Working with motor locked-rotor test data

Figure 1. Worksheet setup.

By Mike Howell
EASA Technical Support Specialist challenge – how to extrapolate the test extrapolation and that the data should
data to rated voltage with reasonable be plotted on log-log paper and cor-
Editor’s Note: PDFs of this article are assurance of accuracy. If the extrapo- rected to rated voltage using a least
available in English and Spanish in the lation is too far off, we run the risk of square curve fit for maximum accuracy
“Resource Library” of www.easa.com. either rejecting a good motor or accept- [2]. This is where service centers often
ing a bad one. run into trouble – they gather good
l l l l l
The purpose of this article isn’t to reduced-voltage test data but struggle
Locked-rotor testing of three-phase provide detailed procedures for per- with the data analysis and sometimes
squirrel cage induction motors is forming locked-rotor tests, but rather draw the wrong conclusion accord-
used for design validation and qual- to present a practical approach for ingly.
ity control; it also can be a valuable analyzing the reduced-voltage data us-
diagnostic tool. But, this testing isn’t a ing tools that most service centers have Procedure
common task for most service centers. access to at their facilities. Additionally, For this procedure to provide mean-
Two challenges service centers often while this article will focus on locked- ingful results, a locked-rotor test must
face are dynamometer torque capac- rotor test data, the methodology used be performed at a minimum of three
ity and test panel electrical capacity. can certainly be extended to other tests different voltages. This is because we
The work-around is usually reduced- where similar conditions and relation- will use a statistic to indicate how well
voltage testing, which presents another ships exist. our data fits the model and since any
two points define a straight line, we
Background

“
need at least three points to see how
T he pur pose of this It is typical for a reference text to well our model will work. Also, as with
state that the locked-rotor current var- any test, the closer the test voltage is to
ar ticle isn’t to provide ies directly with the voltage and that rated voltage, the better off we’ll be.
detailed procedures for per- the locked-rotor torque varies directly For the data analysis, we will use
forming locked-rotor tests, with the square of the voltage. This is Microsoft Excel due to its prevalence,
but rather to present a a reasonable approach for explaining but most spreadsheet applications
the general concepts. However, for have very similar capabilities and
practical approach for ana-
the purpose of extrapolating reduced- worksheet setup would be similar.
lyzing the reduced-voltage voltage data to rated voltage, this The parrot method will be used – that
data using tools that most simplistic model is far too inaccurate is, inexperienced users should be able
service centers have access for many machines, resulting in cal- to repeat what’s shown here and get a
to at their facilities. Addi- culated locked-rotor current up to 25% working spreadsheet. By inputting the
lower than actual at rated voltage and sample data provided in the article,
tionally, while this article locked-rotor torque up to 60% lower functionality can be verified. If more
will focus on locked-rotor than actual at rated voltage [1]. introductory info is needed for Micro-
test data, the methodology One test procedure used for locked- soft Excel, many free tutorials can be
used can certainly be ex- rotor testing is included in IEEE Std. found on YouTube or other locations
tended to other tests where 112-2004. With regards to reduced- on the web.
voltage testing, it states that a more
similar conditions and re-

”
exact method than the general model
lationships exist. described above should be used for Continued on Page 2

Electrical Apparatus Service Association, Inc. • Copyright © 2016 • www.easa.com • February 2016 1
Working with motor locked-rotor test data
Continued From Page 1

Microsoft Excel worksheets are di- at rated voltage. In

vided into rows and columns, similar B8 and C8, we’ll cal-
to a table or grid. Columns are identi- culate the difference
fied by a letter from left to right (A, B, between our calcu-
C…). Rows are identified by a number lated values and the
from top to bottom (1, 2, 3…). A cell is a machine’s published
specific location within the worksheet values. In B9 and C9
and identified by its column and row. we’ll calculate the
For example, the upper most left cell coefficient of deter-
would be referred to as A1. Cells can mination (R2). The R2
contain different types of input data statistic is used to tell
or they can contain the result of a us how well our data
calculation. fits our model. For
Figure 1 shows one setup that will our purposes, if R2 is
work for our analysis. In this setup, 0.990 or higher, the
row 1 is used for header information model is an accept- Figure 3. Log-log plot of test data.
so text has been entered in A1, B1 and able fit. If R2 is less
locked-rotor torque of 1880 A and 2130
C1 describing what will be input be- than 0.990, repeated testing is recom-
ft-lb respectively. Three locked-rotor
low (voltage, current and torque). Our mended and increasing the number
tests were performed at 155V, 175V
locked-rotor test data will be input for of test voltages to four or more may
and 210V. The linear regression model
each of the three tests into (A2,B2,C2), help. In B10 and C10, we’ll calculate
calculated values in B6 and C6 are
(A3,B3,C3) and (A4,B4,C4). In A5, the index or exponential relationship
both within 5% of the manufacturer’s
we’ll input our rated voltage or the that defines how the locked-rotor cur-
published values and the R2 values for
voltage we want to extrapolate our rent and locked-rotor torque vary with
torque and current both exceed 0.990.
data to. In B6 and C6, we’ll enter applied voltage. This index is simply
It’s reasonable to conclude in this case
functions that will use the test data the slope of the best-fit line.
that the model is a good fit for the
entered to calculate our extrapolated Again, the sample test data used
data and that the motor’s locked ro-
locked-rotor current and locked-rotor are entered in the beige-shaded cells in
tor torque and current are satisfactory
torque at the rated voltage we entered. Figure 1. The machine rated values are
when compared to the manufacturer’s
Note, in calculated cells, the formula entered in the blue-shaded cells and
specifications. If the current varied
entered must begin with “=” as this the calculated results are displayed in
directly with the applied voltage, the
tells the application that you intend the green-shaded cells. The colors have
index would be 1, but as shown in
to perform a calculation. And, values no meaning other than clarification.
Figure 2, the index for this machine
used in the calculations are referred to Once the values shown in Figure 1
is 1.3377. Further, if the torque varied
by their respective cell names. In B7 have been entered into the worksheet,
directly with the square of the voltage,
and C7, we’ll enter the manufacturer’s the values shown in Figure 2 should
the index would be 2, but as shown in
published locked-rotor current and be returned. Don’t become frustrated
Figure 2, the index for this machine
locked-rotor torque for the machine if you have problems at first – use web
is 2.8063.
tutorials as previously men-
Although we are using Microsoft
tioned or feel free to contact
Excel’s built-in functions to do our data
the author for a template.
analysis, Figure 3 is a log-log plot of the
We won’t cover topics like
test data along with linear regression
protecting cells here, but it’s
best fit lines and R2 values provided
important to note that if you
here just to demonstrate graphically
type in a calculated cell after
what was done. The equations and R2
your formula is entered, the
values are generated using Microsoft
value you type will replace
Excel’s trend line tool. The calculations
the formula.
used in our worksheet above give the
Now that the worksheet
same results – graphical representation
is set up, let’s discuss the
is unnecessary. You can see that the ex-
sample test data and results.
ponents shown in Figure 3 (1.3377 for
The sample data is for a 460
current and 2.8063 for torque) are the
V rated induction motor with
Figure 2. Worksheet completed. rated locked-rotor current and Continued on Page 3

Electrical Apparatus Service Association, Inc. • Copyright © 2016 • www.easa.com • February 2016 2
Working with motor locked-rotor test data
Continued From Page 2

same as calculated in our worksheet locked-rotor test data

and are much higher than the general and extrapolate it to
value of 1 for current and 2 for torque. rated voltage. However,
If the general relationship of current a general understand-
varying directly with voltage and ing of the principles
torque with the square of the voltage involved can be use-
had been used, the extrapolated values ful. Due to the physical
would have been significantly lower behavior of the induc-
than published, bringing the motor tion machine, it is well
into question. documented that power
models can be used to
Need locked-rotor current only? accurately describe the
When looking to verify locked- relationship between
rotor current only, it is possible to do applied voltage and the
so approximately without mechani- resulting locked-rotor
cally locking the rotor. If we consider current and locked-rotor Figure 4. Power Curves.
a Y-equivalent circuit model for the You may remember the basic equa-
torque. The model will typically be writ-
three-phase winding, we know that tion for a straight line (y = mx + b).
ten as shown below where “y” is torque
the phase voltage will be equal to the Well, the natural log of the power
or current, x is applied voltage and “B”
line voltage divided by the square root model equation works exactly the
and “M” are parameters that define the
of three. Likewise, if we apply single same way since the natural log of any
relationship.
phase voltage to two line terminals, number is just another number. Once
the phase voltage will be half of the we solve it, we are however left with
applied voltage. For this reason, if we When you look at a power model ln(y) instead of y. It was previously
apply single-phase power to any two graph, it can vary widely from a linear mentioned that taking a logarithm
of the machine line terminals, the re- appearance to a steep curve (see Figure is the inverse operation to exponen-
sulting current will be approximately 4). The power model really isn’t easy tiation. Euler’s constant (denoted e) is
86% of what would be found by ap- to work with when trying to fit data, the base of the natural logarithm. This
plying three-phase power. This can be which is where the “log-log” paper rec- means that if you raise e to the natu-
shown by: ommended in IEEE 112-2004 comes in. ral log of any number, you’re simply
Unfortunately, we do need to use a left with that number. Put another
little math to see how this works but way, eln(y) = y. So, when we solve the
we’re not going through rigorous deri- equation for ln(y) above, whether for
vations or proofs. The term log here is torque, current or another parameter,
short notation for logarithm, which in we raise e to our answer and we get
mathematics is the inverse operation the desired quantity. Again, you don’t
So, if you perform this test and to exponentiation. When log-log paper really need to know this information
measure 100 A, the three-phase current is referenced, that just means to graph to use the tool described earlier in the
would be approximately 100/0.86 = the logarithm of your values (y and x) article but it can be of use sometimes
116 A. If you don’t perform this test instead of the values themselves, or to understand at least some of what’s
at rated voltage though, you need to to use a graph paper whose axes are behind an automated calculation. l
apply the same procedure originally logarithmic scales. We’re going to use
reviewed and do at least three reduced- the natural logarithm, denoted “ln” Bibliography
voltage tests and extrapolate to rated because it is easy to work with. Now, [1] J. D y mo nd, R . O n g a nd P.
voltage. when you take the natural log of the McKenna, "Locked-rotor and
This test can also be used to es- power model equation, something accelerat ion test ing of large
timate locked rotor torque by using interesting happens. No matter what induction machines-methods,
the input power and losses, but this is the shape of the curve, it will resolve problems, and interpretation of
fairly difficult and beyond the scope into a straight line in the form of the the results," IEEE Transactions on
of this article. equation below. And, when you are Industry Applications, pp. 958-964,
working with straight lines instead 2000.
Why this works (optional reading) of complex curves, the data analysis [2] IEEE, Standard Test Procedure for
First, you don’t need any of the becomes much simpler. Polyphase Induction Motors and
information below to analyze your ln y = M ln x + ln B Generators, IEEE 112-2004, 2004.

Electrical Apparatus Service Association, Inc. • Copyright © 2016 • www.easa.com • February 2016 3