Click Image to Enlarge
Fig. 4—Non-random winding minimized the proximity effect. The slot cross section shows non-random placement of conductor strands.
Fig. 1—This design for a 75-hp spindle is the result of efforts to increase metal removal rate in cast iron and aluminum machining.
Fig. 3—The stator winding resistance ratio Rac/Rdc for earlier and later motor designs shows the extent to which "proximity effect" losses were reduced.
The pursuit of more productive machining invariably requires a re-examination of spindle technology. Spindle performance directly determines the efficiency of the machining process.
A shared interest in improving spindle tech-nology united the companies listed on the facing page. One of the goals of their shared project, led by the National Center for Manufacturing Sciences (NCMS), was to double metal removal rates during the machining of cast iron and aluminum.
The team recognized that the component virtually dictating the size and weight of a spindle is its motor. Making a motor more compact increases the challenge of temperature control. In fact, two performance factors limiting the design of advanced spindles are both temperature-related. They are the bearing lubricant coking temperature and the melting temperature of the motor winding insulation.
As spindle speed and power increase within a compact design, remaining within these temperature limits becomes increasingly difficult. But practical solutions can still be found. The 75-hp machining spindle prototype illustrated in Figure 1 demonstrates this. Supported on roller bearings, this spindle proved capable of deep cutting in cast iron at 2,500 rpm and rapid machining of aluminum with power to spare at 12,000 rpm.
The motor that delivers this performance is no bigger than a one-gallon paint can. Heat dissipation throughout the spindle was the major obstacle to accessing this motor's full power. To overcome this obstacle, the spindle design incorporated major advances. Jet-lubricated roller bearings enable the spindle to operate at twice the speed originally specified by the manufacturer for this class of bearings. Also, a motor winding innovation overcomes a heat source previously overlooked by all motor designers.
Torque And Speed Capabilities
The 75-hp spindle demonstrated its high-torque capabilities by deck milling cast iron engine blocks and boring their eight cyl-inders at twice the metal removal rate ac-cepted in current production. It demonstrated its high speed metal removal capabilities by milling an aluminum head at 12,000 rpm while delivering up to 95 hp. Relative to a comparable spindle currently used in production, the 75-hp spindle delivered 40 percent more torque at 2,500 rpm and 220 percent more torque at 10,000 rpm, while offering triple the con-trolled speed range and a 20 percent higher maximum speed.
Achieving this level of performance involved innovation and at least one surprising discovery.
Chilled water circulates through a jacket surrounding motor and bearings both. The jacket keeps bearing temperatures below 100°C and the motor winding temperature below 150°C. Test instrumentation confirmed the effective-ness of the jacket in cooling the bearings, but motor temperatures indicated a significant discrepancy between predicted and measured values. At a winding temperature of 150°C, the delivered torque was 30 percent less than the design torque, and the heat gene-rated in the stator was 2 kW more than was forecast. The search for the source of this heat revealed the presence of an electromagnetic loss phenomenon identified as the proximity effect—a condition known in the transformer field but long ago discounted in motor design.
Bearing manufacturer Timken Company is quoted in NCMS's report, An Evaluation of the Design and Performance of Three Advanced Spindles. According to a Timken source, "It is highly likely that the NCMS spindle is the highest speed tapered roller bearing spindle in the world." The bearing operates at a DN (mean diameter in milli-meters times the top speed in revolutions per minute) of 1.4 million—very high for this type of bearing.
The nose and tail bearing designs are shown in Figure 2. Heat is generated in three areas of the bearing: between the rollers and the race that is secured to the shaft; between the rollers and the race that is secured to the housing; and between the roller ends and a "rib" that is secured to the housing. Most of the heat is generated on the races, but the highest temperature occurs in the rib-roller interface, a highly loaded small area with limited heat transfer opportunities.
Bearing system performance was validated on an instrumented rig. As Figure 2 shows, the design includes channels for circulating chilled water through the bearing housing. The bearings are modified 95- and 75-mm Timken "HydraRib" tapered roller bearings. In the initial design, the bearings were lubricated by an air-oil system. However, tests indicated that additional heat removal was needed above 6,000 rpm. To permit higher speeds, a jet lubrication system replaced the air-oil system, with slots machined in the outer race to eject hot oil that might other-wise become overheated.
At the inner race, little can be done to improve cooling. However, measurements showed that the maximum shaft temperate under the inner race never exceeded 90°F—a low temperature resulting from the high efficiency of the motor applied to this spindle. The cool shaft provides a heat sink at the bearing inner diameter to complement the cooling jacket at the outer diameter. The bearings' various cooling effects enable the spindle to run continuously under full load to 12,000 rpm.
When a heat balance was computed for the motor, designers could not explain the results. Test data revealed the presence of unexpected heating. A specialist in electromagnetic analysis was added to the team, and the presence of a strong proximity effect was discovered.
The proximity effect is a power-dissipating phenomenon occurring when AC current creates a magnetic field around a conductor strong enough to induce eddy currents in neigh-boring conductors. If the AC frequency is 60 hertz or less, and the current is less than about 50 amps, the induced eddy currents are negligible. But when the frequency ap-proached 1 kilohertz and the current approached 100 amps, the induced currents became capable of producing enough heat to overwhelm the cooling system.
As a consequence of the circulating current Ic and the wire's resistance R, heat generated is proportional to Ic2R. However, current Ic circulates within the conductors and cannot be measured at the motor leads. The effect therefore appears to be an increase in the winding resistance. This increased resistance Rac can be referenced to the copper resistance Rdc. The factor of increase is Rac/Rdc.
To achieve higher spindle speed and cutting load, the motor drive current and frequency have to be increased. The effect of this increase on heat generation can be seen in Figure 3.
The motor was modeled using finite element analysis. This analysis correctly identified eddy currents from the proximity effect as the source of the losses not identified in the initial design. The analysis also showed that the motor could be wound in a way that would minimize these losses.
By this point, two motors had been built, "75 hp-1" and "75 hp-2." Neither met the torque target. The winding wire was stripped from motor 75 hp-2 and the laminations were re-used to build motor "75 hp-3." This third motor was wound in the manner suggested by the analysis.
The new winding is illustrated in Figure 4. The key step in reducing the proximity effect losses was to abandon the random winding technique used for decades in the construction of standard industrial electric motors. Non-random winding is used in its place. Strands belonging to the same turn occupy layers that offer the minimum feasible radial dimension. Minimizing this dimension minimizes the proximity effect.
Figure 3 shows the effect: a much lower rate of increase in Rac/Rdc for the newer motor. The improvement this graph shows is the basis for seeking a patent on the new winding technique. Thanks to the efforts of the spindles project team, a major unrecognized cause of motor heating has been identified and cir-cumvented—and relegated to the status of just one more motor design consideration.
About the author: Jack McCabe is vice president of technology for the National Center for Manufacturing Sciences.blog comments powered by Disqus