SPE

Society of Petroleum Engineer'S

SPE 18186
Beam Pump Motors: The Effect of Cyclical Loading
on Optimal Sizing
by M.O. Durham, U. of Tulsa, and C.R. Lockard, OXY USA Inc.
SPE Members

Copyright 1988, Society of Petroleum Engineers

This paper was prepared for presentation at the 63rd Annual Technical Conference and Exhibition of the Society of Petroleum Engineers held in
Houston, TX, October 2-5, 1988.
This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the
author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the
author(s}. The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers
presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Permission to copy is
restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment of
where and by whom the paper is presented. Write Publications Manager, SPE, P.O. Box 833836, Richardson, TX 75083-3836. Telex, 730989 SPEDAL.

II

steady design conditions used by the manufacturer

to rate the motor.
The performance of motors changes when the motor
is applied to cyclic loads. The effective horsepower
and efficiency is reduced. A method of calculating
the. effective ratings is presented. A comparison of
operations on both conventional and unconventional
pumping units is outlined.

Research into cyclic loading effect on motors is

beginning at the Power Applications Research
Center of The University of Tulsa.
The authors
have also previously published papers that address
the problem [1,2].
Lea discusses some of these
problems in another paper at this session.
Gibbs
has approached parts of this problem in some of
his works [3]. This paper will illustrate the effect
of the cyclic loading on motor efficiency.

Artificial lift by beam pumping units has existed

almost as long as the petroleum industry.
It
comprises about 85% of the artificial lift systems
used today.
In the past twenty years, electric
motors
have
become
the
predominant
drive
mechanism for these units.
Literally thousands of
motors
have
been
applied
to
the
units.
Unfortunately, the performance of the motors
operating under the cyclic loading of a beam
pumping unit remains elusive.

To effectively understand the impact of cyclic

loading, the electric power relationship must be
comprehended. Several different manifestations of
power are used in a motor. The input to the motor
is electrical power. The output is mechanical shaft
power. Within the motor are both mechanical and
electrical losses [ 4].
EleCin - Elec1oss = Mech1oss + Mechsbaf\ . ( 1)

The economics of operating motors at the most cost

effective point should be apparent. Electrical costs

The input power is determined by the electrical

equation.
Pin = .{3 V I cos a ........ (2)

range :from twent;y to thirt;y-:five percent o:f the

direct operating cost :for petroleum production.

Furthermore, the initial capital costs of motors is

minimal compared to the cost of operating the
equipment.
With
this
large an impact on
profitability, it behooves the operator to seriously
consider the motor performance.

The square root of 3 is a factor that arises because

of the symmetry of three-phase power.
For a
balanced power system, the factor is a constant. If
the system is unbalanced, an alternate factor must
be used.
The voltage is the potential that exists
between two different phase wires. The current is
the flow that exists in a phase wire.

Motors are designed for use on a steady load. The

beam pumping unit causes a motor to operate under
cyclic conditions. During part of a cycle the motor
is heavily overloaded.
During other parts of the
cycle, the machine is operating not as a motor but
as a generator.
This type operation will radically
change the performance of the motor from the
References

and

illustrations

at

end

of

Cos a is the power factor for the circuit.

The
angle, a, exists between the voltage and the
current. The power factor will be reduced as the
motor becomes heavily overloaded or underloaded.
Hence the angle between the voltage and current

paper.
57

BEAM PUMP MOTORS:

THE EFFECT OF CYCLICAL LOADING ON OPTIMAL SIZING

will increase. Voltage is a constant for the power

system.
For a fixed load, if the power factor
decreases, then the current must increase as
illustrated in Figure 1.

SPE18186

some
corrections
must
be
made
for
cyclic
conditions.
Because of the large number of
variables already under consideration, only NEMA D
motors
will
be
considered
in
this analysis.
Furthermore, speed-torque characteristics are not
readily available for the various designs of ultrahigh-slip motors.

The rna jor electrical losses in a motor are core

losses and copper losses.
The core losses are
dependent upon the iron of the machine and
electromagnetic fields.
This loss is primarily
influenced by the voltage.
Since voltage is
constant with load, this loss remains constant.

Several
approximations
are
necessarily
made
because of the present inadequate information
within the industry. Nevertheless, the procedure is
completely accurate and provides a method of
comparison between two motors applied to the same
load.
It also provides a comparison between two
different loads applied to the same motor. What is
not available is an absolute value to compare the
performance of any arbitrary pieces of equipment.
As we develop more data and better methods of
instrumentation, a more complete representation will
be offered.

The copper losses are dependent on the wire size

and result in heating,
The resistance is a
characteristic of the motor and will not change
significantly over the normal operating range of the
motor.
While
the
input power is
linearly
proportional to current, the electrical heat losses
have an exponential dependance on current.
Pbeat
= I 2 R (3)
The electrical power that is internally converted to
mechanical power uses the familiar relationship for
rotation. The torque and angular speed determine
the power.
Pinternat
= (T whnternat , , .. , ........ , , ...... , , , , , ... ( 4)

The
basic
pumping
unit
data required
for
determining motor effective efficiency is the cyclic
load information of net torque versus crank angle.
The data is taken from one of the three sources
identified earlier.
The data is normalized to the
torque
required
for
one
horsepower.
The
normalization permits using the same curve for
multiple different loads.
The torque curve is
simply multiplied by a scaling factor to obtain any
other loads.

The mechanical losses are comprised primarily of

friction and windage. This is influenced by speed.
Since speed changes only a limited percentage with
load, this loss is not as dramatic as the effect of
!ZR heating,
However, for high slip motors, this
value can become substantial because of wide
variations in shaft speed.

The motor performance data is shaft torque, shaft

speed and efficiency, The data is digitized so that
the exact motor torque corresponding to the scaled
unit torque can be obtained. The shaft speed and
efficiency of the motor is determined from the
motor performance data at the load torque.

To obtain the shaft power in terms of torque and

speed, the inertia of the motor's rotor and sheave
must be removed from the internal performance.
(Tw)sbaft = (Twhnt -Mech1oss -Inertiaint
(5)

The manufacturer's motor performance data is based

on a steady-state load.
However, the electrical
response time is fast enough that the instantaneous
steady state parameters can be applied for any
dynamic loading.

The output shaft power of the motor is the

mechanical power that goes into the beam pump.
The loss in the belts and the inertia of the gear
pulley must be considered to obtain power into the
gear box.

Each of the calculated data points are instantaneous

values corresponding to a pumping-unit crankangle. The output horsepower of the motor at that
crank angle is the product of the instantaneous
torque and speed of the motor.
hpsbaft = T(in-lbs)
w(RPM) I 63025 ........ (6)

From this short tutorial, it is apparent that there

are many effects on the power train that have not
been readily quantified.
This paper will only
address the efficiency of the motor. The efficiency
will change as the load on the motor cycles with
the pumping unit. The other effects on the power
train will necessarily have to be addressed in
future research.

Since the load is cyclic, the duration at each crank

angle is slightly different. The time duration for a
crank angle is the reciprocal of the speed in RPM
multiplied by the number of discrete crank angle
segments used.
The crank position was divided
into 100 increments, which results in 36 segments
per cycle.
t = 1/w(RPM)
1/36(seg/rev)
60(sec/min) ..... (7)

A fundamental requirement for determining the

effect of cyclic loading on motor efficiency is to
define the load. Because of the wide divergence of
information, several different ~;~ources of pumping
unit torque data are available.
These include the
static torque information from API llE [5], computer
simulated torque such as provided by Nabla [3],
and dynamic torque measurements from operating
pumping units [2].

The instantaneous energy is the product of the

instantaneous horsepower and the time duration at
that speed.
The losses within the motor are derived from two
sources. The no-load loss is a characteristic of the
motor.
It is published by the manufacturer.
Alternately u can be measured using a wattmeter.

Similarly, the
motor
losses
and
speed-torque
characteristics must be known. The manufacturer's
motor data is for steady state operation, therefore

58

MARCUS 0. DURHAM and CLARK LOCKERD

SPE18186

The instantaneous load losses of the motor are

calculated from the motor efficiency, n, and the
motor shaft horsepower.
Ploss

Ploss

Pin

Pout

= [hPshaft

(8)

I n] - hpshaft

........... (9)

reasonable
approximations
are
used
in
Two
determining the motor losses. First, the losses are
based on absolute torque.
The internal losses are
effectively symmetrical whether the machine is
consuming
power
or
is
generating
power.
Deviations are primarily affected by the variations
in voltage drop and resulting current that exists
within the machine because of direction of power
flow [6]. Second, if the efficiency falls below 50%,
the losses are equivalent to no load losses.
Because the curves are so compressed at the lower
loads, it is difficult to determine any significant
variation from the no-load losses.
The energy consumed in losses is calculated from
the load losses and the time duration at the point.
Wlosses
= Flosses
d t (10)

By integrating the time at each crank angle over

the entire cycle, the total time in seconds per
stroke oan be found.
Inverting the value and
multiplying by 60 seconds per minute yields the
effective unit speed in strokes per minute.
36
SPM = 60 I E dt ... ( 11)
1
The effective cyclic energy from the motor shaft is
the integration of the discrete energy over a
complete cycle.
36
E dW (12)
kWh = 2685.6 kWhlhp-sec
1
The effective horsepower on . the motor shaft is the
rate of doing work over a complete cycle of the
pumping unit. The effective energy over the total
time for a cycle provides the power.
This is not
calculated from root-mean-square (RMS) voltage and
current. Rather it is the numerical integration of
the instantaneous values of torque and motor
speed.
The total losses in the motor is simply the
summation of all the discrete losses at each crank
angle.
The effective cyclic efficiency of the motor is the
total energy out the shaft divided by the total
energy plus the total losses.
neff

= kWhshaft

I [kWhshatt+kWhloss]

...... (13)

These calculations are dependent

on several
parameters. The performance is for a given motor
operating on a particular type pumping unit. The
normalized pumping unit torque is scaled to provide
a known effective load on the motor.
Figure 2
illustrates a
10
hp
motor operating
on a
conventional unit with an effective load of 5.8 hp.
Figures 3 provides similar performance for the 10
hp motor on an unconventional unit with a load of
5.8 hp.
Figure 4 is a 25 horsepower motor on a
cqnventional unit with an effective load of 13.1
horsepower.

Observation of the instantaneous losses compared to

the shaft horsepower illustrates the impact of
loading on the motor. The 10 hp motor in Figure 2
is operating at an effective horsepower of 5.8
horsepower
over
the
entire
cycle
on
the
conventional unit.
At first, it would appear that
the motor is lightly loaded to little more than onehalf its capacity.
The instantaneous values of horsepower reveal an
entirely different explanation.
The peak load on
the shaft is 22 horsepower. Conversely, the motor
generates 10 horsepower at the minimum output.
The range of load is over 32 horsepower on a
machine designed for
a
steady load
of 10
horsepower.
The losses also encompass a wide swing in values.
First, the losses are never negative.
Even when
the machine is generating power back into the
system, energy is consumed by internal losses of
the motor.
Second,
the losses increase much
quicker for peak overloading of the motor, than for
loads near the rated horsepower.
Figure 5 portrays the exponential second order
nature of the losses.
The plot is load versus
losses for a particular motor.
The losses are
comparatively flat near no-load on the motor shaft.
These low-load losses are predominated by the
electromagnetic core losses as well as the friction
and windage .mechanical losses.
As the load
increases, the current increases.
Then the IZR
heat losses tend to dominate.
The analysis is based on the assumption that motor
shaft output horsepower is equal to pumping unit
shaft input horsepower. This does not include the
effect of belt-slippage. The effect of belt-slippage
will be to lower the peak requirements on the
motor.
Another parameter that has not been included is
the effect of average temperature rise in the motor
windings.
With cyclic loading, the
internal
temperature will be expected to increase because of
the high instantaneous losses.
Moreover, even
when the unit is coasting, it still has losses.
Because
of
the
temperature
coefficient
of
resistivity, the resistance also increases when the
temperature of the conduction. increases.
In turn,
the losses have additional increase. Hence the total
losses are probably greater than this projection
indicates.
.~JlJ.'lJJ.. ....Ql!.....C..Y.HY.lJ.Ji

The calculations have involved a particular motor

operating at only one effective horsepower. If the
torque scaling factor for the unit is adjusted over
a range of values, a suite of performance points
can be obtained for the motor.
When the scaling factor has a very low value, the
effective cyclic motor load is essentially no-load.
The maximum scaling factor is determined by one of

59

BEAM PUMP MOTORS:

THE EFFECT OF CYCLICAL LOADING ON OPTIMAL SIZING

two limits. The scaling factor times the unit torque

can not exceed the motor torque; neither can the
motor horsepower be exceeded,

SPE18186

kWh of energy at 100% efficiency, 83782 kWh at 78%,

and 96103 kWh at 68% efficiency, The difference in
efficiency is equivalent to 12321 kWh per year. If
energy cost is $.06/kWh, the savings is $739 per
year. This provides a payout in the neighborhood
of one year simply from energy savings.

The effective motor horsepower and efficiency at

each of the scaling points can be used to plot a
particular motor's cyclic efficiency curve.
Similar
calculations can be made for other pumping unit
configurations. The result is depicted in Figure 6.
The lower curve represents a conventional unit
from API llE.
The middle curve is a conventional
unlt performance from a computer simulation by
Nabla.
The
upper
curve is
the
API
llE
performance for an unconventional unit.

An even greater improvement in efficiency is

realized when a larger frame motor is implemented.
Typically the losses do not increase proportional to
the motor rating.
A motor rated at 25 hp has a
maximum efficiency of 90%, while one rated at 10 hp
has a maximum efficiency of 85%.
A load of 8
horsepower represents 63% efficiency using a 10
horsepower motor and 84% using a 25 horsepower
motor. This is an improvement of over 21 points or
a 33% improvement in energy usage.

The API 11E curves are based on static loading for

the
unit.
The
computer
simulation closely
represents
dynamic
running
conditions
which
include inertia.

The information indicates motors should be sized to

allow for peak gear box torque and a cyclical load.
The motor rating will be significantly larger than
average polished rod horsepower.
The pumping
unit performance and the actual motor performance
should be compared to arrive at the optimum
selection.

From these curves several observations can be

made. On a conventional unit, the maximum cyclical
horsepower which can be started is 50% of the
motor rating, as shown at the maximum load point
of curve 1.
However, the motor can drive an
average cyclical load at 90% of its rating as shown
by the maximum point of curve 2. The phenomenon
is familiar to those who have had to "rock" a
pumping unit to start it.

Although the calculations are tedious by hand, with

computer programs it is very viable to consider
actual unit geometry and actual motor curves.
Computer hardware is also being developed that will
permit direct monitoring of the true electrical
horsepower and unit load.'
This will further aid
maintaining minimum operating expenses.

The best efficiency point occurs near 40% of the

balanced cyclical load.
This is at the peak of
curve 2.
The best efficiency point of an
unbalanced unit is achieved by restricting the
cyclical load to 25-30% of the motor rated load. To
obtain the best efficiency, a motor should have a
rating about 2.5 times the polished rod horsepower
requirement on a conventional unit.

In applying motors to projected loads, it is not

always feasible to develop sophisticated models.
Because of the shape of the motor performance
curve and the pumping unit torque curve, in
general, the motor should be sized to be 40% of the
average mechanical horsepower calculated from
producing rates, head, and downhole efficiency,

The curves indicate the unconventional geometry

unit has significantly better performance than a
conventional unit.
The
static and
dynamic
performance are very close.
This is the result of
the better average to peak torque relationship.
Therefore, a motor can start a load with an average
cyclical load equal to the motor rating.
The
maximum efficiency of the motor occurs when its
cyclical load is 60% of its rated size,

There are always risks to making generalizations.

Nevertheless, the extensions can be useful in daily
operations where detailed data is not always
available. For reasonable efficiency over the cyclic
load range of a NEMA D motor operating on a
conventional unit, the instantaneous motor load
should not exceed 150% to 175% of the motor
nameplate rating. This is very apparent by looking ,
at the loss breakpoint on Figure 5.

Using
the
same
motor,
the
efficiencies are
consistently greater operating on an unconventional
geometry.
Five percent less energy will be
consumed by the same load on an unconventional
unit compared to a conventional beam pumping unit
when properly sized.
If the motor is heavily
loaded, the unconventional geometry has as much as
30% improved efficiency over the conventional unit.

When checking the pumping unit balance, the

ammeter should not exceed 1.5 - 1.75
times the
nameplate current. The unit should be balanced to
reduce the current, if possible, If the motor load
continues to be too large during the peaks, a
larger motor should be installed. The smaller motor
may be adequate to start and run the unit, but the
larger
motor
will
operate
more
efficiently.
Subsequently, the power cost will be reduced.

If a

motor is sized so that the effective cyclical

load is 40% of the motor rating, the amount of
electricity consumed is significantly reduced.
If
the motor has a 75% load, a typical efficiency is 68%
while a 40% load has a typical efficiency of 78%.
The 10 point improvement in efficiency will more
than offset the investment in the larger equipment.

.Q.'lllB.H...C.QN.fi.J.JJ.B.H!!..'f..l.QN.Ji
In the preceding analysis only the peak
effective horsepower effect on motor efficiency
been compared. If the motor is sized to have
a 40% average load, other performance criteria
be considered.

As an example, an average cyclical load of 10 hp

running continuously for one year consumes 65350

60

and
have
only
must

SPE18186

MARCUS 0. DURHAM and CLARK LOCKERD

One detrimental effect of overs1z1ng motors is the

lowering of the power factor from .87 to . 7 4.
However, this is easily corrected with capacitors.
One of the most positive effects is that available
starting torque becomes approximately twice as
great. This assures the unit will start even if it is
unbalanced. By using the larger sizing procedure,
starting torque is always available.
Another
significant
improvement
is
reliability.
Larger frame size motors have larger bearings
capable
of
longer
life
with
less
loading.
Furthermore the larger units have less I2R heating.
Since insulation life is reduced by approximately
one-half for each 10oc rise in temperature, the
mean time between failure should improve.
Because of the slope and shape of the speed curve,
the speed change of the unit and resulting motor
slip is greater at full load than one-half load. At
20 hp the speed is 1120 RPM for a slip of 6.6%,
while at 10 hp the speed is 1160 RPM for a slip of
only 3.33%.
This increased slip is one reason for
decreased motor efficiency as loading increases.

Electric energy consumption can be dramatically

improved by properly sizing the motor to consider
the consequence of cyclic loading on efficiency.
Motor performance curves are designed for steady
state operation. By using the motor torque, speed
and efficiency curves with the pumping unit torque
and crank angle curves, the optimum motor size can
be calculated.
1.
The best efficiency will be achieved with a
motor operating at 40-50% of its rating.
2.
When checking the pumping unit balance, the
ammeter should not exceed 1.5 - 1. 75
times the
nameplate current.
3.
A motor provides adequate starting torque
for a conventional unit only when the motor rating
is at least twice the effective cyclic load.
4.
Unconventional geometry units are at least 5%
more efficient than a conventional unit at the same
load.
5.
When
a
motor
is
heavily
loaded,
an
unconventional unit is as much as 30% more
efficient than an equivalent conventional unit .
6.
Efficiency performance curves
were
not
available
for
various
ultra-high-slip
motors.
Therefore additional work remains to be done in
that area.
We have addressed only one area of cyclic loadin~
effects on equipment performance.
It is obvioue
the characteristics are considerably different fran
what is often expected. As indicated, many othez
electrical parameters and mechanical characteristic!
remain to be identified. These will be topic of somE
of our future research.

61

1.

Durham, Marcus 0.
and
Lockerd,
Clark:
"Effect of Cyclic Loading on Motor Efficiency,"
IEEE Transactions on Industry Applications,
New York, Nov /Dec 1988.

2.

Durham, Marcus
0.
and
Lockerd,
Clark:
"Optimal Sizing of Motors for Beam Pumping
Units,"
IBBB Proceedings or Petroleum and
Chemical
Industry
Conference,
Calgary,
September, 1987.

3.

Gibbs, Sam:
May, 1987.

4.

Chapman, Stephen J.:

Electric Machinery
Fundamentals, McGraw-Hill, New York, 1985.

5.

Data Curves, Nabla, Midland, TX,

.S.P..~.gJfiG~t.i9.n. . . . . .f9.r.: . . . . . .P.Y..mP.i.ng. . . . . . P..ni.t..~.;, . . . . . . S..P.~.c..:::J.lE.:

American Petroleum Institute, Dallas, 1986.

6.

Durham, Marcus 0.:

Analysis or Induction
Generators on Unbalanced Power Systems,
Oklahoma State University, Stillwater, OK, 1985.

MOTOR CURRENT AT DIFFERENT POWER FACTORS

OUTPUT HP AND LOSSES

API CONY. TORQUE / 10 HP NEMA D

25
EFFECTIVE LOAD ON THE MOTOR 5.8 HP

1-

20

w
n::
n::
:J
u
w

rnrn

15

9
D:

CURRENT INTO
THE MOTOR

10

>

15

~Q.

...

1-

rn

D:

:J

J:

-5

-10

-15

CURRENT IN PHASE

~ITH

VOLTAGE

20

40

60

Fig. 1-Effect of power factor.

...

- - LOAD ON THE MOTOR

80 100 120 140 160 180 200 220 240 260 280 300 320 340 360

CRANK ANGLE
+ LOSS IN THE MOTOR
Fig. 2-10-hp motor losses on API conventional.

OUTPUT HP AND LOSSES

API UNCONY. TORQUE
13

OUTPUT HP AND LOSSES

10 HP NEMA D

API CONY. TORQUE

60

12

rnrn

10

rnrn

D:

D:

~
...

EFFECTIVE LOAD ON THE MOTOR =5.8 HP

a
(
d

...

D:

~
Q.

...rn
D:

0
J:

25 HP NEMA D

'(/)

50

11

EFFECTIVE LOAD ON THE MOTOR -13.1 HP

rc

...

40

30

Q)
J-&

20

Q)
10

D:

...rn

-10

Q.

D:

0
J:

-20
-30

-1

--40
0

20

40

60

- - LOAD .ON THE MOTOR

80 100 120 140 160 180 200 220 240 260 280 300 320 340 360

CRANK ANGLE
+ LOSS IN THE MOTOR
Fig. 3-10-hp motor losses on API unconventional.

20

40

60

- - LOAD ON THE MOTOR

80 100 120 140 160 180 200 220 240 260 280 300 320 340 360

CRANK ANGLE
+ LOSS IN THE MOTOR
Fig. 4-25-hp motor losses on API conventional.

seE 18186
HP VS LOSSES
10 HP NEMA D
26

24 -

22 20 -

18 16 14 -

12 10 8

c
c

C CD

[j]

-5

-15

25

15

HP OUT OF MOTOR
Fig. 5-10-hp motor loss as a function of load.

25 HP NEMA D MOTOR EFFICIENCY

0.88
0.86
0.84
0.82
0.8
0.78

0.76

Ill

0.74

0.72

0.68

Ill
II:

:I

0.7

0.66
0.64
0.62
0.6
0.58
0.56
0.54
0

12

16

PUMPING LOAD - HP
Fig. 6-25-hp motor efficiency on cyclic load.

63

20

24