1162 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 25, NO.

4, DECEMBER 2010

Machines and Drives Comparison for Low-Power

Renewable Energy and Oscillating Applications
Ted K. A. Brekken, Member, IEEE, Hannes Max Hapke, Student Member, IEEE, Chad Stillinger, and Joe Prudell

Abstract—The objective of this paper is to analyze, test, and e Electrical.

compare machines and drives in oscillating applications. In par- o Steady-state oscillating operation.
ticular, this paper is focused on low-power wave energy generator L Load.
applications, such as autonomous weather and monitoring buoys
with power requirements in the 100 W and less range. Due to the
oscillating motion of waves, the ocean environment can require Operators
bidirectional and variable speed operation of the generator. In this · Time average.
research, the efficiency of a set of small brushed dc, induction, ·T w Time average over one wave period.
brushless dc, and synchronous reluctance drives and machines
were compared in constant and oscillating operation. The pre-
sented results show that drives and machines used in low-power I. INTRODUCTION
oscillating applications (e.g., ocean wave energy) should not expect
ITH RISING energy costs and improvements in power
a significant derating with respect to their nameplate rating. In
addition, it is shown that the frequency of oscillation (e.g., ocean
wave frequency) has little impact on efficiency.
W electronics and advanced materials in recent years,
wave energy is becoming a more attractive renewable energy
Index Terms—AC machines, dc machines, drives, marine tech- source [1]. Although companies are already installing high-
nology, power generation, reluctance generators. power commercial-scale prototypes in both North America and
Europe, major design and technology questions still have to
be answered to determine the optimal and most efficient wave
NOMENCLATURE energy converter [2]–[9]. In addition to the burgeoning utility-
Parameters and Variables scale market, there are ample opportunities for the application
B Viscous friction coefficient in newton meter/(radian of wave power for small self-powered autonomous buoys.
per second). Autonomous measurement buoys are used by organizations
C Fluid friction coefficient in newton meter/(radian per such as the National Oceanic and Atmospheric Administration
second)2 . (NOAA) to record information about coastal and open ocean
p Power in watts. weather and wave conditions [10]. These buoys range from
T Torque in newton meter. 1.5 m diameter up to 12 m diameter. Self-powered autonomous
Tw Wave period in seconds. buoys are also used for safety and navigation. These buoys
khyst Core hysteretic power loss coefficient in watt per are not grid connected, and require relatively low power to run
hertz. data measurement, data acquisition, lighting, and other auxiliary
keddy Core eddy current loss coefficient in watt per square systems. Typical power requirements may range from 10 W up
hertz. to several hundred Watts.
ηc Efficiency in constant operation. Autonomous buoys are typically powered with a combination
ηo Efficiency in oscillating operation. of solar panels, batteries, and occasionally small wind turbines.
ω Mechanical speed in radian per second. In addition, the wave motion itself has been used to harvest
ωe Machine electrical frequency in radian per second. power for these small buoy systems. An early pioneer in wave
energy conversion, Yoshio Masuda, developed wave-powered
navigational buoys in Japan in 1947 [11]. These early devices
Subscripts
were based on the oscillating water column principle [12]. De-
c Steady-state constant operation.
spite this early work, wave energy extraction is not commonly
used in modern small buoys due to the extra mechanical and
conversion complexities associated with reliably and efficiently
Manuscript received October 26, 2009; revised January 28, 2010; accepted extracting energy from the oscillating input. However, with ad-
March 21, 2010. Date of publication November 8, 2010; date of current version
November 19, 2010. This work was supported by the Department of Energy, vances in power electronics, machines, and energy storage tech-
the Navy, and Columbia Power Technologies. Paper no. TEC-00457-2009. nologies, there is a revived interest in using wave power for this
T. K. A. Brekken, H. M. Hapke, and C. Stillinger are with the Depart- market [13], [14].
ment of Electrical Engineering and Computer Science, Oregon State Uni-
versity, Corvallis, OR 97331 USA (e-mail: [email protected]; Either a linear or rotational generator can be used to convert
[email protected]). the reciprocating motion of a heaving buoy to electrical en-
J. Prudell is with the Columbia Power Technologies, Kelley Engineering ergy. The reciprocating motion can come from floats that move
Center, Oregon State University, Corvallis, OR 97331 USA (e-mail: jprudell@
columbiapwr.com). relative to one another in the wave, or from the motion of an
Digital Object Identifier 10.1109/TEC.2010.2046641 inertial mass contained within the buoy. The inertial approach is
0885-8969/$26.00 © 2010 IEEE
BREKKEN et al.: MACHINES AND DRIVES COMPARISON FOR LOW-POWER RENEWABLE ENERGY AND OSCILLATING APPLICATIONS 1163

Fig. 3. Rotational test bed.

Fig. 1. Weather buoy with rack-and-pinion-based wave energy converter.

Power requirements will also vary. This research focuses on
100 W average, which is toward the high end of the estimated
power needs for this type of application.
In this application, the speed profile of the rotational generator
is very different from standard constant unidirectional speed
typical for most electrical generators. For the example system
shown in Figs. 1 and 2, the mechanical motion of the generator
is an oscillating motion at the ocean wave period, typically 8 s
for swells and 4 s or less for local wind chop waves [12].
This research studies the effect of this oscillatory motion
on the generator efficiency compared to standard unidirectional
constant speed. There are two main objectives:
1) Investigate whether oscillatory operation causes a de-
crease in machine or drive efficiency such that applica-
tion for wave power generation requires a derating of the
machine compared to the nameplate ratings.
2) Investigate if the frequency of oscillation affects the effi-
ciency or overall performance of the machine or drive.
Four different machines and drives have been analyzed:
brushed dc, induction, brushless dc, and synchronous reluc-
tance. The presented research builds upon [16] by adding the
theoretical background to predict and explain the observed ef-
fect of oscillation on machine efficiency.

Fig. 2. Close up of rack and pinion and generator. II. MATERIAL AND METHODS
To compare the different drives, a small-scale (≈200 W) test
attractive as all parts of the system can be sealed from the bed has been used, as shown in Fig. 3. On the rotational test
corrosive salt–water environment. If a rotational generator is bed, there is a fixed permanent magnet brushed dc “generator”
used, it may be necessary to use a linear-to-rotary conversion coupled to four interchangeable “motors.” The available motors
mechanism to convert the heaving, reciprocating motion to ro- are as follows:
tational motion, such as the following: rack and pinion; chain, 1) permanent magnet brushed dc;
belt, or cable drive; or ball screw [15]. This research focuses 2) induction;
on small rotational generators, assuming a linear-to-rotary con- 3) permanent magnet brushless dc; and
version system such as a rack and pinion. A concept of such 4) synchronous reluctance.
a system is shown in Figs. 1 and 2. This is a conceptual ex- The test bed and machines are made by MotorSolver to ap-
ample for illustration; in actual application, the placement and proximately the same physical size and rated speed. The ma-
orientation of the generator, the nature of the linear-to-rotary chines were originally developed for educational use and as
mechanism, and the geometry and size of the buoy may vary. such they are rated conservatively. The induction machine and
1164 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 25, NO. 4, DECEMBER 2010

Fig. 4. Hardware system block diagram.

the synch. rel. machine have lower nameplate-rated power than It should also be noted in the case of a permanent magnet
the brushed and brushless dc machines, but due to the very sim- (or wound field) synchronous machine, and it is simple to use
ilar size and construction and conservative rating, the functional a diode rectifier to convert the varying frequency, amplitude,
difference in nameplate-rated power is not expected to be sig- and direction of the output to dc [2], [20], [21]. This has the
nificant. Any of the four machines used in this research could be advantage of being simple and robust, but has disadvantages in-
reasonably employed for the simulated application (100 W aver- cluding poor THD in the machine and limited control of torque.
age, as explained in Sections I and III). The machine parameters The dSPACE-based setup used in this research allows for rela-
are given in the Appendix. The drive board is made by HiRel tively simple implementation of vector control and synchronous
and has two bidirectional three-phase inverters running off of a rectification, which lends itself well to speed and torque control.
common 42 V dc bus. The switching frequency for pulsewidth
modulation (PWM) is 10 kHz. The drive board has transducers III. DEFINITION OF OSCILLATING OPERATION
for current and voltage sensing. The generator is equipped with
Three tests were conducted for each of the four machines and
an optical encoder for position and speed sensing, and a torque
corresponding drives are as follows
transducer is mounted between the machines for torque mea-
1) Machine and drive efficiency for constant operation.
surements. A dSPACE DS1104 system is used for control and
2) Machine and drive efficiency for oscillating operation.
data acquisition.
3) Machine and drive efficiency for oscillating operation at
For this experiment, the brushed dc “generator” acts as the
rated speed and torque but sweeping the frequency of
load and its armature is connected to a variable resistive load
oscillation.
(i.e., high-power potentiometer). The machine to be tested
In each case, the system is allowed to reach a steady-state op-
(“motor”) is speed controlled using the well-known speed and
erating temperature. In all cases, flux weakening (i.e., extended
torque cascade loop structure, as shown in Fig. 4. The exact
speed operation) was not used to allow a more direct comparison
structure of the torque control varies between the four machines.
between the machines.
For the brushed dc machine, armature current was used to con-
A major objective of the testing was to compare constant-
trol torque. For the brushless dc machine, vector control was
speed efficiency with oscillating efficiency. Therefore, the con-
used with the d-axis aligned with the rotor flux, and the q-axis
cept of rms speed and rms torque is introduced. For constant op-
current used to control torque [17]. For the induction machine, a
eration, rms speed and rms torque are simply the constant speed
V/f control with a slip input was used to control torque. Lastly, for
and constant torque, and the constant machine shaft power is
the synchronous reluctance machine, vector control was used.
equal to the speed multiplied by the torque.
The rotor saliency was aligned with the d-axis, and q-axis and
For the oscillating case, the machine is controlled to follow a
d-axis currents were used in equal proportion to control torque.
sinusoidal speed profile to approximate actuation from an ocean
As the load machine is a dc type connected to a resistance, it
wave
will produce a load torque proportional to speed. This is a good  
approximation to optimal control for ocean wave energy [12]. 2π
ω(t) = ω̂ sin t (1)
It is noted that the power flow is from the motor to be tested Tw
to the “generator” (as labeled in Fig. 3) and to the load resistors. where Tw is the wave period and is typically on the order of 8 s.
This research is focused on evaluation of the effects of oscilla- As the speed profile is sinusoidal, the rms speed is then
tion on the four machine types (“motor” in Fig. 4) when those
ω̂
machines and associated drive control are used in oscillating ωrm s = √ . (2)
wave power generation. It is assumed that general observations 2
about the effects of oscillation on motoring can be drawn for The load motor is a brushed dc machine connected to a load
generation as well, as is standard in machine testing [18], [19]. resistor. Therefore, it produces an opposing torque proportional
BREKKEN et al.: MACHINES AND DRIVES COMPARISON FOR LOW-POWER RENEWABLE ENERGY AND OSCILLATING APPLICATIONS 1165

constant values (as a function of time) are equal to the rms

value of the oscillating waveforms. Using the concept of rms
speed and torque, constant and oscillating applications can be
compared side-by-side, as the average power over one wave
period is the same. However, this also means that for oscillating
applications, the speed and torque will exceed that of constant
operation for the same average power level by the square root
of two at the maxima.

IV. THEORETICAL EFFECT OF OSCILLATION ON EFFICIENCY

A. Friction Loss
Friction is modeled as a static friction term, a viscous friction
term, and a windage friction term, as shown in (15) and (16)
[22]–[24]
Fig. 5. Speed, torque, and power for constant and oscillating modes of oper-
ation, assuming that T w = 8 s for oscillation. Tfric (ω) = Tstatic sign(ω) + Bω + Cω|ω| (15)
pfric (ω) = Tstatic sign(ω)ω + Bω 2 + Cω 2 |ω|. (16)
to speed. The load torque will then be
    To compare the theoretical difference in frictional power loss,
2π 2π the constant speed case and the oscillating speed case are defined
TL (t) = k ω̂ sin t = T̂L sin t (3)
Tw Tw in (17) and (18), where W is a constant
where k is dependent on the value of the load resistor connected ωc = W (17)
to the load machine armature. The rms torque is then  
√ 2π
T̂L ωo = W 2 sin t . (18)
TL ,rm s = √ . (4) Tw
2
The average frictional power loss of the constant speed case
Finally, the instantaneous shaft power is equal to the load is then
torque times the speed. Using the concept of rms speed and
torque, the average shaft power over one wave period (Tw ) is pfric (ωc ) = Tstatic W + BW 2 + CW 3 . (19)
ω̂ T̂L The average frictional power loss of the oscillating speed case
pshaft (t)T w = = TL ,rm s × ωrm s . (5) is
2
√ √
The rated speed and torque for these tests are 283 rad/s and 2 2 8 2
0.354 Nm. Therefore, for constant operation at rated power pfric (ωo )T w = Tstatic W + BW +2
CW 3
π 3π
ω(t) = ωrm s = 283 rad/s (6) ≈ 0.9Tstatic W + BW 2 + 1.2CW 3 . (20)
TL (t) = TL ,rm s = 0.354 Nm (7) It is observed that the power loss terms dependent on the
square of speed are the same between the two cases, as would be
pshaft (t) = pshaft (t) = 100 W. (8)
expected given the definition of rms speed. However, the static
For oscillating operation at rated power friction term is underemphasized in the oscillating case, whereas
  the fluid friction term is more significant. This is expected as
2π
ω(t) = 400 sin t rad/s (9) the oscillating case momentarily reaches higher speeds than
Tw
the constant case, though the rms speed is the same, and the
400 dependence of the fluid friction power loss term on the cube of
ωrm s = √ = 283 rad/s (10)
2 the speed makes these momentary high speeds very significant.
  There are no cooling fins on the machines used in this re-
2π
TL (t) = 0.5 sin t Nm (11) search, and their full rated speed is not unusually high, so
Tw
that it is expected that the fluid friction loss term is not sig-
0.5 nificant. If the fluid friction coefficient C is negligible, then
TL ,rm s = √ = 0.354 Nm (12)
2 pfric (ωo )T w < pfric (ωc ).
 
2 2π
pshaft (t) = ω(t)TL (t) = 200 sin t W (13) B. DC Machine Conduction Loss
Tw
As stated in Section III, the load will provide a torque pro-
pshaft (t)T w = 100 W. (14)
portional to speed. Assuming that the machine torque is propor-
Fig. 5 graphically shows the speed, torque, and power in tional to current, the constant speed operation and oscillating
the constant and oscillating modes of operation. Note that the speed operation currents are defined in (21) and (22), where I
1166 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 25, NO. 4, DECEMBER 2010

is a constant
idc,c = I (21)
 
√ 2π
idc,o = I 2 sin t . (22)
Tw
The power loss due to heating of the windings is the same
between the constant and oscillating cases
pcond,dc (i) = i2 R (23)
pcond,dc (idc,c ) = I R = pcond,dc (idc,o )T w .
2
(24)

C. AC Machine Conduction Loss

In the case of the ac machines, constant operation is ac steady
state, and the oscillating case is ac operation with a long time- Fig. 6. (a) ia c , c . (b) ia c , o . (The constants I = 1 pu and W e = 4 Hz for
scale modulation at the wave period Tw . The current magnitude illustration.)
is assumed to be proportional to torque. The current electrical
frequency is assumed to be directly proportional to rotational The conduction power loss for the constant steady-state op-
frequency. For induction machines, this means that the slip is eration is
assumed to be small, and the magnetizing current is small com-
pared to the torque producing component of the current. pcond,ac (iac,c ) = I 2 R. (30)
For the constant operation case For the oscillating application
  ⎛ ⎞2
√  Tw
iac,c = I 2 sin ωe,c dt (25) 1
pcond,ac (iac,o )T w = ⎝ i2ac,o dt⎠ R. (31)
Tw 0
where I is a constant and ωe,c is the electrical frequency in radian
per second for constant operation. By (17), the mechanical speed The electrical frequency constant (We ) is much larger than the
in the constant mode of operation is constant. The electrical wave oscillation frequency (2π/Tw ). Under this condition, it
frequency is proportional to the mechanical frequency by the can be shown by numerical integration that
number of machine pole pairs and, therefore,
pcond,ac (iac,o )T w ≈ pcond,ac (iac,c ). (32)
ωe,c = We (26) Therefore, the conduction losses due to the torque producing
where We is a constant. Equation (25) then becomes component of the current are effectively the same for an ac
√ machine in either constant or oscillating mode.
iac,c = I 2 sin (We t). (27)
D. AC Machine Core Loss
For the oscillating case, both the current magnitude and the
electrical frequency are modulated at the wave frequency. This For an ac machine, the core loss in the stator due to the rotat-
is because the torque and speed change together at the wave fre- ing field is made up of two components: eddy current losses and
quency, as shown hysteretic losses. The hysteretic loss is proportional to the fre-
√ in Fig. 5. As the mechanical speed and torque
have maxima 2 larger in the oscillating case compared to the quency of field change, and the eddy current loss is proportional
constant case, to the frequency of field change squared [25]
√ the current magnitude and electrical frequency
are likewise 2 larger at the maxima ωe,c = We (33)
       
√ 2π √ √ 2π
iac,o = 2 sin t I 2 sin ωe,o dt (28) ωe,o = We 2 sin t (34)
Tw Tw
 
√ 2π Pcore (ωe ) = khyst |ωe | + keddy |ωe |2 (35)
ωe,o = We 2 sin t . (29)
Tw
where We is a constant.
The constant and oscillating ac currents are illustrated in For the constant case, the average core power loss is
Fig. 6. Note that the oscillating operation√exceeds the constant
Pcore (ωe,c ) = khyst We + keddy We2 . (36)
operation magnitude and frequency by 2. An electrical fre-
quency of We = 4 Hz is used for clarity of illustration, but in For the oscillating case, the average core power loss is
actuality, the electrical frequency would likely be much higher, √
2 2
from 50 Hz up to several hundred hertzs, depending on the Pcore (ωe,o )T w = khyst We + keddy We2
number of poles and the effective gear ratio of the mechanical π
conversion system. ≈ 0.9 khyst We + keddy We2 . (37)
BREKKEN et al.: MACHINES AND DRIVES COMPARISON FOR LOW-POWER RENEWABLE ENERGY AND OSCILLATING APPLICATIONS 1167

Fig. 7. Brushed dc machine efficiency: (a) constant and (b) oscillating at 8 s period.

Therefore TABLE I
EFFICIENCY AT 1 PU SPEED, 1 PU TORQUE, AND 1 PU POWER
Pcore (ωe,o )T w < Pcore (ωe,c ). (38)

The core power loss should be less for the case of oscillating
operation for the brushless dc machine, the induction machine,
and the synchronous reluctance machine.
TABLE II
E. Theoretical Efficiency Summary EFFICIENCY AT 0.71 PU SPEED, 0.71 PU TORQUE, AND 0.5 PU POWER

The theoretical calculations show that almost all of the major

causes of power loss are equal or less for oscillating opera-
tion, compared to constant operation. The only exception is for
friction, depending on the significance of higher order friction
terms. Therefore, it is expected that the machines should actu-
ally be more efficient in the oscillating operation compared to TABLE III
constant operation. EFFICIENCY AT 1 PU SPEED, 0.5 PU TORQUE, AND 0.5 PU POWER

V. TESTING PROCEDURE
As outlined at the beginning of Section III, three tests are
conducted: efficiency for constant operation, efficiency for os-
cillating operation, and efficiency as a function of oscillation
TABLE IV
(i.e., wave) frequency. EFFICIENCY AT 0.5 PU SPEED, 1 PU TORQUE, AND 0.5 PU POWER

A. Efficiency Versus Speed Versus Torque

1) Constant Operation: The drive speed is set to the desired
value and the load machine resistive load is adjusted until the
shaft torque reaches the desired value. The drive and machine
efficiency is recorded at this point. This process is repeated for
five evenly spaced points between 0 and 1 per unit (pu) in speed 2) Oscillating Operation: The drive speed is set to follow
and five evenly spaced points between 0 and 1 pu in torque. This a sinusoidal reference point with a period of 8 s such that the
yields the machine and drive efficiency at 25 evenly spaced rms value of the sinusoidal speed reference is the desired value.
points in the torque-speed plane. (Note that for the constant The load machine resistive load is adjusted until the rms value
operation, a speed of 1 pu means that ω(t) = 283 rad/s and a of the shaft torque reaches the desired value. The average drive
torque of 1 pu means that TL (t) = 0.354 Nm.) and machine efficiency over 8 s is recorded at this point. This
1168 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 25, NO. 4, DECEMBER 2010

Fig. 8. Efficiency as a function of wave period at rated power for (a) drive and (b) machine.

process is repeated for three evenly spaced points between 0 B. Efficiency Versus Wave Period
and 1 pu in speed and three evenly spaced points between 0 and The dependency of efficiency on wave period Tw has been
1 pu in torque. This yields the machine and drive efficiency at tested at rated power. The result is shown in Fig. 8. The rms
nine evenly spaced points in the torque–speed plane. (Note that
speed and torque are both 1 pu, and the average power over one
√ oscillating operation, a speed of 1 pu means that ω(t) =
for the wave period is 1 pu.
283 2 sin((2π/T√ w )t) rad/s and a torque of 1 pu means that
TL (t) = 0.354 2 sin((2π/Tw )t) Nm, such that the rms speed
and torques are the same between the constant and oscillating
operation.)
VII. DISCUSSION
B. Efficiency Versus Wave Period A. Machine Efficiency
The drive speed is set to follow a sinusoidal reference such The theoretical calculations predict that machine efficiency
that the rms value of the sinusoidal speed reference is 1 pu. should be higher for oscillating application. Tables I–IV indeed
The frequency of the sinusoidal reference is the variable to be show that this is the case. In nearly every combination of speed
swept. The load machine resistive load is adjusted until the rms and torque for the brushed dc, induction, and brushless dc ma-
shaft torque is 1 pu. The average drive and machine efficiency chines, the efficiency is higher in the oscillating operation than
over one wave period (Tw ) is then recorded for eight different in constant operation.
periods of sinusoidal drive speed reference between 2 and 16 s. There are a few exceptions. In every case, the synchronous
reluctance machine was less efficient in oscillating operation.
VI. RESULTS The theory would predict this if for that particular machine, the
windage losses are much larger than for the other machine types.
A. Efficiency Versus Speed Versus Torque This is unlikely as each machine is manufactured to approxi-
As outlined in Section V, the efficiency of the drive and mately the same size and mass. Therefore, it is likely that there
machine in each mode of operation (constant and oscillating) is an additional heat-related or core-loss-related loss mechanism
was recorded over many points in the torque and speed plane. that is not adequately accounted for.
From this, a contour plot can be created, as shown in Fig. 7, Another anomaly is the poor oscillating performance of the
for the brushed dc machine. The machine efficiency at 1 pu induction machine for low speed and high torque (Table IV). It is
speed and 1 pu torque is 75.0, and in oscillating operation, it is hypothesized that this is due to reduced cooling associated with
76.1. Linear interpolation is used between points to create the the lower speed. A more complex machine model and analysis
contour. This testing procedure was applied to all four machines that accounts for increased rotor bar and stator conductor resis-
and associated drives. For brevity, only the contour plots for the tance due to heating would be necessary to test this hypothesis.
dc machine are shown. All contour plots can be found in [16]. This is a good topic for future work.
The results for all four machines are summarized in Tables Lastly, in Table III, it was not possible to collect data for the
I–IV at four different operating points representing full power, brushless dc machine in oscillating application at high speed and
half power at equal speed and torque, half power with speed low torque. This was due to friction in the load motor exceeding
emphasized, and half power with torque emphasized. the desired torque in this lightly loaded condition.
BREKKEN et al.: MACHINES AND DRIVES COMPARISON FOR LOW-POWER RENEWABLE ENERGY AND OSCILLATING APPLICATIONS 1169

B. Drive Efficiency There are several areas for future research. The results con-
A detailed treatment of the theoretical drive efficiency is com- clusively showed a decrease in drive efficiency for oscillating
plex as it requires full modeling and analysis of both the conduc- operation. There are opportunities for detailed theoretical treat-
tion and switching losses of the drive as a function of duty ratio ment of power electronics efficiency in oscillating operation,
and current, where these duty ratios and currents are themselves especially for oscillating AC drives. Also, this research can be
dc, steady-state ac, and ac modulated at the wave frequency. extended to include higher power machines and drives, such as
This is a complicated analytical task that is an excellent candi- would be used for utility-scale generation.
date for future research. However, the experimental results of
the drive efficiency measurements are included in Tables I–IV. APPENDIX
The trend is clear: contrary to the machines, the drives univer- MACHINE PARAMETERS
sally suffer a decrease in efficiency in the oscillating mode. This
suggests there is a high-order loss√ mechanism in the drive that
becomes very significant at the 2 higher speeds and torques
momentarily experienced in the oscillating application.

C. Overall Efficiency
Although the drive efficiency decreases in the oscillating
mode, in general, the decrease in efficiency is not as large as the
gain in efficiency in the machine. The net effect is an overall in-
crease in efficiency when operating in the oscillating mode. The
exception is the synchronous reluctance machine, which suffers
a large overall decrease in efficiency as both the machine and
drive perform poorly in the oscillating mode.

D. Effect of Oscillation Period on Efficiency

The result is conclusive that for wave periods of 4 s and
up, efficiency has no significant dependence on wave period.
However, the data do suggest that the efficiency of both the
drive and machine will decrease for periods of 3 s and less.
This is likely due to the extra current required to meet the
torque demand caused by the acceleration and deceleration of
the system inertia, as this becomes more significant at higher
wave frequencies (i.e., lower periods).

VIII. CONCLUSION
The goal of this research was to determine the effect of oscil-
lating operation on small power machines and drives used for
wave energy generation and oscillating applications. The salient
conclusions are as follows:
1) In general, machine efficiency is improved in the oscil-
lating operation, as predicted by the theoretical analysis.
This is true for the brushed dc, induction, and brushless
dc machines. The synchronous reluctance machine was an
exception, as it showed a strong reduction in efficiency in
the oscillating operation.
2) Drive efficiency universally decreased for the oscillating
operation.
3) In general, the increase in efficiency in oscillation for the
machines was greater than the decrease for the drive, re-
sulting in a net increase in system efficiency in oscillating
operation.
4) The frequency of oscillation over large ocean swell periods REFERENCES
(several seconds and greater) has no effect on efficiency
[1] J. Vining and A. Muetze, “Economic factors and incentives for ocean wave
for oscillating machines and drives. At small periods (i.e., energy conversion,” IEEE Trans. Ind. Appl., vol. 45, no. 2, pp. 756–763,
high frequencies), there is a decrease in efficiency. Mar./Apr. 2009.
1170 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 25, NO. 4, DECEMBER 2010

[2] M. Leijon, O. Danielsson, M. Eriksson, K. Thorburn, H. Bernhoff, [23] F. Al-Bender and J. Swevers, “Characterization of friction force dynam-
J. Isberg, J. Sundberg, I. Ivanova, E. Sjöstedt, O. Agren, K. Karlsson, ics,” IEEE Control Syst. Mag., vol. 28, no. 6, pp. 64–81, Oct. 2008.
and A. Wolfbrandt, “An electrical approach to wave energy conversion,” [24] A. Harnoy, B. Friedland, and S. Cohn, “Modeling and measuring friction
Renewable Energy, vol. 31, no. 9, pp. 1309–1319, 2006. effects,” IEEE Control Syst. Mag., vol. 28, no. 6, pp. 64–81, Oct. 2008.
[3] M. Leijon, H. Bernhoff, O. Agren, J. Isberg, J. Sundberg, M. Berg, [25] P. Sen, Principles of Electric Machines and Power Electronics, 2nd ed.
K. Karlsson, and A. Wolfbrandt, “Multiphysics simulation of wave en- New York: Wiley, 1997.
ergy to electric energy conversion by permanent magnet linear generator,”
IEEE Trans. Energy Convers., vol. 20, no. 1, pp. 219–224, Mar. 2005.
[4] M. Mueller and N. Baker, “A low speed reciprocating permanent mag-
net generator for direct drive wave energy converters,” in Proc. Power
Electron., Mach. Drives Int. Conf., Jun. 2002, pp. 468–473.
[5] J. Vining, T. Lipo, and G. Venkataramanan, “Design and optimization of a Ted K. A. Brekken (M’06) received the B.S., M.S., and Ph.D. degrees from the
novel hybrid transverse/longitudinal flux, wound-field linear machine for University of Minnesota, Minneapolis, in 1999, 2002, and 2005, respectively.
ocean wave energy conversion,” in Proc. Energy Convers. Congr. Expo. He is currently an Assistant Professor in Energy Systems at Oregon State
(ECCE), 2009, pp. 3726–3733. University, Corvallis, where he is also a Codirector of the Wallace Energy Sys-
[6] T. K. A. Brekken, J. Prudell, A. von Jouanne, and M. Stoddard, “A novel tems and Renewables Facility (WESRF).
permanent magnet tubular linear generator for ocean wave energy,” in He was involved in studing electric vehicle motor design at Postech in
Proc. Energy Convers. Congr. Expo. (ECCE), Sep. 2009. Pohang, South Korea, in 1999. During 2004-2005, he was also involved in
[7] S. Narayanan, B. Murthy, and G. Rao, “Dynamic analysis of a grid- studing wind turbine control at the Norwegian University of Science and Tech-
connected induction generator driven by a wave-energy turbine through nology (NTNU), Trondheim on a Fulbright scholarship. His current research
hunting networks,” IEEE Trans. Energy Convers., vol. 14, no. 1, pp. 115– interests include control, power electronics, and electric drives; specifically dig-
120, Mar. 1999. ital control techniques applied to renewable energy systems.
[8] H. Polinder, M. Damen, and F. Gardner, “Linear pm generator system Dr. Brekken is a recipient of the National Science Foundation (NSF)
for wave energy conversion in the aws,” IEEE Trans. Energy Convers., CAREER Award.
vol. 19, no. 3, pp. 583–589, Sep. 2004.
[9] H. Polinder, B. Mecrow, A. Jack, P. Dickinson, and M. Mueller, “Con-
ventional and TFPM linear generators for direct-drive wave energy con-
version,” IEEE Trans. Energy Convers., vol. 20, no. 2, pp. 260–267, Jun.
2005.
[10] National data buoy center. (2010). [Online]. Available: http://www.
ndbc.noaa.gov/ Hannes Max Hapke (S’04) received the Intermediate Diploma in electrical
[11] J. Brooke, Wave Energy Conversio (Elsevier Ocean Engineering Series). engineering from the University of Stuttgart, Stuttgart, Germany, in 2006, and
Amsterdam, The Netherlands: Elsevier, 2003. the M.S. degree in electrical engineering in 2009 from Oregon State University,
[12] J. Falnes, Ocean Waves and Oscillating Systems. Cambridge, U.K.: Corvallis, where he is currently in the Department of Electrical Engineering and
Cambridge Univ. Press, 2002. Computer Science.
[13] S. Bastien, R. Sepe Jr,, A. Grilli, S. Grilli, and M. Spaulding, “Ocean wave His current research interests include renewable energy technologies and
energy harvesting buoy for sensors,” in Proc. Energy Convers. Congr. modern control.
Expo. (ECCE), 2009.
[14] A. Grilli, S. Grilli, M. Spaulding, and J. Cheung, “Experimental and
numerical study of spar buoy-magnet/spring oscillators used as wave en-
ergy absorbers,” presented at the Int. Offshore Polar Eng. Conf., Lisbon,
Portugal, 2007.
[15] E. B. Agamloh, A. K. Wallace, and A. von Jouanne, “A novel direct-drive
ocean wave energy extraction concept with contact-less force transmission
system,” Renewable Energy, vol. 33, no. 3, pp. 520–529, 2008. Chad Stillinger received the B.S. degree in mechanical engineering from
[16] T. K. A. Brekken, H. Hapke, and J. Prudell, “Drives comparison for George Fox University, Newberg, OR, in 2004. He is currently working to-
reciprocating and renewable energy applications,” in Proc. Appl. Power ward the Ph.D. degree at Oregon State University, Corvallis.
Electron. Conf. (APEC), Feb. 2009, pp. 732–738. His current research interests include designing wave energy converters,
[17] N. Mohan, Advanced Electric Drives: Analysis, Control, and Modeling implementing different motor/generator control schemes, analyzing prototype-
Using Simulink, 2001 ed. Saint Paul, MN: MNPERE, 2001. testing data, and researching new designs for wave energy extraction methods.
[18] IEEE Guide: Test Procedures for Synchronous Machines, IEEE Standard
115-1995, 1995.
[19] IEEE Standard Test Procedure for Polyphase Induction Motors and Gen-
erators, IEEE Standard 112-2004 (Revision of IEEE Standard 112-1996),
2004.
[20] P. Brooking and M. Mueller, “Power conditioning of the output from a
linear vernier hybrid permanent magnet generator for use in direct drive Joe Prudell received the B.S. and M.S. degrees in electrical engineering in
wave energy converters,” Generation, Transmiss. Distrib., vol. 152, no. 5, 2005 and 2007, respectively, from Oregon State University (OSU), Corvallis,
pp. 673–681, Sep. 2005. where he is currently with Columbia Power Technologies LLC, researching new
[21] C. Boström, E. Lejerskog, M. Stålberg, K. Thorburn, and M. Leijon, technologies in ocean wave energy converters.
“Experimental results of rectification and filtration from an offshore wave He has studied electrical systems design for CH2M Hill, and controls design
energy system,” Renewable Energy, vol. 34, no. 5, pp. 1381–1387, 2009. for NACCO Materials Handling Group. He has also worked for PGE Marking
[22] B. Armstrong and Q. Chen, “The z-properties chart,” IEEE Control Syst. as a Transmission Engineer performing extensive studies on high-voltage (HV)
Mag., vol. 28, no. 5, pp. 79–89, Oct. 2008. and HV dc transmission lines throughout the Pacific Northwest.