Sie sind auf Seite 1von 5

IEEE Transactions on Energy Conversion, Vol. 11, No.

3, September 1996

495

Efficiency Testing of Medium Induction Motors A Comment on IEEE Std 112-1991


Gerald G. Gray
GE Motors & Industrial Systems, Owensboro, KY
Abstract - The energy crisis of 1973 focused attention on the efficiency of polyphase inductiod motors, and the need to have a test procedure that was repeatable and accurate. Medium induction motors, especially those in the size range 1 through 125 horsepower, were important because of the large number in service and the fact that a significant efficiency improvement through redesign was economically feasible. In response to this the ANSVIEEE Std. 112, IEEE Standard Test Procedure for Polyphase Induction Motors and Generators Method B was chosen for testing medium ac motors for efficiency. The current version IEEE Std. 112-1991 Method B contains several requirements for accuracy improvement which have been found to be effective. Even though the Method B procedure with the additional accuracy requirement has been in use since 1984 no rational has been published to support these extra requirements. The paper will discuss the purpose of these improvements, and also point out some weakness in procedure.

Walter J. Martiny, Life Senior IEEE


Ft. Wayne, In The new higher efficiency motors are attractive to users whose applications require considerable running time or run continuously. The value added because of the higher efficiency can be computed using the saving in electric energy. Some process industries and others with continuous duty applications evaluate the savings at over $l/hp for each 0.1% improvement in efficiency. This can be derived as follows. A motor with 95% efficiency operating at rated load will require 0.785 kW input for each 1 hp of output. If the operation is continuous, 8760 hrs/year and the evaluation is for 3 years; then the total energy required is 0.785 X 8760 X 3 = 20,630 k W o u t p u t hp. Therefore a 0.1% improvement in efficiency will result in a decrease in input of (1/.95-1/.951) X 20,630 = 22.8 k w p . At $O.O6/kWh the cost increment is $1.37/hp. Because of the very high value of small differences in efficiency it is necessary to be able to test with precision and repeatability.
11. TEST PROCEDURE

I. BACKGROUND The significant increases in the price of electricity that occurred in the period 1972-1974 focused attention on the cost of operating induction motors. In 1976 the Arthur Little Co. published a report Energy Efficiency and Electric Motors [I] that identified alternating current, ac, polyphase induction motors 1- 125 hp, as the ones most likely to provide an opportunity to save energy. In response to this situation electric motor manufacturers have developed and marketed motors with higher values of efficiency as an alternative to the normal efficiency motors previously offered. The National Association of Electrical Manufacturers, NEMA, issued an addition to their standard, NEMA MG1 [2] defining an energy eflcient motor, and providing values for the efficiency required. The test procedure specified to measure the efficiency of a medium induction motor, motors rated 1 to 500 hp, in this standard is IEEE Std. 112 Method B [3]..

There are many test standards in use for testing polyphase induction motors [4]. Within Std. 112 there are five test procedures, Methods A, B, C , E, and F. These tests require measurements that include all of the machine losses. All but Method F require loading to full load. In addition there are two alternate procedures, El and F1 which use an assumed value for the stray-load loss. The different procedures are offered to provide for the testing of machines which are greatly different in size and mechanical construction. Each procedure requires test facilities suitable for the size and construction of the test machine. Within the US, however, almost all testing of medium ac motors, sizes 1-500 hp, for efficiency is specified to use the IEEE Std 112-1991 Method B. It is intended that the resultant efficiency will be same regardless of the method used. This is illustrated in the results tabulated in Table I. The data in Table I was taken from the testing done on one motor. Most of the elements in the comparison are based on the same test readings to minimize the effect of test variance and emphasizes the difference due to the choice of test procedure.

96 WM 131-3 EC A paper recommended and approved by by the IEEE Electric Machinery Committee of the IEEE Power Engineering Society for presentation at the 1996 IEEWPES Winter Meeting, January 21-25, 1996, Baltimore, MD. Manuscript submitted July 10, 1995; made available for printing November 13, 1995.

088.5-8969/96/$05.00 0 1996 IEEE

496
TABLE I 250 hp 1785 rpm 4000 v 3 phase 60 Hz motor
Test Method Efficiency at rated load (%) Power Input (kW) Stray-load
loss(kW)

A 95.79 194.618

B 95.69 194.812

F 95.56 195.086

F1 94.75 196.746

Method C was not studied for this paper because it requires two identical machines, one operating as an induction motor driving the other as an induction generator. Hence, the data on one motor could not be used to obtain a direct comparison based on the same data. Any attempt to include the Method C test would introduce the effect of test error into the comparison. In theory the results of Method C should be identical to those of Method B.
111. ACCURACY IMPROVEMENTS IN METHOD B

0.671

0.865

1.209

2.796

The difference in efficiency determined by Methods A or B and Methods F or F1 is due to the method of determining the stray-load loss. In Method B the value of stray-load loss is determined by the indirect method. In this method the total loss is measured by subtracting the output power from the input power, then the residual loss is determined by subtracting the sum of the conventional losses from the total loss. The residual is claimed to be the stray-load loss. Similarly in Method A the total loss is the difference between the input and the output. The losses are not segregated as in Method B, but a value for stray-load loss was determined in the same manner for this study. In Method F the stray-load loss is determined directly by the reverse rotation test [5], sometimes called the Morgan test after its author. In Method F1 the stray-load loss is an assumed value, taken from Table 2 of IEEE Std. 112-1991, in this case it is 1.5% of the output. The use of Methods F or F1 avoids the need to test the motor under load which can be a problem for very large machines. Since the motor in this ihstance has significantly less stray-load loss than the Std.112 Table 2 value, there is a large difference between the result from Methods F and F 1. Two other methods, Method E and E 1, not shown in Table I, will produce a result almost identical to that of Method F or F 1. Method E requires the use of the reverse rotation test as in F and E l uses the same assumed value for stray-load loss as for F1. However, the calculation procedures of Forms EE l and F-F l of IEEE Std. l 12- 199l do differ in the handling of the core loss. In Method F the core loss is adjusted from the no-load value to account for the small voltage drop in the stator impedance due to the load current for each load point. In the Method E procedure the core loss is held constant for each load point: The effect on the efficiency in this case in minimal; the result is 95.51% by Method E and 95.56% by Method F.

Test Method B was originally developed to assist the motor designer in identifying the source of the loss in a new machine in order to make improvements and calibrate the mathematical model. It has been found that Method B also provides some advantages in test accuracy not found in Method A. Consequently Method B has been chosen as the referee method in proving the value of efficiency for the energy efJicient motors. The difference in the value for efficiency of the tests in Table I between Method B and Method A is due to test variance. The curve of Fig. 1 will illustrate this.
1400

1200 -1000

200000

400000

600000

800000

100(1000

Torque Squared

Test Points + Least Square regression - - Adjusted Stray-load Loss Line

Fig. 1. Plot of a)tested values of residual loss vs. load in pound- feet squared of torque, b) the least square regression [6]and c) the adjusted value for stray-load loss.

The test points for the values of residual loss shown in Fig. 1 do not form a straight line when plotted against the value of torque squared. The single point at full load used in the determination of full load efficiency by Method A is

497

significantly below (in this case) both the dashed regression line and the solid adjusted stray-load loss line used in Method B to determine the full load efficiency. In theory the stray-load loss is the result of harmonic losses in the conductors and the magnetic material as a result of the leakage flux. The mmf for this leakage flux is directly proportional to the load component of current. Hence it can be assumed that a plot of the stray-load loss against the square of the load current should result in a straight line passing through zero. In a constant voltage machine the load component of current is directly proportional to the torque; therefore it is also proper to expect the stray-load loss to be linear with torque squared. The equation for stray-load loss takes the following form:

The graph in Fig.2 illustrates the concept. If there is a random error in the value for each point, then the distribution of the readings about the mean will take the form of the bell curve of normally distributed data. The regression then finds the best fit of a line through the means for each point when only a few samples of data are available. While this is not a normal use for correlation statistics it has proven to be very effective in defining a good set of data. One of the most frequent errors in motor testing is the measurement of mechanical output, especially the torque. Even though the devices used today are improved over the spring scale of past eras, it is still difficult to eliminate the zero bias in the readings. The procedure of Methods A and B requires a determination of the zero torque by means of a dynamometer correction; however the calibration of many torque transducers can change after determining the zero correction due to a sudden surge in torque as when starting across the line. This procedure by forcing the stray-load point through zero virtually eliminates this most common source of error. Unfortunately when there is a ratio error such as a current transformer ratio error or an error in the length of the torque a m this procedure does nothing to correct it. r Another error which is not corrected is a phase angle shift in the current or voltage due to the use of current transformers or potential transformers. These errors must be corrected by careful calibration. IV. CHANGES IN MOTOR CONDITION Section 4.3.1.1 of IEEE Std. 112-1991, Bearing Loss Stabilization, attempts to address one of the most troubling aspects of test repeatability and correlation between testing laboratories. In Std 112 Stabilization (of the bearing losses) can be considered to have occurred whenever the power input at no load... does not vary by more than 3% between two successive readings at the same voltage at hayhour intervals. . Canadian Std. C390 Sect. 5.1.4 [7] is an identical requirement, the IEC 34-2 standard does not address the issue. The authors have found this requirement to be inadequate in order to assure repeatability in efficiency testing to the level of accuracy demanded by the current uses of the labeled value of efficiency. The data provided by the following test will illustrate the errors possible and the recommended caution in interpreting the results of any test.

LL = AT^ Eq. 1 where: LL is the corrected stray-load loss in kW or W Tis the shaft torque in nm or ft-lb A is a constant o proportionality f
Those familiar with testing motors under load will quickly recognize that the input and output are not constant, but pulsate due to the effects of the phase belt and slot permeance harmonics. Consequently any readings taken at a single point are subject to some test error no matter how carefully the instruments are calibrated and how much effort is put into averaging the data. For this reason Method B requires that the points must form a straight line and that the line must go through zero at no-load. The amount that the scatter of the individual points deviate from a straight line is determined by the coefficient of correlation for the regression line. To qualify as an accurate test the coefficient must be 2 0.90.

Torque Fig. 2 Use of Least Square Regression to Improve Accuracy

498

A . Test Procedure

A typical medium size induction motor rated 60 hp 1800 rpm 3 phase 60 Hz 460 V open construction was used for the testing. The test motor employs two single shielded antifriction bearings, size 3 14 on the drive end and size 213 on the opposite drive end. The bearings are grease lubricated with a polyurea modified grease. This is a very common construction for industrial motors. It provides a very reliable yet economical construction. Provision is made for relubrication to extend the grease life beyond that of the original supply. The motor was then run at no load to monitor the no load power input as proscribed by IEEE Std. 112 as a function of the running time. In order to magnify the effect of the grease loss the test was done at 1/2 voltage so that the other components of no load loss ( core loss, and stator conductor loss) would be smaller than normal.
Three conditions of grease fill were tested. The first test was made with the bearing and grease cavity full. A second test was made with a small amount of grease in the bearings and the remainder of the grease cavity 1/2 full. A third test was made with the bearing lubricated with a small amount of light oil of the same viscosity as the oil in the grease and no grease in the cavity.

The amount of grease is also important to the reliability of the motor. It is normal practice to build motors for stock and for them to remain in storage for several months, in some cases for over a year, while a new facility is being built and put into operation. The electric motor may spend considerable time in transit between the motor manufacturer, the driven equipment manufacturer where it is assembled, as well as transit to the site. During this time the motor bearings are exposed to cycles of moisture and condensation which could cause rust and contribute to early \failure of the bearings. In order to maximize bearing reliability it is desirable to fill t k grease cavity to exclude the air space away from the bearing surfaces. The bearing cavity can also be filled to capacity during relubrication. For this reason the testing included testing with 100% fill of the cavity. An extended run with the bearings 100% full of grease is shown in Fig. 4. This graph shows that the bearings will eventually purge the excess grease and the losses will approach that of a lightly oiled bearing just as the case of the partially filled bearing.

B. Test Results

'#

700 600

I
0

200

400 600 Time hrs

800

1000

Fig. 4. Extended running with 100% grease fill in the bearing cavity
10 20

30

Run Time hrs


+Oil

+1/2 Full - - 100% Full

Fig. 3. Effect of Quantity of Grease in the Bearing and the Cavity on No Load Watts Input

The result of the testing shows that the amount of grease in the bearing and the grease cavity has a dramatic effect on the bearing loss during the efficiency test. In Fig. 3 the effect of the grease loss is shown as a function of time. In all three tests the motor met the requirement of Std. 112 Sect 4.3.1.1 after the first two hours of run time in spite of the fact that in the case of the full packed bearings the loss was nearly 900 W compared to 637 W for the 1/2 packed test and only 433 W when the bearing was free from the effects of grease friction.

The test motor for the grease test is a 60 hp motor with a nominal efficiency of 93.6%. A difference in the losses of 47 Watts equates to a change in efficiency of 0.1%. After eight hours of running the difference in the loss due to the effect of grease fill is 224 watts between a pair of 1/2 full bearings and a pair lightly lubricated. It is 470 watts between a pair of full packed and a pair lightly lubricated. The effect on efficiency of this amounts to a 0.4% to 1.0% decrease in the full load efficiency of the motor. Using the $1.37/hp this evaluates to $329 and $822 per motor for this 60 hp motor.
V. CONCLUSIONS

The accuracy improvements now required by IEEE Std. 112-1991 Method B are adequate to provide repeatable results. The greatest variable in measurement is the

499

condition of the motor at the time of test. This can not be corrected by a change in the test procedure. It is inappropriate to require the motor manufacturer to redesign the bearing system or the recommended regreasing practices to accommodate the occasional need for testing for efficiency. There is however a need to recognize that the condition of the grease in the bearing can have a significant influence on the result of an efficiency test and that the bearing must be in a reasonable condition before the test is conducted. If for reliability reasons the amount of grease is increased for shipment and storage; then the excess must be removed before testing in order to obtain repeatable results . The electrical industry is spending an enormous amount of money to improve the accuracy of the testing equipment in order to measure motor efficiency within f0.5%. In the range of 95% efficient motors this is equivalent to f10% of the total losses of the motor. In order to accomplish this accuracy it is necessary to use precision instruments and frequent calibration. Since motor torque includes transients it is also necessary to employ data logging devices to collect repeated samples of the data at each point to compute an average value. In spite of all of this care the motor condition, for example the amount of grease in the bearings, at the time of the test will have a large impact on the result.

Walter Martiny (LS 91) was born in New Orleans LA on March 28,1926. He graduated from Tulane University in 1948 with a BEEE degree. He was employed by the General Electric Co. as a design engineer in the Industrial AC Motor Department until retirement in 1992. He is now a consulting engineer specializing in electric machines. He is a past chairman of the Induction Machines Subcommittee of the PES, chairman of the Working Group for IEEE Std. 112, and has served on committees of both PES and IAS. He is a registered P.E. in Indiana.

Gerald Gray was born in Amsterdam, NY on July 25, 1946. He attended Hartsville Tech in Hartsville, TN. and Brescia College in Owensboro, Ky. He had been employed by the General Electric CO for 30 years, of which the past 28 years have been with the Industrial AC Motor Department. He is presently the Resident Product Engineer at the GE Motors facility in Owensboro, Ky.

VI. REFERENCES Arthur D. Little, Inc. Acorn Park, Cambridge MA, Energy Efficiency and Electric Motors. Prepared for Federal Energy Administration Washington D.C. National Electrical Manufacturers Association, 2 101 L Street N.W. Washington D.C. NEMA Standards Publication No. MG 1- 1993 IEEE Std. 112-1991 IEEE Standard Test Procedure for Polyphase Induction Motors and Generators Paul Cummings, W.D. Bowers, and Walter J. Martiny, Induction Motor Efficiency Test Methods IAS Transactions Vol. 1A-17, No. 3 MayIJune 1981 pgs. 253-272 Theodore H. Morgan, William E. Brown, and Arthur J. Schumer, Reverse-Rotation Test for the Determination of Stray Load Loss in Induction Machines Transactions AIEE July 1939, Vol. 58 pgs. 3 19-324 Paul G. Hoel, Introduction to Mathematical Statistics Second Edition, John Wiley Sons, Inc. Canadian Standards Association C390-M 1985 Energy Efficiency Test Methods for Three-phase Induction Motors

Das könnte Ihnen auch gefallen