IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 34, NO.

2, APRIL 2009 123

Peer-Reviewed Technical Communication

Study of a Wave Energy Converter Connected to a Nonlinear Load
Cecilia Boström, Student Member, IEEE, Rafael Waters, Student Member, IEEE, Erik Lejerskog, Olle Svensson,
Magnus Stålberg, Student Member, IEEE, Erland Strömstedt, and Mats Leijon, Member, IEEE

Abstract—This paper presents experimental results from a

wave energy converter (WEC) that is based on a linear generator
connected to a rectifier and filter components. The converter-filter
system is installed onshore, while the linear wave generator oper-
ates offshore a few kilometers from the Swedish west coast. The
power from the generator has been rectified with a diode bridge
and then filtered using a capacitive filter. Performance of the
whole conversion system was studied using resistive loads con-
nected across the filter. The aim was to investigate the operational
characteristics of the generator while supplying a nonlinear load.
By changing the value of the resistive component of the load, the
speed of the translator can be changed and so also the damping of
the generator. The power absorbed by the generator was studied
at different sea states as well. The observations presented in this
paper could be beneficial for the design of efficient wave energy
conversion systems.
Index Terms—Electrical system, experimental results, linear
generator, power absorption, wave energy.

I. INTRODUCTION

T HE threat of climate change in combination with in-

creasing energy demand globally has intensified the
research and development in renewable energy systems during Fig. 1. Illustration of the WEC and the electrical system.
the last decades. Many similarities about the importance in
finding alternative energy resources exist between the ongoing
debate today and the debate that was carried out during the higher today. The technological development has also made
1970s. At that time, however, the main cause of the debate was it easier for researchers to share technical information and
a drastic increase in oil price. Ocean wave energy is one of research results; see, for example, [2]. One of the challenges in
several renewable energy technologies that emerged in Europe coming up with a marketable wave energy technology is to find
during the 1970s [1]. In spite of this, wave energy technologies an efficient and economically viable system that can survive
are still in a relatively early stage of development, but with the the harsh climate conditions of the sea. Several strategies to
technological developments and the experiences that have been overcome these problems have been attempted and have re-
gained over the past 40 years, the possibility of succeeding is sulted in a number of different wave energy technologies, some
of which were presented in [3]–[6]. In the latest report from the
International Energy Agency (IEA), 81 different wave energy
projects were examined, and out of them, 13 were estimated
Manuscript received October 07, 2008; accepted February 03, 2009. First
published April 07, 2009; current version published May 13, 2009. This to have reached the stage of having a full- or near-full-scale
work was supported in part by the Swedish Energy Agency, Vattenfall AB, prototype at sea [7].
the Gothenburg Energy Research Foundation, Draka Cable AB, the Göran The wave energy converter (WEC) presented in this paper is
Gustavsson Research Foundation, Statkraft AS, Ångpanneföreningen, Fortum,
Vargöns Research Foundation, Falkenberg Energy AB, and the Wallenius
based on a direct drive linear generator placed at the ocean floor
Foundation. as shown in Fig. 1. The translator, which is mounted with per-
Associate Editor: H. Maeda. manent magnets, is connected to a buoy at the ocean surface, and
C. Boström, R. Waters, O. Svensson, M. Stålberg, E. Strömstedt, and M. thereby, the motion of the waves is transferred to the generator.
Leijon are with the Swedish Centre for Renewable Electric Energy Conver-
sion, Department of Engineering Science, Uppsala University, Uppsala 751 21, The electrical characteristics of the generator have been calcu-
Sweden (e-mail: [email protected]; Rafael.Waters@angstrom. lated in a finite element method (FEM) simulation tool. Based
uu.se; [email protected]; [email protected]; Er- on these results, the generator is designed to produce 10 kW at a
[email protected]; [email protected]).
E. Lejerskog is with the Seabased Industry AB, Uppsala 751 83, Sweden
translator speed of 0.7 m/s, a line-to-line voltage of 200 V [root
(e-mail: [email protected]). mean square (RMS) value] when it is connected to a nominal
Digital Object Identifier 10.1109/JOE.2009.2015021 load that is 4 . The efficiency of the generator in this case is

0364-9059/$25.00 © 2009 IEEE

124 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 34, NO. 2, APRIL 2009

Fig. 2. Circuit diagram of experimental setup.

calculated to be 86%. The design was made to match the signif- addition of the rectifier and capacitor. In addition to the infor-
icant wave height on the Swedish west coast. However, the gen- mation given on the operation of a single WEC, this study will
erator is both electrically and mechanically designed to handle give a hint on how a farm of WECs will operate when connected
large overloads. More information about the numerical model to the grid. Simulations and laboratory experiments have been
of the generator can be found in [8]. The numerical model is done on this topic before (see, for example, [10] and [19]), but
verified with experimental results in [9]. the area lacks offshore experimental results.
A traditional high-speed rotating generator requires several
stages between the generator and the source, the waves. This II. EXPERIMENT
will result in a system with mechanical moving parts such as
The main object of this paper is to study how the generator
gear boxes that need maintenance and have a limited lifetime.
works when it is connected to a nonlinear load. These results
By using a direct drive generator that is directly coupled to the
will be the foundation for the design of a grid connected system
waves, the needed mechanical parts can be reduced. Having a
containing several WECs.
construction that uses few moving parts can result in a robust
The resistance in the load is varied between four different
generator and this is believed to increase the lifetime and reduce
values. A circuit diagram of the system is shown in Fig. 2 and
the cost of the WEC [10]. A WEC based on this technology was
a more detailed description of the electrical system used in the
launched outside the Swedish west coast in March 2006 [11].
experiment is given in [20]. The WEC is connected to a diode
Initially, the linear generator was connected to a resistive load
rectifier and a capacitive filter with a capacitance of 24.3 F on
and the power absorption of the generator was studied in detail
both positive and negative dc sides. The capacitance is designed
[9], [12], [13].
according to the sea state at the site and with help of voltage data
Linear generators, unlike a conventional generator running
from the WEC. This is described in more detail in [20].
at constant speed, result in somewhat more complicated energy
During shorter periods of time (approximately 10 s), due to
conversion system. The motion of the translator will vary in
their large power storage capacity, the ultracapacitors will give
speed and direction, and as a result, the voltage and current will
a similar effect as if the generator were connected to a dc bus
have irregular amplitude and a varying frequency, and the output
with a constant dc level.
power peaks will reach levels several times higher than the av-
Different resistive loads are connected in parallel with the
erage power production. This results in a generator that has to
filter. The total resistive values that can be selected are
be overdimensioned in relation to its average power generation
9.17, 13.75, 18.34, and 27.5 . These values were
in order to handle intermittent overloads. Consequently, compo-
measured before operation with an accuracy of four significant
nents in the electrical system also need to have a higher power
numbers. According to the resistor characteristics, a resistance
rating compared to the average power.
variation of 5% can be assumed.
There are various energy conversion schemes for direct drive
By changing the resistive load, different damping conditions
generators [14]–[17]. In our work, the basic strategy is to rectify
for the whole conversion system are achieved. The three-phase
the voltage from several WECs with a passive diode rectifier and
currents , , and can be used to calculate the power losses
then interconnect them in parallel on a common direct current
in the transmission cable and generator windings
(dc) bus. Passive diode rectifiers are cheaper and are less com-
plex compared to active rectifiers, therefore, we chose to use a (1)
diode rectifier in this study. A capacitive filter is then added.
After this step, the dc voltage will be inverted, transformed, and where is the generator winding resistance 0.44 , and
connected to the grid. Several layouts for this are possible de- is the cable resistance 0.54 . If is added to the power
pending on, e.g., farm size and distance to shore [18]. The in- that is measured before the rectifier , an approximate value of
terconnection of several WECs in parallel on a common dc bus the total absorbed power can be obtained. Mechanical and
will lead to a possible smoother power output [19]. iron losses in the WEC will be neglected in these calculations
In this study, the same offshore generator is used as in [12],
but it is now connected to a nonlinear load resulting from the (2)
BOSTRÖM et al.: STUDY OF A WAVE ENERGY CONVERTER CONNECTED TO A NONLINEAR LOAD 125

Fig. 4. Power P produced over the loads measured over 48 h.

Fig. 5. Average values of absorbed power calculated over 30-min periods and
plotted against the average energy flux during the same time. The data represents
a total of 713.5 h of measurements.

Fig. 3. Data from the generator sampled during 30 s. (a) and (b) The three phase
voltages and three phase currents measured before the diode rectifier. (c) The
A 48-h sequence of the power is plotted in Fig. 4,
translator position in the generator. (d) Calculated and filtered speed of the trans- where was 13.75 . The power on the dc side has been
lator. (e) Generator output power. smoothened considerably compared to the generator output
power in Fig. 3(e).
In Fig. 5, the average mean value of the absorbed power by the
(3) generator is calculated over a 30-min period and plotted versus
the mean sea state during the same time. The data represents a
The position of the translator can be calculated from the gen- total of 713.5 h of measurements. The sea state is calculated
erator pole configuration and the phase order of the windings. using wave data from a wave measuring buoy located about
A change in phase order means that the translator has changed 50 m from the generator [21].
its direction. Every period of the voltage represents the fact that
the translator has moved the length of two pole widths, which IV. DISCUSSION
is equivalent to 0.1 m. Because the time and the position of the
translator are known, the translator speed can be calculated. The One of the most important issues studied in this paper is the
speed curve is filtered to clear out some transients due to noise effect of the rectifier and filter on the power output from the
and to get a smooth shape. WEC. When the generator was directly connected to resistive
loads as in [12], the extreme fluctuations and quality of the pro-
duced power made it incompatible for grid connection. The re-
III. RESULTS sults of Fig. 4, however, indicate that long term nearly stable
Voltage and current data were sampled during three months power output generation is possible.
and the results form the study that is presented here. Fig. 3 Ideally, the linear generator would produce power all through
shows a typical data for generator operation characteristics for the translator motion with an exception at the turning points.
a period of 30 s. is 13.75 in the results in Fig. 3. The Consider the three phase voltages and three phase currents
three phase voltages , , and , and three phase currents plotted in Fig. 3(a) and (b) between 4.7 and 13.5 s, 16 and
, , and are shown in Fig. 3(a) and (b), respectively. The 21.5 s, and 25 and 28 s, where this is the case. However, during
dc link voltage was approximately 70 V during this period. other time intervals, no power is produced during translator
The corresponding position and speed of the translator can be motion. This is due to the level of the dc bus voltage. A conse-
seen in Fig. 3(c) and (d), respectively. Finally, the power pro- quence of a high dc bus voltage is that the translator must have
duced by the generator (sea cable losses are included) is shown a high speed to produce energy to the system. The connection
in Fig. 3(e). between the speed of the translator and the generated power is
126 IEEE JOURNAL OF OCEANIC ENGINEERING, VOL. 34, NO. 2, APRIL 2009

clearly visible in Fig. 3(d) and (e). The control of the dc bus V. CONCLUSION
voltage in the electrical system will be a key parameter.
This paper has studied the electrical output and motion of a
Furthermore, with a fixed dc voltage, the three phase currents
linear generator-based WEC connected to a rectifier and filter
distribution decides the power production; compare the curves
in offshore conditions. The most significant finding, in regards
in Fig. 3(b) and (e). Thus, by regulating the dc level, the speed of
to the technology’s viability as a supplier of energy to the elec-
the translator and the generated current can be controlled appro-
tric grid, was that the output power is smoothed to a great ex-
priately. Due to this effect, it would be desirable to have different
tent compared with the fluctuating power of the waves. The
dc voltages during different sea states if the power production
power absorption for the studied load cases continued to in-
is to be optimized. This is possible to achieve if, for example,
crease with increasing damping of the generator. An optimal
a dc/dc booster or a variable (tap changed) transformer is used
level of damping was thus not found and would require more
together with the inverter. Then, the output voltage can be kept
tests at higher levels of damping. An increase in damping, how-
constant despite the variation in dc voltage level.
ever, will lead to more losses in the system, as a result of higher
At the upper turning point, the translator is generally motion-
currents, and is a factor that needs to be taken into account. The
less for longer periods of time compared to the lower turning
overall energy absorption from the waves shows a trend of lev-
points as shown in Fig. 3(c) between 14 and 15.3 s and 21.8
eling off towards more energetic sea states, a phenomenon that
and 23.8 s. This result may indicate a nonoptimized generator
is in agreement with previous studies.
operation. The difference in time of standstill can be attributed
In future studies, a system of several generators will be con-
to ill-dimensioned springs, i.e., the spring constant is too small
sidered. When this is done, the allocation of the generators has
and is unable to force the translator to faster downward motion
to be chosen carefully and the number of generators to be con-
as in the optimal case. Another possible explanation could be
nected to each dc bus should be investigated in detail. By con-
that the buoy moves in the surging direction before it starts to
necting several generators on the dc bus in a suitable way, the
descend. The phenomena that can be seen in Fig. 3(c), where
needed energy storage will decrease exponentially and the costs
the translator sometimes loses its speed and stops in its upward
of the capacitor will be substantially reduced. Costs and power
motion, e.g., 3.8 and 4.5 s and 28.3 and 29 s, may also be
optimization algorithms are also going to be considered at this
caused by the shape of the wave.
stage.
Fig. 5 shows that the highest power absorption is achieved at
the lowest load, i.e., at the highest damping factor. Therefore,
finding an optimum load resistance for optimal energy produc- ACKNOWLEDGMENT
tion is difficult from these results. In two of the four load cases, The authors would like to thank J. Goncalves and K. Yuen for
there are only samples in the region with relatively mild sea their contribution to the experimental setup. They would also
states. As a result of this, the shape of the fitted trend lines in like to thank O. Danielsson for his help with figure layouts and
Fig. 5 can be different. If the optimal damping of the gener- calculation tools and N. Theethayi for the help with the written
ator is to be found, more experiments with lower resistive loads material.
are needed. Fig. 5 further indicates that the relative power ab-
sorption decreases at more powerful sea states. This indicates
REFERENCES
a power limit due to the WEC design. When the energy flux in
the waves increases, in general, the wave period will be larger [1] J. Falnes, “A review of wave-energy extraction,” Mar. Struct., vol. 20,
no. 4, pp. 185–201, 2007.
and the proportion of the absorbed energy by the buoy will de- [2] C. M. Johnstone, K. Nielsen, T. Lewis, A. Sarmento, and G. Lemonis,
crease because of its dimensions [1]. This is also in agreement “EC FPVI co-ordinated action on energy: A European platform for
with previous studies [9], [12]. When studying the amount of sharing technical information and research outcomes in wave and tidal
energy systems,” Renewable Energy, vol. 31, pp. 191–196, 2006.
absorbed power, it is of interest to compare it to the wave height [3] A. Muetez and J. G. Vining, “Ocean wave energy conversion—A
and wave period. These two parameters will have different im- survey,” in Proc. IEEE Conf. Ind. Appl./41st Ind. Appl. Soc. Annu.
pacts on the absorbed energy and such a study will give a more Meeting, Oct. 2006, vol. 3, pp. 1410–1417.
detailed picture of the optimal damping at a certain sea state. [4] R. Henderson, “Design, simulation, and testing of a novel hydraulic
power take-off system for the Pelamis wave energy converter,” Renew-
This study is suggested to be done when the system is upgraded able Energy, vol. 31, pp. 271–83, 2006.
with more loads. [5] J. P. Kofoed, P. Frigaard, E. Friis-Madsen, and H. C. Sorensen, “Pro-
The authors also acknowledge that there is always an uncer- totype testing of the wave energy converter wave dragon,” Renewable
Energy, vol. 31, pp. 181–189, 2006.
tainty in measurements due to the accuracy of measuring instru- [6] M. Leijon, H. Bernhoff, O. Ågren, J. Isberg, J. Sundberg, M. Berg, K.
ment and data or the accuracy of electrical system devices in the E. Karlsson, and A. Wolfbrandt, “Multiphysics simulations of wave
considered range of operation. Those effects are assumed to be energy to electric energy conversion by permanent magnet linear gen-
small. However, since the measurements were carried out during erator,” IEEE Trans. Energy Convers., vol. 20, no. 1, pp. 219–224, Mar.
2005.
a longer period, the value of the load resistance could have been [7] Implementing Agreement on Ocean Energy Systems, “Review and
changed due to deteriorations caused by the system and deteri- analysis of ocean energy system development and supporting policies,”
orations caused by the ambient environment. A more accurate Jun. 2006, A report by AEA energy and environment on behalf of
Sustainable Energy Ireland for the IEA’s implementing agreement on
result could have been obtained if the value of the load would ocean energy systems.
have been measured continuously during the test period. The [8] O. Danielsson, M. Eriksson, and M. Leijon, “Study of a longitudinal
speed and position curve was filtered to clear out some noise flux permanent magnet linear generator for wave energy converters,”
that occurred in the calculations. Therefore, some points have Int. J. Energy Res., vol. 30, pp. 1130–1145, 2006.
[9] M. Eriksson, R. Waters, O. Svensson, J. Isberg, and M. Leijon, “Wave
been smoothened out, and the curve gives a rough value of the power absorption: Experiments in open sea and simulation,” J. Appl.
speed. Phys., vol. 102, 2007, 084910.
BOSTRÖM et al.: STUDY OF A WAVE ENERGY CONVERTER CONNECTED TO A NONLINEAR LOAD 127

[10] M. A. Mueller, “Electrical generators for direct wave energy con- Erik Lejerskog received the M.Sc. degree in engi-
verters,” Inst. Electr. Eng. Proc. Generation Transm. Distrib., vol. neering physics from Uppsala University, Uppsala,
149, no. 4, pp. 446–56, 2002. Sweden, in 2007.
[11] J. Engström, R. Waters, M. Stålberg, E. Strömstedt, M. Eriksson, J. Is- From 2006 to 2007, he was a Research Engineer at
berg, U. Henfridsson, K. Bergman, J. Asmussen, and M. Leijon, “Off- the Division for Electricity, Uppsala University. He
shore experiments on a direct-driven wave energy converter,” in Proc. is now working at Seabased Industry AB, Uppsala,
7th Eur. Wave Tidal Conf., 2007, pp. 1078–1082. Sweden, where he is developing electrical systems
[12] R. Waters, M. Stålberg, O. Danielsson, O. Svensson, S. Gustafsson, E. for wave power.
Strömstedt, M. Eriksson, J. Sundberg, and M. Leijon, “Experimental
results from sea trials of an offshore wave energy system,” Appl. Phys.
Lett., vol. 90, 2007, 034105.
[13] M. Stålberg, R. Waters, O. Danielsson, and M. Leijon, “Influence of
generator damping on peak power and variance of power for a direct
drive wave energy converter,” J. Ocean Mech. Artic Eng., vol. 130, no.
3, 2008, 031003. Olle Svensson received the University Diploma
[14] P. R. M. Brooking and M. A. Mueller, “Power conditioning of the with specialization in electrical engineering from
output from a linear vernier hybrid permanent magnet generator for use Blekinge Institute of Technology, Karlskrona,
in direct drive wave energy converters,” Inst. Electr. Eng. Proc. Gen- Sweden, in 1998 and the M.Sc. degree in engi-
eration Transm. Distrib., vol. 152, no. 5, pp. 673–681, 2005. neering physics from Uppsala University, Uppsala,
[15] E. A. Mahrous, N. A. Rahim, W. P. Hew, and K. M. Nor, “Proposed Sweden, in 2007, where he is currently working
nine switches five level inverter with low switching frequencies for towards the Ph.D. degree at the Department of
linear generator applications,” in Proc. Int. Conf. Power Electron. Electricity.
Drives Syst., 2005, vol. 1, pp. 648–653. He was a Test Engineer at Ericsson Energy
[16] M. Molinas, O. Skjervheim, P. Andreasen, T. Undeland, J. Hals, T. Systems from 1988 to 2003. He joined the
Moan, and B. Sørby, “Power electronics as grid interface for actively Department of Electricity, Uppsala University,
controlled wave energy converters,” in Proc. Int. Conf. Clean Electr. in 2004, as a Research Engineer.
Power, 2007, pp. 188–195.
[17] K. Rigbergs, P. Lürkens, M. Wendt, S. Schröder, U. B. ouml;ke, and
R. W. De Doncker, “High-efficient soft-switching converter for three-
phase grid connections of renewable energy systems,” in Proc. Int. Magnus Stålberg (S’06) received the M.Sc. degree
Conf. Power Electron. Drive Syst., 2005, pp. 246–250. in energy systems engineering from Uppsala Univer-
[18] K. Thorburn, H. Bernhoff, and M. Leijon, “Wave energy transmission sity, Uppsala, Sweden, in 2005, where he is currently
system concepts for linear generator arrays,” Ocean Eng., vol. 31, pp. working towards the Ph.D. degree within the field of
1339–49, 2004. wave power at the Division for Electricity, special-
[19] K. Thorburn and M. Leijon, “Farm size comparison with analytical izing in underwater power transmission systems for
model of linear generator wave energy converters,” Ocean Eng., vol. linear generator WEC arrays.
34, pp. 908–16, 2006.
[20] C. Boström, E. Lejerskog, M. Stålberg, K. Thorburn, and M. Leijon,
“Experimental results of rectification and filtration from an off-
shore wave energy system,” Renewable Energy, vol. 34, no. 5, pp.
1381–1387, 2009.
[21] S. Gustavsson, “Measuring of waves at islandsberg and literature
survey of wave measurement technology,” M.S. degree project, UPTE
F04048, Div. Electricity, Uppsala Univ., Uppsala, Sweden, 2004. Erland Strömstedt received the B.S. degree in
business administration and economics (BBA) from
Stockholm University, Stockholm, Sweden, in 2003
and the M.Sc. degree in materials technology, with
a specialization in light weight structures, from the
Royal Institute of Technology, Stockholm, Sweden,
in 2003. He is currently working towards the Ph.D.
degree at the Department of Electricity, Uppsala
University, Uppsala, Sweden.
He joined the Department of Electricity, Uppsala
Cecilia Boström (S’07) received the M.Sc. degree in University, in 2005, as a Research Assistant.
engineering physics from Uppsala University, Upp-
sala, Sweden, in 2006, where she is currently working
towards the Ph.D. degree at the Division for Elec-
tricity, and is involved in a wave energy converter Mats Leijon (M’83) received the Ph.D. degree in
project. electrical engineering from Chalmers University of
Technology, Gothenburg, Sweden, in 1987.
From 1993 to 2000, he was a Head of the
Department for High Voltage Electromagnetic
Systems, ABB Corporate Research, Västerås,
Sweden. In 2000, he became a Professor of Elec-
tricity at Uppsala University, Uppsala, Sweden.
Currently, he supervises 11 Ph.D. students within
wave power and marine current power and he has
Rafael Waters (S’06) received the M.Sc. degree in supervised ten students to a doctoral degree and
energy systems engineering and the Ph.D. degree in two students to a licentiate degree.
engineering physics from Uppsala University, Upp- Prof. Leijon received the Chalmers award, John Ericsson Medal, in 1984, the
sala, Sweden, in 2005 and 2008, respectively. Porjus International Hydropower Prize in 1998, the Royal University of Tech-
He is now working as a Researcher in the wave nology Grand Prize in 1998, the Finnish Academy of Science Walter Alstrom
energy project discussed in this paper. Prize in 1999, and the 2000 Chalmers Gustav Dahlen Medal. He also received
the Grand Energy Prize in Sweden and the Polhem Prize and the Thureus prize.
He is a Member of the Institution of Electrical Engineers (IEE), World Energy
Council (WEC), the International Council on Large Electric Systems-Cigre, and
the Swedish Royal Academy of Engineering Science.