1

AC Propulsion – Ramco - MIT - ICA

Improving the Efficiency of High Speed Induction Motors

Using Die Cast Copper Rotors

Version 1.5 – 11 July 2013

Paul Carosa & Wally Rippel – AC Propulsion Dan Seger & Jay Sanner - Ramco James Kirtley, Jr. - MIT Malcolm Burwell - ICA

2

Executive Summary Traction electric motors are becoming more common in automotive applications, due to the growing popularity of hybrid and electrical vehicles. Generally two different types of motors are used in electric vehicle (EV) applications, DC brushless and induction motors. DC brushless motors are more common in EV applications, but high prices and price volatility in the rare earth market has been making it difficult for motor manufacturers to control and manage costs. These price issues are making induction motors more attractive for EV applications. Induction motors using copper rotors are more efficient than those employing aluminum rotors, and over the past decade the International Copper Association has focused on commercializing technology to reduce the costs associated with die casting copper rotors. However, it has recently been recognized that, when the rotor is operating at high speeds, the efficiency of inverter-driven motors incorporating die cast copper rotors is often slightly lower than those using fabricated copper rotors. The goals of this project therefore were twofold:

a) Identify procedures that can be used to produce die cast copper rotors that develop efficiencies at high rotational speeds similar to fabricated copper rotors.

b) Identify the mechanism that is responsible for the lower efficiency of the die cast copper rotors.

A series of copper rotors were die cast using a range of process parameters. Casting parameters examined included coating of the lamination stack, quenching immediately after casting, performing a heat treatment after cooling the cast rotors to room temperature, and changing the skew of the rotors. Dynamometer testing was performed to determine the impact of rotor condition on motor efficiency. The project met its first goal: die casting parameters were identified that produced copper rotors that improve cast-rotor motor efficiencies in-line with motors employing fabricated rotors. The two casting parameters that appeared to have the biggest impact on improving efficiency were quenching of the rotor immediately upon ejection from the casting die, along with the use of an insulating coating applied to the lamination stack to better isolate the conductor bars from the laminations (physically, thermally and electrically). The project was unable to completely meet its second goal: rather than identifying a single causal mechanism, the following three mechanisms were identified that could be the reason for the lower efficiencies of the die cast rotors:

3

(i) Increased electrical shorting between the rotor bars and the laminations of die cast rotors could increase circumferential parasitic currents from either bar-to-bar or inter-laminar currents.

(ii) The high temperatures associated with copper die casting could cause thermal breakdown of the lamination coating, allowing lamination-to-lamination and lamination-to-end-ring shorting.

(iii) The end rings of fabricated rotors have higher electrical resistance than die cast rotors due to the brazing material used in their fabrication. It is possible that this high resistance is reducing frequency-dependent losses that are apparent in the pure copper cage.

It is suggested that a next step in narrowing to one causal mechanism would be the sponsoring of a Ph.D. student at the University of Wisconsin to examine these effects. In addition, a series of bench-scale tests are described. The goals of this testing were to identify improved coatings for the laminations and stack, and to characterize changes in the electrical properties of the lamination steel that occur during heating of the laminations to the high temperatures associated with copper die casting. The coating test work is anticipated to continue beyond the close of this project, and will be reported separately in the future.

4

TableofContentsExecutive Summary ................................................................................................................................ 2

Index ....................................................................................................................................................... 4

1. Introduction ..................................................................................................................................... 5

2. Background ...................................................................................................................................... 7

3. Goals of Project ................................................................................................................................ 9

4. Project Participants ........................................................................................................................ 10

5. Procedures ..................................................................................................................................... 11

6. Results ............................................................................................................................................ 17

6.1 Quality of Die Castings ............................................................................................................. 17

6.1.1 6‐inch Copper Rotors ...................................................................................................... 17

6.1.2 6‐Inch and 3‐inch Aluminum Rotors .............................................................................. 17

6.1.3 3‐inch Copper Rotors ...................................................................................................... 17

6.2 Electrical Tests to Evaluate Quality of the Cast Conductor Bars .............................................. 18

6.3 Chemical Analysis of Die Cast Rotors ....................................................................................... 18

6.4 Results from Dynamometer Testing ........................................................................................ 20

6.5 End Ring to Lamination Resistance Measurements................................................................. 23

6.6 Further Evaluation of Coatings for the Lamination Stack ........................................................ 24

6.7 Evaluation of Alternate Lamination Coatings .......................................................................... 25

6.8 Changes in Electrical performance of Laminations after Casting ............................................ 27

7. Discussion ...................................................................................................................................... 30

8. Summary & Conclusions ................................................................................................................ 32

9. Future Work ................................................................................................................................... 34

10. References/Bibliography ................................................................................................................ 35

Appendix 1: Brief Description of the Die Casting and Fabrication Methods for Producing Rotors ..... 36

Appendix 2: Detailed Description of Motor, Rotor and Laminations ................................................... 38

Appendix 3: Sections Cut through 6‐Inch Long Copper Die Cast Rotors .............................................. 39

Appendix 4: Sections Cut through 6‐Inch Long Aluminum Die Cast Rotors ......................................... 41

Appendix 5: Sections Cut through 3‐Inch Long Copper Die Cast Rotors .............................................. 43

Appendix 6: Efficiency Data .................................................................................................................. 46

5

1. Introduction Due to the growing popularity of hybrid and electrical vehicles (xEVs), traction electric motors are becoming common in automotive applications. There are a number of different terminologies associated with these types of vehicles, including hybrid electric vehicles (HEV, such as the Toyota Prius), battery electric vehicle (BEV, such as the Tesla Model S), plug in hybrid electric vehicles (PHEV, such as the recently released Toyota Prius Plug-In Hybrid) and extended range electric vehicles (EREV, such as the Chevy Volt), but all utilize an electric motor to provide some (or all) of the traction supplied to the vehicle’s drive wheels. Historically two different types of motors have been used in xEV applications, induction and DC brushless motors. For an induction motor, rotation is generated via magnetic fields from the stator inducing current in the rotor. The most common form of induction motors utilize a squirrel-cage rotor (such as the one shown in Figure 1), which comprises a series of longitudinal conductor bars (usually made of aluminum or copper) inserted into slots located towards the periphery of a stack of steel laminations. The conductor bars are connected at both ends by shorting “end” rings, producing what is known as a squirrel cage structure. In contrast, DC brushless motors typically employ permanent magnets in the rotors. Both motors use stationary coils in the motor housings to create the rotating magnetic fields that cause the rotor to spin.

Figure 1: Copper die cast rotor The high energy permanent magnets used in DC brushless motors are made from neodymium and other rare earth materials, and while the prices of these materials were relatively low and stable until 2010, since then they have been subject to enormous price volatility (Figure 2), making it more difficult for xEV manufacturers to control and manage costs. In addition, induction motors typically have a significantly lower acquisition cost for xEV applications, compared with DC brushless motors. The price

6

volatility in the rare earth markets has made induction motors more attractive for xEV manufacturers.

Figure 2: Historical pricing information for rate earth materials used in high energy permanent magnets

7

2. Background As induction motors are typically less efficient than permanent magnet motors, efforts are continuously underway to improve their efficiency. One approach has been to develop more economic processes for making the squirrel cage from a highly conductive material such as copper. The squirrel cages for induction motors are typically produced from either aluminum or copper, and rotors can be manufactured by either casting or a fabrication approach, where the squirrel cage is brazed together from a large number of machined pieces (the details of the die cast and fabrication methods are described in more detail in Appendix 1). The lowest cost approach for producing rotors is by die casting aluminum in the squirrel cage, but as the electrical conductivity of pure copper is more than 60% greater than that of aluminum, rotors constructed from copper generally produce motors of higher overall efficiency. However, the high melting point of copper (1083oC/1982oF for copper versus 660oC/1220oF for aluminum) makes copper rotors more difficult and more costly to die cast. Historically, copper rotors (especially large rotors) have generally been produced using the fabrication approach, despite the high costs associated with assembly and brazing. Over the past ten years, the International Copper Association has focused on commercializing technology to reduce costs associated with die casting copper rotors. During that period, however, it has been recognized that, when the rotor is operating at the high rotational speeds associated with xEVs (up to 15,000 rpm), the efficiency of inverter-driven motors incorporating die cast copper rotors is slightly lower than motors using fabricated copper rotors (Figure 3). This phenomenon is not apparent when the rotor is rotating at lower speeds (1,800-3,000 rpm), or with motors that are not inverter driven. This difference is surprising, as it might be expected that the efficiency associated with fabricated copper rotors would be lower, as the cages of fabricated rotors generally have higher electrical resistance due to the brazing material used in their manufacture. This is emphasized by the data in Table 1, which shows the theoretical resistance for 75 mm long aluminum and copper squirrel cages. When comparing measured and calculated values, the aluminum and copper die cast rotors are close to theoretical, while the measured values for the fabricated rotors are significantly higher, again probably due to the brazing materials in their cages. Accordingly the main purpose of this project was to identify the cause of the lower efficiencies of the copper die cast rotors, and identify processing conditions that produce copper die cast rotors having equal or even higher efficiencies than their fabricated counterparts.

8

Figure 3: Efficiencies of different motors at high rotational speeds

Item

Resistance (μΩ)

Calculated Measured

75 mm long fabricated copper rotor 1.86 3.00

75 mm long cast copper rotor 2.16 2.00

75 mm long cast aluminum rotor 3.52 3.33

Table 1: Effective internal cage resistance analysis for different squirrel cage materials†

† Note that the resistance calculated for the fabricated rotor was lower than that of the cast rotor as the fabricated rotor was assumed to have a 1.0‐inch long end ring, while the end ring of the cast rotor was 0.6‐inches long (this is discussed in more detail in Section 5 of this report).

9

3. Goals of Project The goals of this project were twofold:

1) Identify procedures to produce die cast copper rotors that develop efficiencies at

high rotational speeds similar to fabricated copper rotors. 2) Identify the mechanism that is responsible for the lower efficiency of the die cast

copper rotors.

10

4. Project Participants Four organizations/individuals participated in this project. Their roles are summarized below.

AC Propulsion (ACP) is located in San Dimas, CA and is a leader in the development, design and manufacture of drivetrain systems for electric vehicles. ACP designed the motor used in this project and supplied the laminations used to produce the cast rotors. That motor design is currently made with fabricated copper components, so ACP was able to supply a fabricated baseline version of the motor with which to compare cast versions. In addition, ACP did much of the performance testing, particularly efficiency measurements via dynamometer testing of the copper and aluminum rotors.

Located in Greenville, OH, Ramco Electric Motors, Inc. (Ramco) is a

manufacturer of electric motors, stators & rotors for use in industrial, commercial, military & aerospace applications. Ramco used a proprietary die casting process for the production of the copper and aluminum rotors examined in this study.

Dr. Jim Kirtley, Jr. is a professor of electrical engineering at the Massachusetts

Institute of Technology in Cambridge, Massachusetts. Dr. Kirtley provided motor design, consulting and magnetic testing services for this project.

The program was led and managed by Malcolm Burwell of the International

Copper Association (ICA), New York, NY. The ICA also provided funding for tooling and materials used in this work.

11

5. Procedures Initially it had been planned to employ the 6-inch long rotors used in ACP’s AC-150 motor for the testing performed in this project, but due to casting issues described later in this report, the focus was changed to the shorter 3-inch long rotors used in ACP’s AC-75 motor. The AC-75 is an air cooled three-phase induction motor designed specifically for automotive applications (Figure 4). Specifications for the motor are listed in Table 2, and details of the motor construction are provided in Appendix 2.

Figure 4: AC-75 motor used in this study

Item Magnitude

Peak shaft torque >115 Nm

Shaft power – peak >75 kW

Shaft power – continuous >25 kW

Rotor length 3.0 inches

Rotor diameter 4.95 inches

Number of conductor bars in rotor 68

Maximum speed 13,000 rpm

Dry weight 34 kg

Table 2: Specifications for the AC-75 motor This motor typically uses a fabricated copper rotor. For the die casting trials described here, the design of the rotor was essentially unchanged, except that the length of the

12

end rings were reduced from 1.0-inches used with the fabricated rotor to 0.6-inches for the die cast rotor. However, as the electrical resistance of a cast copper end ring will be lower than that of a fabricated end ring (due to the high number of brazed joints in a multi-part brazed end ring of the AC-75), this was not expected to negatively impact efficiency. The copper rotors were all die cast at the Ramco plant in Greenville, Ohio using a proprietary die casting processes marketed as Ramcast†. During the course of this project a total of 37 rotors were cast. The first 18 rotors were 6-inches long, and were used to attempt to identify casting parameters that would fully fill the end rings and slots. The long, thin conductor bars made it difficult to attain repeatable bar-fill quality. Process parameters used to produce these initial 18 rotors are summarized in Table 3. Due to the poor quality of these castings, only limited testing was performed on these 6-inch long rotors.

Rotor No. Alloy Stack Preheat

(oF) Coating

1 Cu Room temp No

2 Cu Room Temp No

3 Cu Room Temp No

4 Cu Room Temp Yes

5 Cu 775 Yes

6 Cu 735 Yes

7 Cu 785 Yes

8 Cu 780 Yes

9 Cu 770 Yes

10 Cu 788 Yes

11 Cu 782 Yes

12 Al 785 Yes

13 Al 782 Yes

14 Al 788 Yes

15 Al 810 Yes

16 Al 812 Yes

17 Al 784 Yes

18 Cu 1145 Yes

Table 3: Processing conditions for first 18 rotors

† Note that the Ramcast process does not use the vertical THT machines described on the Ramco web site.

13

After these initial 18 castings were produced, the rotor length was shortened to 3-inches, and casting quality improved significantly. This allowed the study to systematically vary a number of process parameters, to determine the impact upon both casting quality and electrical efficiency of each of these parameters. The parameters used to produce these 19 rotors are summarized in Table 4, and each of the variables examined is discussed in more detail below.

Rotor No.

Quench after Cast

Post Cast Heat

Treatment

Coated Slots

Skew (No. of bars)

Alloy Evaluation Method

19 N N Y 0 Cu Band saw

20 N N Y 0 Cu Elect/Weight/Band Saw

21 N N Y 0 Cu Elect/Weight/Band Saw

22 N N Y 0 Cu Elect/Weight

23 N N Y 0 Cu Elect/Weight

24 Y N Y 0 Cu Elect/Weight

25 Y N Y 0 Cu Elect/Weight

26 N Y Y 0 Cu Elect/Weight

27 N Y Y 0 Cu Elect/Weight

28 N N N 0 Cu Elect/Weight

29 N N N 0 Cu Elect/Weight

30 N N Y 3 Cu Elect/Weight

31 N N Y 3 Cu Elect/Weight

32 N N Y 5 Cu Elect/Weight

33 N N Y 5 Cu Elect/Weight

34 N N Y 0 Al Elect/Weight

35 N N Y 0 Al Elect/Weight

36 Y N Y 0 Cu Elect/Weight

37 Y N Y 0 Cu Elect/Weight

Table 4: Testing matrix. Note that rotors 19 thru 23 used the “standard” casting conditions,

and for rotors 24 thru 37 one variable was adjusted for each casting, as highlighted in the Table. Rotors 19 thru 21 were destructively tested to confirm high casting quality

Rotor length – As described above, the rotors tested for electrical efficiency in this study were all 3-inch long.

14

Materials - The majority of the rotors were cast using copper alloy C102 (high purity oxygen-free copper). The chemical specification for this alloy is listed in Table 5(1).

Element Composition (wt.%)

Nominal Minimum Maximum

Copper (incl. silver) -- 99.95 --

Residual Deoxidants -- None

Table 5: Chemical composition specification for copper alloy C102(1) For comparative purposes, several rotors were also die cast from aluminum alloy 170.1, and the chemical specification for this alloy is listed in Table 6(2).

Al min Si Fe Mn Cr Zn Ti

Unspecified Other Elements

Each Total

99.7 (a) (b) (b) (c) (c) 0.05 (c) 0.03 (c) 0.10

(a) The Al content is the difference between 100.00% and the sum of all other metallic elements present in amounts of 0.010% or more each, expressed to the second decimal before determining the sum

(b) Fe/Si ratio = 1.5 minimum (c) Mn+Cr+Ti+V = 0.025%max

Table 6: Chemical composition specification for aluminum alloy 170.1 (in wt.%)(2) Laminations – The main 0.014-inch (0.35 mm) thick laminations used in this study were supplied by JFE Steel Corp. through ACP’s factory in China. Two thicker (0.031-inch, 0.79 mm) laminations fabricated by Proto Laminations from cold rolled steel were placed at each end of the stack to prevent lamination bending during high pressure filling with molten copper. More details about the laminations are provided in Appendix 2. Coatings Used on Lamination Stack - As shown in Table 3, the first three rotors were cast without applying any additional coating to the lamination stack. However, after experiencing problems filling the rotor, all other rotors were cast after Ramco applied a proprietary, white-colored ceramic coating to the whole stack. This coating was applied by dipping the assembled lamination stack into a container of the coating material, and a photograph of the typical coating thickness on the inside of the slots is shown in Figure 5. The coating was approximately 0.05 mm (0.002-inches) thick. This coating is

15

typically used to provide a thermal barrier between copper and steel to prevent premature solidification during casting. Skew - The fabricated copper rotors normally used with both the AC-75 and the AC-150 motors do not have any conductor bar skew. However, an analysis performed at the start of this project suggested that incorporating skew could improve performance and efficiency. Therefore, rotor castings were included that implemented three different skews – the zero skew normally used with this rotor, along with skews corresponding to the width of three and five conductor bars.

Figure 5: Photograph of a slot in a non-ACP lamination stack, after the application of the white coating

Stack Preheat - The stack preheat temperatures for the first 18 rotors are listed in Table 3. For the remaining 19 rotors, the stack preheat temperature was standardized to 425±28oC (800±50oF). Quench after Ejection from Die - Once the cast rotors were ejected from the casting die, two different cooling regimes were examined, air cooling and a rapid quench. In each case, the rotors were at a temperature of about 350-370oC (650-700oF) when ejected from the die, and the air cooled samples were simply allowed to naturally cool to room temperature. For the water quenched samples, approximately 45 to 60 seconds elapsed from opening the die to plunging the castings into a tank of water. The tank contained about 170-190 liters (45-50 gallons) of water at an approximate temperature of 20oC (65-70oF). No quenching additives were present in the water. Post Cast Heat Treatment & Quench - The impact of a post casting heat treatment upon rotor efficiency was also evaluated. Select rotors that had been air cooled to room temperature following ejection from the die were subsequently placed into a furnace

16

pre-heated at 427oC (800oF) for a period of 90 to 120 minutes. After removal from the furnace, the rotors were water quenched as described above. Performance testing of the die cast rotors was carried out using an in-house dynamometer at ACP (Figure 6). Efficiency was calculated by comparing input power to the mechanical power developed, and measurement accuracy is believed to be 0.1% of full scale. All efficiencies measured were DC-to-shaft efficiencies and included the efficiency of the inverter, which was assumed to be constant for all tests. The same stator was used for all efficiency measurements, and each measurement presented in this report is the average of five separate measurements. Efficiency measurements were performed at two power levels (10 kW and 50 kW), two rotational speeds (8,000 rpm and 10,000 rpm), and at two rotor temperatures (50oC and 100oC). These speeds were chosen because it was known in the industry that they would demonstrate the difference between fabricated and cast rotors. Rotor temperature measurements were performed with an infrared pyrometer using the following procedure - rotor temperatures were monitored as they increased during operation, and efficiency measurements were performed once the rotors achieved a temperature of either 50oC or 100oC. Both end rings were painted black to ensure stable temperature readings.

Figure 6: Dynamometer used at AC Propulsion for efficiency testing

17

6. Results This section describes the results from various tests performed during the project. 6.1 Quality of Die Castings This initial section describes the quality of the die cast rotors produced in this study, including the 6-inch long copper rotors, the 6-inch and 3-inch long aluminum rotors and the 3-inch long copper rotors. Photographs of sections through a number of the die cast rotors are shown in Appendix 3 through Appendix 5. 6.1.1 6-inch Copper Rotors As noted earlier, Ramco experienced severe problems while attempting to die cast the six inch long copper rotors. As summarized in Table 3, 13 attempts were made to cast the 6-inch long copper rotors, and although some variation in casting quality was observed, none of the 6-inch long copper rotor castings were fully filled. Photographs of sections through some of these rotors are shown in Appendix 3. Figure A3-1 shows sections through rotor number 3, and large holes and non-filled areas can be seen in both the end rings and the conductor bars. Rotor quality improved somewhat with rotors 4 and 5 (Figure A3-2), but the quality of other castings (numbers 7, 9 and 18) was also poor. 6.1.2 6-inch and 3-inch Aluminum Rotors Similar problems were not experienced when die casting both 6-inch and 3-inch long aluminum rotors. X-rays and photographs of sections through some aluminum rotors are shown in Appendix 4. Sections cut through the rotors showed that the conductor bars were essentially fully filled (Figure A4-1), and little-to-no porosity was observed on x-rays through either the end rings or the slots (Figure A4-2). 6.1.3 3-inch Copper Rotors Once the rotor length was reduced from 6-inches to 3-inches, the quality of the die cast copper rotors improved significantly. Appendix 5 shows a series of photographs of sections cut through two of the 3-inch long copper rotors. The quality of the these rotors is very high, with very little porosity or non-fill observed in either the end rings or the conductor bars. The quality of these copper die cast rotors was considered adequate to perform the testing outlined in Table 4.

18

6.2 Electrical Tests to Evaluate Quality of the Cast Conductor Bars Additional testing was performed to evaluate the quality of the conductor bars for each of the die cast rotors. The experimental set-up is shown in Figure 7. A 50 amp DC current was applied between opposite sides of the two end rings, and a pair of probes used to measure the potential difference along each conductor bar. If a conductor bar had poor cast quality, or had a total break (due to non-filling during casting), this would be revealed by a high potential difference.

Figure 7: Experimental set-up of test similar to that used by ACP to evaluate the quality of the

conductor bars in the die cast rotors This is a rapid, non-destructive test procedure used on manufacturing lines to confirm the quality of rotor cage integrity. In all the 3-inch rotors tested in this study, the results of these electrical measurements were consistent, with acceptable quality of rotor integrity. 6.3 Chemical Analysis of Die Cast Rotors The chemical composition of the copper used to produce the rotor was analyzed, to determine if impurities dissolved in the copper could be the cause of the observed efficiency reduction. Both the copper used to produce the rotors, as well as the copper from one of the cast rotors were analyzed. All elements (except oxygen) were analyzed using inductively coupled plasma mass spectrometry, while oxygen concentrations were determined using a Leco oxygen analyzer. The measured chemistries are listed in Table 7. The only element showing any significant increase after casting is oxygen. However, only those elements that dissolve

19

into a solid solution of copper will have a significant impact on electrical conductivity, and elements known to have a major impact on reducing electrical conductivity are shown in Figure 8. If an element, such as oxygen, does not dissolve to any significant extent in a copper solid solution, but instead forms second phase particles with copper, such an element has only a minor impact on electrical conductivity.

Element Composition at Different Locations (wt.%)

Raw Material Pre-Casting

Gate End RingSurface of

Conductor Bar Opposite Gate

End Ring

Iron <0.001 <0.001 0.003 <0.001

Oxygen <0.001 0.12 -- 0.12

Phosphorous <0.005 <0.005 <0.005 <0.005

Manganese <0.001 <0.001 <0.001 <0.001

Silicon 0.008 <0.005 0.015 <0.005

Nickel 0.001 0.001 0.001 0.001

Chromium <0.001 <0.001 <0.001 <0.001

Table 7: Chemical composition measurements – samples taken from an early 6-inch long rotor

casting

Figure 8: Impact of different elements in solid solution on the electrical conductivity of

copper(3)

20

Figure 9 provides information regarding the solubility of oxygen into a copper solid solution(4,5). Data published by Horrigan(5) summarizes measurements by various researchers (Figure 9b). Those data show that the maximum solubility of oxygen in solid copper at 1000oC is most likely less than 0.02%, and at 600oC is most likely less than 0.01%. Therefore, the remainder of the oxygen shown in Table 7 must be present as copper oxide particles. It is unlikely that such low solubilities of oxygen as shown in Figure 9b would have a significant impact on electrical conductivity. This conclusion is backed-up by the electrical performance of copper alloy C110 – this alloy allows a maximum oxygen content of 0.04%, but it still has excellent electrical conductivity(1).

Figure 9: Cu-O phase diagram

a) Copper end of the phase diagram(4) b) Solubility of oxygen in solid copper(5)

To further analyze the impact of contamination of the copper by oxygen (or any other element), electrical conductivity measurements were performed on the runners of each of the 3-inch long rotors. The results are listed in Table 8, showing that conductivity values for the copper rotors range between 99.0 and 100.7% IACS. This confirms the analysis discussed above, that the copper is not being contaminated by any impurities that reduce conductivity. 6.4 Results from Dynamometer Testing Die cast rotors 22 through 37 were all dynamometer tested at ACP using the equipment shown in Figure 6. The casting parameters used to produce these rotors are summarized in Table 4. To provide baseline data, two fabricated copper rotors were

21

also tested, one before and one after the cast rotors (to confirm absence of drift in the dynamometer test equipment).

Rotor Number Electrical Conductivity

(% IACS)

19 100.0

20 99.8

21 100.3

22 100.1

23 100.0

24 99.0

25 99.8

26 100.2

27 100.4

28 100.6

29 100.4

30 100.3

31 100.3

32 100.0

33 100.4

34 61.2*

35 61.2*

36 100.7

37 100.1

*Aluminum rotors

Table 8: Electrical conductivity measurements performed on the runners of the 3-inch long cast

rotors The efficiency data at two rotational speeds (8,000 rpm and 10,000 rpm), two temperatures (50oC and 100oC) and two power levels (10 kW and 50 kW) are listed in Table A6-1 in Appendix 6. Table A6-2 in the same appendix shows the difference in efficiency for each condition compared to the baseline measurement for the fabricated copper rotor shown on the first row of the Table. The average loss in efficiency for each condition examined in this study is then summarized in Table A6-3, and these data are reproduced in Figure 10 below. This shows that the four rotors that were water quenched immediately after ejection from the casting die have the lowest loss in efficiency, having an efficiency only 0.2% lower on average than the fabricated copper

22

rotor. The “standard” casting conditions (including air cooling), and those rotors given a post casting heat treatment have slightly lower average efficiencies, 0.50% and 0.57% lower than the fabricated copper, respectively. The rotors cast without the Ramco proprietary ceramic coating had even lower efficiency, while the rotors produced with skew both exhibited poor efficiencies (this latter result is to be expected, given the rotor bar shorting that was present, which is known to counteract any benefits from skewed rotor bars).

Figure 10: Average efficiency loss for each condition examined The data shown in Appendix 6 and Figure 10 indicate that quenching immediately after casting had the biggest impact upon improving efficiency, while casting without the proprietary ceramic coating on the lamination stack degraded efficiency. These two observations suggest that minimizing physical and electrical contact between the die cast copper conductor bars and the lamination stack had the biggest impact on efficiency improvements. It seems likely that quenching immediately following casting “breaks” the cast conductor bars away from the lamination stack, due to the difference in coefficient of thermal expansion between the copper and the steel. The presence of the ceramic coating is believed to lower the copper-to-steel adhesion and promote breakaway. Performing an elevated temperature heat treatment after cooling the rotor to room temperature did not have a similar effect, possibly due to the relatively low treatment temperature examined in this study.

0.00

1.00

2.00

3.00

4.00

Quench Standard PCHT No coat Skew 5 Skew 3

Loss in

Efficiency

23

Figure 11 summarizes these conclusions: it is a modified version of a Figure shown earlier in this report, but now includes the data for the efficiency of copper die cast rotors incorporating water quenching immediately after casting. This shows that the efficiency of motors utilizing die cast copper rotors is now essentially the same as those using fabricated rotors.

Figure 11: Efficiencies of different motors at high rotational speeds, incorporating the data for

the rotors water quenched immediately after die casting 6.5 End Ring to Lamination Resistance Measurements To further evaluate the impact of casting parameters on the amount of physical and electrical contact between the cast copper and the lamination stack, additional measurements were made of the electrical resistance between the end ring and the lamination stack for rotors cast under different conditions. A 100 A current was applied between the shaft rear end and the front end ring. The electrical resistance was measured between the front end ring and the lamination surface using a digital volt meter. The results, which are summarized in Table 9, show very little difference for any of the copper rotors cast in this study, while the resistance measurements were about three orders of magnitude higher for the fabricated rotor. This appears to contradict the analysis presented in the previous section, which suggested that the use of the ceramic coating together with quenching immediately after casting had improved motor efficiency by minimizing the electrical contact between the

24

die cast copper conductor bars and the lamination stack. Additional work appears to be required to better characterize the state of the interface between the conductor bars and the laminations.

Rotor Number Rotor Material Condition Resistance

(μΩ)

22 Cu Normal 1.2

23 Cu Normal 1.2 – 4.6

24 Cu Quench 0.15 – 3.0

25 Cu Quench 0.65 – 6.8

29 Cu No coat 1.0

31 Cu Skew 3 1.3

34 Al -- 13

35 Al -- 5 - 7

Fabricated Cu -- 1,450

Table 9: Electrical resistance measurements between the front end ring and the lamination

surface for rotors produced under different conditions 6.6 Further Evaluation of Coatings for the Lamination Stack The proprietary ceramic coating applied to the outside of the lamination stack prior to casting has three main functions – (1) to minimize heat loss from the molten copper to the cooler lamination stack (thereby maximizing castability), (2) to provide some level of electrical insulation between the copper conductor bars and the lamination steel, and (3) to provide a weakness zone from which the bar can pull away from the steel during the forces generated from differential contraction quench. The data in Appendix 6 appears to show that this coating was partially successful in improving the efficiency of the die cast copper rotors, by providing a barrier between the liquid copper and the steel laminations. To examine how well this coating was standing up to the injection of molten copper, several rotors were sectioned to directly examine the interface between the cast copper and the lamination stack. One such sample is shown in Figure 12, which indicates that some of the white coating material did survive the copper die casting process, and was still present on the surfaces of both the copper

25

conductor bar and the slot. A brown material, possibly iron oxide, was also present on the surfaces of the conductor bar and the laminations.

Figure 12: Sectioned samples from a copper rotor, showing the surface of the cast copper conductor bar (right) and inside surface of the slot in the lamination stack (left)

Figure 13 shows a section of a cast copper conductor bar, and the ceramic coating material is still present over at least a portion of the cast surface. The sample also clearly shows witness marks from individual laminations within the rotor stack, confirming that the copper does indeed make intimate contact with each individual lamination edge during the die casting operation.

Figure 13: Piece of copper conductor bar sectioned from a die cast rotor 6.7 Evaluation of Alternate Lamination Coatings Ramco has also performed some first-pass bench-scale testing to identify potential replacements for the C5 lamination coating. Two coatings were examined, Remisol EB 5308 and EB 5620, both manufactured by Rembrandtin (www.rembrandtin.com). The

26

EB 5308 material is described as an inorganic/organic hybrid coating with inorganic insulating pigments that is resistant to continuous exposure to temperatures of above 270°C. Phosphoric acid in the liquid coating forms an iron phosphate layer on the steel lamination, which should survive the molten copper temperatures. Samples of both coatings have been obtained from Rembrandtin, and simple testing has been performed where small rectangular coated steel samples were dipped into baths of either molten aluminum (at 705oC/1300oF) or molten copper (at 1300oC/2375oF). Figure 14 shows the results from the testing, indicating that while both the aluminum and the copper adhered somewhat to the samples, the copper buildup was more extensive and thicker (about 0.25 mm/0.01-inches per side).

a) b)

Figure 14: Surface condition after immersion of samples EB 5308 and EB 5620 into molten metals a) Molten aluminum b) Molten copper

To further quantify this testing, electrical resistance measurements were performed to determine the degree to which the coatings were still intact after immersion. Probes were used to perform the measurements at the following three locations (see Figure 15):

Probe location 1 – Coating had been removed by mechanical abrasion Probe location 2 – As received coating Probe location 3 – Portion of sample immersed into molten metal

The electrical resistance results are summarized in Table 10, which shows that the electrical resistance was lower for the samples dipped into molten copper. However, these resistance measurements are still high (22,000 Ω), and would most likely be adequate to insulate the lamination stack from the copper conductor bars. These results suggest, therefore, that both these coatings could be suitable for copper die cast rotors, and it would be worthwhile to perform castings trials with these coatings on the

27

slot surfaces (with and without the proprietary Ramco ceramic coating) further reducing bar-to-lamination resistance to determine the coatings’ impact on overall efficiency.

Figure 15: Locations uses to test electrical resistivity

Coating Liquid Metal Probe Locations Resistance

(ohms)

5308

Aluminum

1 + 2 ∞ 1 + 3 ∞

5620 1 + 2 ∞

1 + 3 ∞

5308

Copper

1 + 2 ∞

1 + 3 22,000

5620 1 + 2 ∞

1 + 3 22,000

Table 10: Results of electrical resistivity measurements on coating samples EB 5308 and EB

5630 after dipping into molten aluminum or molten copper

6.8 Changes in Electrical Performance of Laminations after Casting A separate series of bench-scale tests have been performed to determine whether heating of the lamination steel during the copper die casting operation negatively impacts its electrical performance. For the production of both fabricated copper rotors and aluminum die cast rotors, the laminations stacks are only heated to a relatively low temperature (a maximum of 800oC and 700oC, respectively), while the laminations will be heated to a much higher temperature during copper die casting. This suggest the possibility that the magnetic properties of the lamination steel could be affected differently in copper die cast rotors as compared to both fabricated copper rotors and cast aluminum rotors. To investigate this possible effect, two types of laminations were tested - as-received laminations, and laminators extracted from copper rotors after die casting. The electrical

28

properties of the steel were tested at two locations for each lamination – adjacent to the slot OD and adjacent to the slot ID. Copper wire was wound around each of the laminations (as shown in Figure 16), and the circuit used in Figure 17 used to test performance at each location.

a) b)

Figure 16: Windings used to test electrical performance of the lamination steel

Figure 17: Circuit used to test the electrical performance of the lamination steel Testing showed that the lamination steel located adjacent to both the lD of the slot and the OD of the slot showed only minor changes in electrical performance after die casting (Figure 18). It is unlikely that a change in the steel’s magnetic performance during die casting could explain the lower efficiency of the die cast rotors.

29

a) b)

Figure 18: Comparing electrical performance of the steel laminations before and after casting

with copper a) Testing performed at the ID of the slot using the configuration shown in Figure

16a (at 1 kHz) b) Testing performed at the OD of the slot using the configuration shown in Figure

16b (at 8 kHz)

As received

After casting

30

7. Discussion

The data presented in Figure 10 and Figure 11 of this report indicate that casting conditions have been identified that improve the efficiency of motors using die cast rotors, bringing their efficiencies in-line with motors using fabricated rotors. The two parameters that appeared to have the biggest impact on improving efficiency were quenching of the rotor immediately upon ejection from the casting die, along with the use of a ceramic coating applied to the lamination stack. Both of these parameters appeared to isolate the conductor bars from the lamination stack, physically thermally and electrically. During this study, five possible explanations were generated as possible mechanisms responsible for the lower efficiencies of the die cast rotors, and it is worthwhile discussing each of these mechanisms in relation to the results of this study. In the following section, each of the five mechanisms is initially presented followed by a brief discussion of the results from this study in relation to the proposed mechanism.

(i) During solidification, the liquid copper that has filled the rotor die (and the lamination slots) is subjected to a very high pressure of several thousand pounds-per-square inch. The liquid copper in the slots will be pushed against the laminations, and so will make much better contact than the copper rods inserted into the slots of fabricated rotors. Electrical shorting between the rotor bars and the laminations of die cast rotors could increase either circumferential parasitic currents from bar-to-bar or inter-laminar currents. The losses associated with this mechanism are likely to be higher when the rotor is spinning at high speeds (10,000 rpm versus 1,800 rpm), as centrifugal forces will push the copper conductors bars against the lamination stack.

The impact upon efficiency of quenching after casting plus the use of the ceramic coating (shown in Figure 10, Figure 11 and Appendix 6) suggest that electrical shorting between the rotor bars and laminations could indeed be the cause of the reduced efficiency. It should be noted that the mechanism of efficiency reduction associated with bar-to-lamination shorting is not fully understood. However, as described in Section 9 below, further work is underway to identify coatings that will improve the level of insulation between the conductor bars and the laminations.

(ii) The web in the laminations at the OD of the slots is very thin and so will be

heated to a very high temperature during casting and solidification of the liquid

31

copper. It was postulated that this high temperature could degrade the electrical and magnetic performance of the steel, leading to higher losses. However, data shown in Figure 18 suggest that heating of the lamination steel during the copper die casting process has not significantly impacted the electrical properties of the steel. This explanation, therefore, has been discarded.

(iii) The laminations have a coating on their surfaces. The high temperatures associated with copper die casting could cause thermal breakdown of the coating, allowing lamination-to-lamination and lamination-to-end-ring shorting.

The data from Table 10 indicates that some breakdown of the C5 coating has occurred during the copper die casting process, but it is likely that lamination-to-lamination insulation is still adequate. The bar-to-lamination resistance is still low, which could be the origin of the lower efficiencies. This effect should be further investigated by coating the interior of the bar slots with inorganic coatings that lower the resistance, such as those from Rembrandtin.

(iv) The end rings of fabricated rotors have higher electrical resistance than die cast rotors due to the brazing material used in their fabrication. It is possible that this high resistance is reducing frequency-dependent losses that are apparent in the pure copper cage.

No attempt was made in this project to examine the impact of the electrical conductivity of the brazed rotors on motor efficiency. This mechanism should be explored in future work.

(v) Minor contamination of the liquid copper by impurity elements would significantly reduce electrical conductivity, thereby reducing overall efficiency of the motors. Testing of both the chemical composition (Table 7) and electrical resistivity (Table 8) indicate that contamination of the liquid copper did not occur, and so this is not the reason for the lower efficiency associated with die cast copper rotors. This mechanism was also discarded.

32

8. Summary & Conclusions

1. The goals of this project were twofold: a. Identify procedures that can be used to produce die cast copper rotors

that develop efficiencies at high rotational speeds similar to fabricated copper rotors.

b. Identify the mechanism that is responsible for the lower efficiency of the die cast copper rotors.

2. A series of copper rotors were die cast using a range of process parameters.

Dynamometer testing was performed to determine the impact of rotor condition on motor efficiency. Testing confirmed the difference in electrical efficiency at high speeds between fabricated and die cast rotors.

3. The project met its first goal: die casting conditions have been identified that

produce copper rotors that bring motor efficiencies in-line with motors employing fabricated rotors. The two casting parameters that appeared to have the biggest impact on improving efficiency were quenching of the rotor immediately upon ejection from the casting die, along with the use of a coating applied to the lamination stack to better isolate the conductor bars from the laminations, both physically and electrically.

4. The project was unable to completely meet its second goal: mechanisms

responsible for the improved efficiencies of the die cast rotors are discussed, and the following three possibilities remain:

a. Electrical shorting between the rotor bars and the laminations of the die cast rotors originating from the high pressures used with the die casting process.

b. Thermal breakdown of the lamination coating due to the high temperatures associated with copper die casting, allowing lamination-to-lamination and lamination-to-end-ring shorting.

c. High resistance of the fabricated end rings reducing frequency-dependent losses that are apparent in the pure copper cages.

5. A series of bench-scale tests are also described, having the goals of identifying

improved coatings for the laminations and stack, and characterizing changes in the electrical properties of the lamination steel due to the high temperatures associated with copper die casting. This work should be continued beyond the current project to determine if standard C5 coatings applied to the slot interiors

33

can completely isolate the bars from the laminations, and, if so, if the efficiency is further increased.

34

9. Future Work Suggested future work on this project involves the following:

1. A focus of future work should be to attempt to produce a cast copper rotor where the cage is fully insulated from the lamination core. This could be used as a baseline for demonstrating the ideal case where parasitic losses between the conductor bars and the laminations are completely eliminated. Certainly data shown in Figure 10 and Figure 11 has shown that coatings applied to the slots of the lamination stack can improve electrical isolation between the conductor bars and the laminations, leading to efficiency improvements. In the current project, a nano-silicon coating and two Rembrandtin coatings have been identified that may provide improved isolation. Coating of a few laminations stacks and production/testing of the copper rotors for efficiency should be pursued.

2. Section 7 of this report discuses which of five possible five mechanisms might be

responsible for the improved efficiency observed in this project. A more detailed study is necessary to evaluate the leading three of these potential mechanisms in more detail. It is recommended that a Ph.D. project be initiated, to allow a student sufficient time to examine each mechanism in detail.

35

10. References/Bibliography

1. Standards Handbook: Wrought Copper and Copper Alloy Mill Products, Part 2 –

Alloy Data, Pub: Copper development Association, 1973 2. ASM Handbook, Aluminum and Aluminum Alloys, Ed: J.R. Davis, 1993, p 25 3. ASM Handbook, Copper and Copper Alloys, Ed: J.R. Davis, 2008, p 4 4. ASM Handbook, Phase Diagrams, p 2-174 5. Virginia Horrigan, Met. Trans. 8A, 1977, p 785

36

Appendix 1: Brief Description of the Die Casting and Fabrication Methods for Producing Rotors

A rotor for an induction motor consists of two portions, a lamination stack and a squirrel cage. The stack consists of a series of thin steel laminations, generally produced by stamping from sheet, each containing a series of slots around their circumference (Figure A1-1). To produce a fabricated rotor, copper conductor bars are inserted into each of the slots and the two end rings are constructed from a series of machined bars. This whole copper assembly is then brazed into place.

Figure A1-1: Lamination stack for a rotor With the die casting method, the lamination stack is loaded into a reusable steel die, and liquid metal is injected into the die to produce the end rings and conductors bars. Figure A1-2 shows the type of die casting machines used to produce the castings. Figure A1-3 shows a model of the casting shape produced when die casting a rotor (the casting die and lamination stack are not shown). The liquid metal is fed from the biscuit along the runner and through the gate into the first end ring. After filling the first end ring, the liquid metal flows through the multiple slots and into the second end ring. Once the liquid metal has solidified, the casting is ejected from the die and the runner system removed.

37

Figure A1-2: Die casting machines

b) Horizontal injection c) Vertical injection

Figure A1-3: Model of the shape cast when die casting a rotor (note that the die and lamination stack are not shown)

Appendix 2: Detailed Description of Motor, Rotor and Laminations Geometric and process information regarding the construction of the stator and rotor of the AC-75 motor are provided in Table A2-1.

38

Item Magnitude

Motor Poles 4

Phase 3

Stator

OD 8-inches

Bore 5-inches

Gap between stator and rotor 0.015-inches

Number of slots 48

Rotor

Overall Length 3-inches

End ring length – fabricated rotors 1.0-inches

End ring length – die cast rotors 0.6-inches

Lamination

Number of slots 68

Thickness 0.014-inches

Lamination supplier JFE Steel Corp.

Condition Fully processed

Annealing after stamping None

Coating C5

Table A2-1: Geometric and processing information for the LCM-150 motor

39

Appendix 3: Sections Cut through 6-Inch Long Copper Die Cast Rotors

Figure A3-1: Sections through rotor number 2

a) Non-fill in the slots b) Large pores in the opposite-gate end ring

40

Figure A3-2: Sections through shot number 4

a) Gate end ring b) Bar sections, close to gate end ring c) Bar sections, close to opposite-gate end ring d) Opposite-gate end ring

41

Appendix 4: Sections Cut through 6-Inch Long Aluminum Die Cast Rotors

Figure A4-1: Sections through shot number 14 (aluminum rotor)

a) GE bar fill b) GE mid stack bar fill c) OGER mid stack bar fill d) OGER bar fill

42

Figure A4-2: X-rays through end rings of 6-inch long aluminum die cast rotors

a) Gate end ring b) Opposite-gate end ring

43

Appendix 5 – Sections Cut through Two 3-Inch Long Copper Die Cast Rotors

44

45

Figure A5-1: Sections through 3-inch long copper rotor numbers 20 and 21

46

Appendix 6 – Efficiency Data The tables in this appendix show the efficiency data from dynamometer testing of the various rotors.

Rotor No.

Condition

50o C rotor 100oC rotor

8k rpm 10k rpm 8k rpm 10k rpm

10kW 50kW 10kW 50kW 10kW 50kW 10kW 50kW

Baseline Cu Fab 91.90 92.41 91.24 90.60 91.38 91.77 90.43 89.90

26 Cu PCHT

90.60 91.69 91.76 90.34 90.04 91.04 91.22 89.18

27 91.41 91.95 90.73 89.32 91.05 91.34 90.19 88.23

28 Cu no coat

90.74 91.64 89.71 89.74 90.04 90.91 89.51 89.00

29 90.99 91.86 90.40 89.96 90.67 91.17 90.13 89.19

30 Cu skew 3

88.13 88.26 87.15 84.71 87.80 87.68 86.76 84.43

31 89.02 88.54 88.49 84.32 88.62 87.79 87.83 84.42

32 Cu skew 5

89.89 87.81 89.43 83.03 89.37 87.07 88.78 81.38

33 90.36 88.34 89.99 83.41 89.79 87.58 89.36 82.67

34 Al

91.50 91.75 91.16 91.02 91.07 90.99 90.81 90.30

35 89.97 91.36 89.39 90.37 89.56 90.41 88.64 89.44

22 Cu Normal.

91.28 92.11 90.75 90.62 91.09 91.38 90.11 89.38

23 90.88 91.78 90.13 90.76 91.00 91.91 90.53 89.42

24 Cu quench

92.04 92.27 91.45 91.07 91.40 91.88 91.56 90.25

25 91.50 92.01 90.37 90.47 91.14 91.35 90.56 89.80

Baseline Cu Fab 91.78 92.31 91.58 90.86 91.44 91.68 91.34 90.04

37 Cu quench

91.54 91.88 90.48 90.38 90.56 91.35 90.05 89.52

36 91.25 91.94 90.80 90.84 90.53 91.33 90.74 89.96

Table A6-1: Efficiency data from dynamometer testing. Note each data point is the average of

five individual measurements, and the data is shown in the order of testing. The figures are percentages and represent the overall DC-to-shaft efficiency of the motor+inverter system

47

Rotor No.

Condition

50oC rotor 100oC rotor

8k rpm 10k rpm 8k rpm 10k rpm

10kW 50kW 10kW 50kW 10kW 50kW 10kW 50kW

Baseline Cu Fab 0 0 0 0 0 0 0 0

26 Cu PCHT

‐1.30 ‐0.72 0.52 ‐0.26 ‐1.34 ‐0.73 0.79 ‐0.72

27 ‐0.49 ‐0.46 ‐0.51 ‐1.28 ‐0.33 ‐0.43 ‐0.24 ‐1.67

28 Cu no coat

‐1.16 ‐0.77 ‐1.53 ‐0.86 ‐1.34 ‐0.86 ‐0.92 ‐0.90

29 ‐0.91 ‐0.55 ‐0.84 ‐0.64 ‐0.71 ‐0.60 ‐0.30 ‐0.71

30 Cu skew 3

‐3.77 ‐4.15 ‐4.09 ‐5.89 ‐3.58 ‐4.09 ‐3.67 ‐5.47

31 ‐2.88 ‐3.87 ‐2.75 ‐6.28 ‐2.76 ‐3.98 ‐2.60 ‐5.48

32 Cu skew 5

‐2.01 ‐4.60 ‐1.81 ‐7.57 ‐2.01 ‐4.70 ‐1.65 ‐8.52

33 ‐1.54 ‐4.07 ‐1.25 ‐7.19 ‐1.59 ‐4.19 ‐1.07 ‐7.23

34 Al

‐0.40 ‐0.66 ‐0.08 0.42 ‐0.31 ‐0.78 0.38 0.40

35 ‐1.93 ‐1.05 ‐1.85 ‐0.23 ‐1.82 ‐1.36 ‐1.79 ‐0.46

22 Cu normal

‐0.62 ‐0.30 ‐0.49 0.02 ‐0.29 ‐0.39 ‐0.32 ‐0.52

23 ‐1.02 ‐0.63 ‐1.11 0.16 ‐0.38 0.14 0.10 ‐0.48

24 Cu quench

0.14 ‐0.14 0.21 0.47 0.02 0.11 1.13 0.35

25 ‐0.40 ‐0.40 ‐0.87 ‐0.13 ‐0.24 ‐0.42 0.13 ‐0.10

Baseline Cu Fab ‐0.12 ‐0.10 0.34 0.26 0.06 ‐0.09 0.91 0.14

37 Cu quench

‐0.36 ‐0.53 ‐0.76 ‐0.22 ‐0.82 ‐0.42 ‐0.38 ‐0.38

36 ‐0.65 ‐0.47 ‐0.44 0.24 ‐0.85 ‐0.44 0.31 0.06

Table A6-2: Difference in % efficiency for each condition compared to the baseline

measurement for the fabricated copper rotor shown on the first row of Table A6-1

Condition Average Efficiency

Loss

Quench 0.20

Normal 0.50

PCHT 0.57

No coat 0.85

Skew 5 3.81

Skew 3 4.08

Table A6-3: Average loss in efficiency for each condition, as compared with the basline data for

a fabricated copper rotor shown in first row of Tables A6-1 and A6-2