136 3050 43.41 design th0 thc speed T weight power Irms fill Jrms Sat 7J T 7J spd Pcont 5.01 15 39 3900 248 47.7 101.4 296 0.429 15.1 17 126 3200 42.2 5.02 15 38.5 3900 244 49 99.9 278 0.435 14.1 17 131 3200 43.87 5.03 14.5 38 3800 250 49.1 99.5 276 0.438 14 17 135 3200 45.21 5.04 16.5 40 3600 247 48.4 93.2 275 0.454 14 17 123 3200 41.19 5.05 15 38.5 3700 262 47 101.4 280 0.422 14.2 18 138 3050 44.05 5.06 16.5 39 3400 274 48.4 97.9 274 0.497 14 18 145 2900 44.01 5.07 15.5 39 3700 252 48.8 97.7 275 0.47 14 18 139 3000 43.64 Rough design Trends Using multiple software sources provides greatest accuracy of results, but the process is slow and labour intensive. Early optimisation was completed in batches within each software, gradually improving upon all variables. It is difficult to prove a given electromagnetic design is optimal with this method, as thermal and mechanical tools will usually offer improved alternatives. Detailed design optimisation must be done within a single unified system. 4
Reluctance Motors Synchrel Design & Optimisation A Switched Reluctance Alternative Incorporating Novel Features The End Result 5
Design score Incorporating Novel Features Weighted Optimisation Utilising Opera s Optimizer to improve the electromagnetic design process. 120 Design weighting and parameterisation By building a parameterised model of the best iterative design, a composite optimisation process was constructed, utilising Opera s Optimizer for the electromagnetic optimisation. Designs were weighted using saturation profile, to allow efficiency monitoring whilst maintaining fast simulation speeds. 100 80 60 40 20 0 9 14 19 50 70 90 radius (mm) 6
Torque Radial Force Incorporating Novel Features Reducing Noise and Vibration A novel approach to the Switched Reluctance problem and utilisation of Opera s toolset to optimise it fully. Flux Shaping One of the greatest issues affecting Switched Reluctance Machines is the noise produced by casing vibration from radial forces at alignment. By introducing voids in the laminations, it is possible to adjust the flux paths, and hence shape the torque produced during alignment. Peak torque is reduced ~20% but overall torque only ~9%, while the radial force is reduced dramatically. As a side effect the machine becomes bias in one direction, however for traction applications a higher noise in the reverse direction is expected. 0 10 20 30 40 50 22.5 32.5 42.5 52.5 62.5 7
Torque Incorporating Novel Features Additional Advantages Effects of the novel approach on machine torque ripple and the advantages thereof. Torque ripple improvements 0 10 20 30 40 50 Machines designed in this way can also be optimised for a reduced torque ripple. Although still higher than that of a Synchronous Reluctance machine or PM, this approach will stack with control strategies and should greatly improve low-speed performance. Reducing the torque peaks will prevent damage to components further down the line, and by boosting torque in trough regions, the machine can avoid dropping below target at any vehicle stall positions. 8
Radial Force Incorporating Novel Features Optimisation Optimisation of the identified novel strategy, and deployment within Opera for best utilisation in the design. Novel feature limits rcs circular rectangular no punch best worst mix By adjusting the optimisation solution used in the machine sizing, the trend between torque loss and radial force reduction can be plotted. Square, circular and smoothed rectangular voids were analysed, at all positions from the air-gap and tooth edge, and at various angles of rotation. Given the trend plotted, it can be assumed that further optimisation of the shape would provide only incremental benefit. Torque 9
Max. Stress Incorporating Novel Features Stress Analysis Improvements to mechanical properties of the machine, and point of diminishing returns. Max. Stress vs Increase in back-iron Stator outer radius Feasibility and further improvements The optimised lamination feature underwent stress analysis to confirm mechanical feasibility, after which further steps were taken to improve upon the electromagnetically optimised noise and vibration profile. Coil retention wedges were thickened and embedded into the stator teeth, a measure which was structurally equivalent to adding 5mm diameter (or 5kg) to the outer diameter. Running a sensitivity analysis on stator back iron thickness, the point of max stress could be seen to shift between zones. Capitalising on this resulted in a stiffer and less risky design. 10
Incorporating Novel Features Winding Design Using Opera to investigate eddy current losses to identify suitable winding strategies. Design for manufacture With the laminations finalised and the thickened coil retention wedge confirmed, eddy-current analysis was carried out to determine optimal winding strategy. Early manufacturability based decisions regarding solid conductors were overturned in favour of twisted strands, and a fitting tool designed to maintain production capability. 11
Reluctance Motors Synchrel Design & Optimisation A Switched Reluctance Alternative Incorporating Novel Features The End Result 12
The End Result Jaguar Land Rover s Design Solution Machine build and final design metrics against targets. The new Rapid SR motor Machine assembled similarly to Ricardo design. Laser cut laminations, rotor assemble and balance, stator housing shrink-fit, winding fit and varnish impregnation. Due to changes in EM and mechanical design to improve noise and vibration, active weight of built machine measured at 63kg (up from 50kg pre NVH optimisation). Total machine weight 79kg. New projects are already reducing this considerably whilst maintaining the NVH reduction. Despite this, the high power capability of the machine still results in a specific power of over 1500W/kg. 13
Torque Torque The End Result Results and Correlation Comparison of test results with simulation. Torque-angle curves 23 28 33 38 43 48 Measurement of the machine on the rig provided good correlation with 2D simulated results (less expected torque, compensated for here). Torque-angle shape is well maintained, especially important in this design! Torque ripple at low speed ~15% Compared to simulations ~11% Compared to baseline ~35% 0 10 20 30 40 50 14
s Time db g The End Result Noise and Vibration Analysis Validation of the novel design strategy s effectiveness, by NVH measurement on the casing, and subjective analysis. 30.00 AutoPower Point3:+Z WF 61 [0-30 s] -60.00 Results and Continuing Work Final validation of the various strategies to reduce noise and vibration, has been measurement of casing vibration of the machine in operation on the rig, which is ongoing. Noise measurement has not been measured objectively given the limitations of the current test area, a suitable location for testing should allow noise measurement in late 2015. Following on from the low noise EM and structural design of the machine, established and innovative techniques for noise improvements in software are now being investigated. These will constitute a follow-on paper. 0.00 0.00 Hz 1000.00-120.00 15
16