Delft University of Technology

Active heave compensation of a floating crane using electric drive

Eijkhout, Thijs; Jovanova, Jovana

DOI
10.1109/AIM46487.2021.9517502
Publication date
2021
Document Version
Final published version
Published in
Proceedings of the IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2021

Citation (APA)
Eijkhout, T., & Jovanova, J. (2021). Active heave compensation of a floating crane using electric drive. In
Proceedings of the IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2021
(pp. 1089-1094). IEEE . https://doi.org/10.1109/AIM46487.2021.9517502

Important note
To cite this publication, please use the final published version (if applicable).
Please check the document version above.

Copyright
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent
of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Takedown policy
Please contact us and provide details if you believe this document breaches copyrights.
We will remove access to the work immediately and investigate your claim.

This work is downloaded from Delft University of Technology.

For technical reasons the number of authors shown on this cover page is limited to a maximum of 10.
Green Open Access added to TU Delft Institutional Repository

'You share, we take care!' - Taverne project

https://www.openaccess.nl/en/you-share-we-take-care

Otherwise as indicated in the copyright section: the publisher

is the copyright holder of this work and the author uses the
Dutch legislation to make this work public.
 2021 IEEE/ASME International Conference on
Advanced Intelligent Mechatronics (AIM)

Active heave compensation of a floating crane using

electric drive
2021 IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM) | 978-1-6654-4139-1/21/$31.00 ©2021 IEEE | DOI: 10.1109/AIM46487.2021.9517502

Thijs Eijkhout Jovana Jovanova

Maritime and Transport Technology Maritime and Transport Technology
Delft University of Technology Delft University of Technology
Delft, Netherlands Delft, Netherlands
[email protected] [email protected]

Abstract—Floating cranes have shown their benefits in port

operations and in offshore tasks. With the increase of offshore
wind farms, installation and maintenance of the wind turbines
require crane vessels for multiple tasks. In order to operate in
open sea regardless the ocean waves, active heave compensation
system is necessary to be installed. In this work an electric drive
is proposed for active heave compensation of a floating crane, as
more sustainable solution among other benefits. Dynamic model
of a floating crane was derived for model based control. Different
control algorithms were developed and numerically tested for
efficiency.
Index Terms—floating crane, active heave compensation, dy-
namic modeling, controller design

I. I NTRODUCTION
Nowadays there is an increased activity in ocean and
offshore engineering. The capacity of offshore wind farms has
increased from 4.117 MW in 2011 to 18.814 MW in 2017 Fig. 1: Offshore crane vessel
[1], and the trend is to install more and bigger wind turbines.
Installing submarine power cables require quite an amount of
resources [2] and sometimes remote-operated vehicles are used instead of hydraulic may be appealing for consumers [6].
for installation. The wind farms rely on regular maintenance to However electric actuators used for PHC and AHC is a field
maintain their high efficiency. The operation and maintenance that is not significantly explored yet [7].
costs are estimated to be 14%-30% of the total life-cycle costs The design of a system for AHC relies on the choice of
of wind farms [3] [4]. For all these offshore and ocean engi- actuation, in this case an electric drive, and a control strategy.
neering operations and their maintenance, there is increased Two commonly applied controllers are the well established
demand for vessels to provide the technicians, components, PID controller and LQR and the more advanced model pre-
spare parts and other resources. The variable costs of a crane dictive controller (MPC) [8]. The main advantages of MPC
vessel can be up to 40.000 USD/day [5]. To increase the over PID are the abilities to handle processes with large time
workability and efficiency of offshore crane vessels, it is delays, non-minimum phase and unstable processes. Also, PID
necessary to design these systems to handle weather conditions controllers tend to have a large phase lag which might be
accordingly [17]. Ocean waves are impeding the handling of another reason for preferring a MPC. Furthermore a MPC
payloads when offshore crane vessels are operating in open is easy to tune and it excels in handling constraints with
sea. When the payload is held, lowered or lifted at sea, active respect to a PID controller. Finally, MPC is also able to handle
heave compensation (AHC) is necessary to disconnect the systematic changes such as the failing of sensors and actuators.
heave motion the vessel endures with respect to the payload. The combination of using an electric drive in combination with
This results in the decoupling of the payload motion from a MPC is a novel approach in AHC [7]. In this work AHC
the vessel motion. Currently, there are companies trying to system is explored by analysing the dynamical system and
find alternatives for hydraulic actuators used for passive heave setting up a control design strategy. Dynamic model had been
compensation (PHC) and AHC. Instead of using hydraulic derived and used for controller design. Thereafter, numerical
systems, it may be possible to use electric drives. The reasons simulation results are presented and discussed using Matlab as
for using electric drives are mainly the lack of an oil reservoir modeling platform. This work explores the benefits of AHC
(which can also cause leaks) and lower motor noise. With the based on electric drive and traditional controllers like PID and
increased attention to sustainability, choosing electric actuators LQR, as well as model based MPC.

978-1-6654-4139-1/21/$31.00 ©2021 IEEE 1089

Authorized licensed use limited to: TU Delft Library. Downloaded on September 24,2021 at 11:43:27 UTC from IEEE Xplore. Restrictions apply.
 II. DYNAMIC MODELING

The design of an AHC system began with a dynamic

model derivation of a floating crane. The system shown in
Fig. 2a consists of a shipboard crane that is driven by an
AC induction motor in combination with a gearbox and a
drum. The drum holds the steel cable which is loaded by Fig. 3: Electro-mechanical system
the payload. When operating at sea, waves are inevitable and
have to be accounted for. The handling of the payload is
important in these circumstances as waves can be dangerous A. Electro-mechanical system
for maintaining a predictable motion of the payload. Hence,
the control objective is to decouple the upward wave motion The mechanical system contains an AC induction motor
of the ship with respect to its payload by using AHC. Fig. which is connected to the drum through a reduction gearbox.
2b shows how the payload is held at a constant position The steel cable around the drum is suspended by the payload
since the drum, together with a gearbox and the AC induction using also a sheave which is found at the tip of the crane.
motor, compensates for the heave. As the vessel sustains an The electro-mechanical system is shown in Fig. 3 where θm
upward heave motion, the drum unwinds to retain the payload is the motor angle, Jm the motor shaft’s mass moment of
at a constant position. On the other hand, when the vessel inertia. Furthermore, Jd is the mass moment of inertia of the
makes a downward have motion, the drum winds up in order drum and θd the drum angle and i is the gear ratio of the
to maintain the payload at the same position. With the aim gearbox. Last but not least, the motor torque is defined as Tm
to reach the control objective, the following assumptions are and whereas the effective torque is Tef f . The motor torque of
made: the AC induction motor is defined by (1).

• The crane structure is a rigid body. p Eg R1

Tm = R
(1)
• The steel cable will be considered as a mass-spring- 2ωr ( s ) + (ωr (L1 + L2 )) s
2 2 2

damper system.
• The mass of the steel cable is neglected. Where p is the number of poles, ωr the angular velocity of the
• The mass and the friction of the sheave is neglected. rotor, Eg the rated phase voltage and s the slip. Furthermore,
R1 is the rotor resistance, R2 the stator resistance whereas
Fig. 2c summarises these assumptions where kc is the spring L1 and L2 are the stator inductance and the rotor inductance,
constant, dc the damping coefficient, Ac the surface area and respectively. In order to control the speed and the torque of
E the Young’s modulus of the steel cable. Moreover, rd and θ̈d the AC induction motor, a variable-frequency drive (VFD)
are the radius of the drum and the angular acceleration of the is connected to the AC induction motor. The motor input
drum, respectively. Furthermore, ẅ is the wave acceleration frequency and voltage could be varied to control the latter.
and g the gravitational constant. The mass of the payload mp The efficient mass moment of inertia Jef f , seen from the AC
has its position measured by using parameter z whereas z0 is induction motor, is presented by (2).
the length of the cable when loaded.
Jd mp · rd2 Tef f
Jef f = Jm + + with i = (2)
i2 i2 Tm
For obtaining the differential equation for the system, La-
grange’s method is applied as shown in (6) where T is
the kinetic energy, R the energy dissipation, U the potential
energy, qj the independent coordinate of the system its motion
and Gj the corresponding load in the independent coordinate.
(a) The independent coordinate qj can be substituted by z. The
kinetic energy of the system is shown in (3).
1
T = mp ż 2 (3)
2
Moreover, the energy dissipation is presented in (4).
1
R= dc ż 2 (4)
2
(b) (c)
Also, the potential energy is given by (5).
Fig. 2: (a) 2D schematic of the system, (b) Active heave
compensation, (c) Kinematic model. 1
U= kc z 2 (5)
2
1090

Authorized licensed use limited to: TU Delft Library. Downloaded on September 24,2021 at 11:43:27 UTC from IEEE Xplore. Restrictions apply.
The three previous mentioned equations are now defined for wave prediction model is the unmeasured disturbance that
(6).   acts on the mechanical system of the offshore crane vessel.
d δT δT δR δU By utilizing MATLAB, the system is tested and is proven to
− + + = Gj (6)
dt δ q̇j δqj δ q̇j δqj be stable, controllable and observable. Then a discretization
Rewriting (6) results in the second order differential equation, process is performed in order to receive the discrete state space
seen in (7). of the system as shown in (10).
EAc Tef f ~x˙ (k + 1) = Ad ~x(k) + B1 ~u(k) + B2 d(k)
~
mp z̈ + dc ż + z = mp (ẅ + rd ) (7) (10)
L0 Jef f ~y (k) = C~x(k)
The aforementioned equation shows the motion of the cable For obtaining the optimal input control sequence at a given
suspended payload. The vibrations of the load are compensated time step k, a cost function VN is derived in the form of (11).
by wave accelerations ẅ and the effective torque of the drum
Tef f . L0 represents the steel cable in unloaded state. min VN (x0 , uN ) =
uN
III. C ONTROL STRATEGY DESIGN  
−1
The differential equation for the load vibrations can be NX 1 1
{ x(k)T Qx(k) + u(k)T Ru(k)

min 
rewritten in the form of a state space shown in (8). 2 {z 2
uN  
k=0 | }
~x˙ = A~x + B~u l(x(k), u(k)) =stage cost
(8)  
~y = C~x + D~u
 1 
+ min  x(N )T P x(N )} (11)
T
 
with the states ~x = [ż z θ̇ θ] . The wave acceleration will uN  2


be considered as an unmeasured disturbance in this work. | {z }
Vf (x(N )) =terminal cost
However, it is possible to predict waves on the sea 2.5 hours
in advance [9] and use a prediction model that will act as a Where both Q and P are positive definite weight matrices
measured disturbance. whereas R is a semi positive definite weight matrix. The
 dr   rd Tef f  Q and the R matrix are chosen, then the P matrix is the
− mp ż − mEA r
p L0
z 0 0 Jef f + ẅ
algebraic solution of the Ricatti equation [10]. Also, uN is
 ż 0 0 0 0
 
~x˙ =  + the optimization variable whereas N is the horizon length.

 0 T
0 0 0  ef f
 
Jef f

(9) The cost function, which is minimised, is subject to certain
0 0 θ̇ 0 0 constraints shown in (12), where Nc is the control and Np the
 
0 z 0 0 prediction horizon.
~y =
0 0 0 θ
Tef f,min ≤Tef f (k) ≤ Tef f,max ∀k = 0, 1, ..., Nc − 1
As shown in the C matrix, the position of the payload is ẅmin ≤ẅ(k) ≤ ẅmax ∀k = 0, 1, ..., Nc − 1
measured and the motor angle is measured. The position of (12)
zmin ≤z(k) ≤ zmax ∀k = 1, 2, ..., Np
the payload is measured with a motion reference unit (MRU)
and the drum angle is measured with a rotary encoder. It is θmin ≤θ(k) ≤ θmax ∀k = 1, 2, ..., Np
also possible to use an inertial measurement unit (IMU) for IV. N UMERICAL SIMULATION
measuring the position of the payload. The block diagram of
the system, presented in (9), is presented in Fig. 4. The numerical simulations are carried out in Matlab where
the system is analysed with ocean wave disturbances and
different control strategies. The list of parameters that are kept
constant for every simulation are shown in Table I. An analysis
is made between MPC, LQR-controller, PID-controller and
open-loop. In total five simulations are carried out. Table II
shows the parameters that are tuned during the simulations,
where KP is the proportional term, KI the integral term and
KD the derivative term of the PID-controller. The system
has to compensate for the wave acceleration ẅ(t). In order
Fig. 4: Block diagram of the system to operate the offshore crane vessel in a safe way, it is
recommended that the wind speed should not exceed above 9.0
It shows that the angle of the drum and the position of the m/s (32.4 km/h or 17.49 knots) [11]. Wind speeds of around
payload are processed in the reference model. The difference 9.0 m/s result in sea waves with an approximate amplitude
between the setpoint and the sensor measurements result in a of 1.5m and a period of about 5s [12]. The function that is
certain error. The model predictive controller will stimulate used as the disturbance is shown in 13 after taking the second
actuator input in order to compensate for the error. The derivative of w(t).

1091

Authorized licensed use limited to: TU Delft Library. Downloaded on September 24,2021 at 11:43:27 UTC from IEEE Xplore. Restrictions apply.
 TABLE I: Parameters

Parameter Value Unit

Jef f 4.0 · 10−3 [kg · m2 ]
mp 1.0 · 10−3 [kg]
E 8.5 · 1010 [P a]
Ac 2.0 · 10−4 [m2 ]
L0 1.0 · 102 [m]
dc 1.8 · 103 [kg/s]
Nc 5 [−]
ts 0.5 [s]

 2  
2π 2π
ẅ(t) = − A sin ts (13) Fig. 6: Performance of the four controllers PID, LQR, open-
T T loop seen from a narrowed interval

TABLE II: Tuned parameters TABLE III: Motion decoupling and settling time of the controllers in the first simulation
Simulation Q R KP KI KD Np Nc Motion decoupling [%] Settling time [s]
1 100 5 35 35 35 20 5 LQR-controller 33.3 10.49
2 200 5 35 10 10 15 5 PID-controller 97.0 11.75
3 400 40 10 35 10 10 5 MPC 92.3 12.32
4 1000 100 10 10 35 7 5 Open-loop control 0 11.34
5 1000 200 10 10 10 5 5

The first simulation, presented in Fig. 5 and Fig. 6, shows The MPC and the PID-controller perform outstanding com-
that all controllers are stable. During the time the wave accel- pared to open-loop control and the LQR-controller when
eration is active the PID-controller is outperforming both the looking at motion decoupling. However, the settling time of the
LQR-controller and the open-loop controller. Even the MPC is LQR-controller is outperforming the other three controllers.
outperformed by the PID-controller. However, when the wave For constant waves at sea, motion decoupling is the decisive
accelerations stop, the settling time tset becomes an important factor for choosing the right controller. Table IV shows all
characteristic of the controller. For the simulation the settling the results of the five simulations. The MPC has a very
time has to be within the ±0, 01m steady state band. The consistent motion decoupling (MD) even when the prediction
LQR-controller shows it has the fastest settling time compared horizon and the weight matrices Q and R are varied. However,
to the other controllers although it has more overshoot and when the wave accelerations stop, the settling time of the
more oscillations then the PID-controller and the MPC. Table MPC depends on the ratio between Q and R. The bigger
III summarises the motion decoupling percentages and the this ratio is, the larger the settling time of the MPC is.
settling time of the controllers. The motion decoupling is Also note that the ratio between Q and R is the same in
calculated by (14). both simulation 4 and simulation 5. This has a minimum
effect on the motion decoupling. Considering the results of
P eakOpenLoop − P eakController the MPC, a final simulation will be done by halving the
· 100% (14) sampling time ts = 0.25 instead of ts = 0.50 and utilising the
P eakOpenLoop
same parameters in simulation 4. The PID-controller performs

TABLE IV: Motion decoupling and settling time of the controllers in the first simulation

Simulation
Parameter 1 2 3 4 5
Q 100 200 400 1000 1000
R 5 5 40 100 200
KP 35 35 10 10 10
KI 35 10 35 10 10
KD 35 10 10 35 10
Np 20 15 10 7 5
M DM P C [%] 92.3 92.3 92.6 92.7 92.5
M DP ID [%] 97.0 95.7 94.3 95.4 90.2
M DLQR [%] 33.3 58.8 23.5 30.2 2.1
tset,M P C [s] 12.32 12.02 10.49 10.49 10.48
Fig. 5: Performance of the four controllers PID, LQR, open- tset,P ID [s] 11.75 9.82 11.18 16.17 13.90
loop tset,LQR [s] 10.49 10.28 10.64 10.50 10.74

1092

Authorized licensed use limited to: TU Delft Library. Downloaded on September 24,2021 at 11:43:27 UTC from IEEE Xplore. Restrictions apply.
 Fig. 7: Performance of of the three controllers in the final comparison

very well considering the motion decoupling percentages. horizon. Furthermore, with a control horizon of Nc = 5,
The first simulation is executed by setting all three PID- the computational effort is reasonably low. Also increasing
controller terms on 35. The second simulation is done by the control horizon results in a more aggressive controller.
holding KP on 35 and the other two terms on 10. One In this scenario the PID-controller performs better. However,
can see that the motion decoupling decreases in the second the PID-controller has no knowledge of constraints. Also it
simulation with respect to the first simulation. However, the does not have the ability to coop with constraints. Due to the
settling time of the second simulation is higher than of the fact MPC has the ability to anticipate in the future, it is an
first simulation. Hence, for the PID-controller it is a trade-off excellent choice when the waves are predicted in advance.
between motion decoupling percentage and the settling time Replacing hydraulic actuation by electric actuation will be
when picking the PID-controller terms. The LQR-controller is a challenge for the future. The power density of hydraulic
not performing very well considering the motion decoupling. actuation systems is higher than AC drives [7]. For example,
The LQR-controller performs the best when the ratio between [13] shows that only hydraulic actuation provides the required
Q and R is very high. A large gain is required for the LQR- power density. The final MPC with ts = 0.25 provides
controller to compete with the MPC and the PID-controller. maximum load position of 0.091m which is comparable and
The settling however is consistent. Now a final comparison sometimes performing better than [14] and [15] in which
is made between the PID-controller of the first simulation hydraulic actuation is used. Considering all the results and
and the MPC of the fourth simulation as they have the best literature, the combination of MPC and AC induction motors
performance looking at motion decoupling percentage. A third is promising. The motion decoupling is consistent and may
MPC (also with parameters used of the fourth simulation) achieve even higher values than approximately 94%. However,
is also added but its sampling time is not 0.50s but 0.25s. the main downside of AC drives is the power density. The
Fig. 7 shows the results of the three controllers. The motion power density is defined as a power-to-weight ratio of the
decoupling of the MPC has improved significantly when the actuator. Hydraulic actuation is unmatched for heavy lifting.
sampling time is halved. With a smaller sampling time of Furthermore, [16] shows that the hydraulic actuators provide
0.25s, the MPC is able to optimise twice as many intervals a higher power density than AC drives. AHC may be achieved
than when using a sampling time of 0.50. However, the using a VFD connected to the AC induction motor. Also, AC
computational time increases. With a prediction horizon of drives should first be considered in offshore support ships for
Np = 7, the MPC is relatively aggressive. This is necessary bringing supplies and utilising maintenance. When AC drives
as for incoming waves, control action have to be taken. On the are used in combination with wide diameter winches, a large
other hand, if the wave frequency is low, one may opt for a torque is required in order to adjust the motor speed. For
more robust controller and therefore increase the prediction example, In a case of a winch with a high inertia, the motor

1093

Authorized licensed use limited to: TU Delft Library. Downloaded on September 24,2021 at 11:43:27 UTC from IEEE Xplore. Restrictions apply.
torque might not be able to surmount the load torque enough in [7] J. K. Woodacre, R. J. Bauer, and R. A. Irani, “A review of vertical
order to achieve an appropriate acceleration. Design of floating motion heave compensation systems,” In Ocean Engineering 104, 2015,
pp. 140–154.
structures for different purposes (living, aqua-farming, energy, [8] A. Maxim, D. Copot, C. Copot, C. M. Ionescu, “The 5W’s for Control
etc.) has raised a lot of attention in the last years. This comes as Part of Industry 4.0: Why, What, Where, Who, and When—A PID
from the need of more space on land, which is not always and MPC Control Perspective,” In Inventions 4.1, 2019, p. 10.
[9] P. Naaijen, ”Deterministic prediction of waves and wave induced vessel
possible, and sustainability in utilising the open sea resources. motions: Future telling by using nautical radar as a remote wave sensor.”
Floating terminals [17] can be beneficial in different scenarios, 2017.
commercially and for disaster relief. All these concepts would [10] S. Bittanti, A. J. Laub, and J. C. Willems, ”The Riccati Equation.”
Springer Science and Business Media, 2012.
benefit from sustainable AHC system based on electric drive [11] I. E. L. P. Troen and E. L. Petersen, ”European wind atlas.” 1989.
and model based control. [12] M. S. Longuet-Higgins ”On the joint distribution of the periods and
amplitudes of sea waves.” Journal of Geophysical Research 80.18, 1975,
V. C ONCLUSION pp. 2688-2694.
[13] P. Gu, A. A. Walid, Y. Iskandarani and H. R. Karimi, “Modeling,
This work explored the combination of MPC and AC simulation and design optimization of a hoisting rig active heave
compensation system”. In: International Journal of Machine Learning
induction motors for AHC of floating crane. The initial results and Cybernetics 4.2, 2013, pp. 85–98.
showed that the feasibility, potential and benefits of such [14] J. Neupert, T. Mahl, B. Haessig, O. Sawodny and K. Schneider, “A
systems. Even though the main downside of AC drives is heave compensation approach for offshore cranes”. In: 2008 American
Control Conference. IEEE, 2008, pp. 538–543.
the power density, other benefits coming from sustainability [15] M. B. Kjelland and M. R. Hansen, “Offshore wind payload transfer
trends and energy transition might encourage the maritime using flexible mobile crane”. In: MIC Journal, Vol 36, No 1, 2015, pp.
industry to look into this solution. AHC could be accomplished 1-9.
[16] A. Nespoli, S.Besseghini, S. Pittaccio E. Villa and S.Viscuso, “The high
with a variable-frequency drive connected to the AC induction potential of shape memory alloys in developing miniature mechanical
motor. For further investigation it is also recommended looking devices: A review on shape memory alloy mini-actuators”. In: Sensors
into multiple, smaller AC drives which work together in and Actuators A: Physical 158.1 2010, pp. 149–160.
[17] J. Jovanova, W. van den Bos and D. L. Schott, ”Design of floating
order to achieve AHC. However, the main problem stays terminals as integrated project for multi-machine systems”. In: L.
that AC drives take a lot of space. Large AC drives might Piatek, R. de Graaf-van Dinther, L. S. Heng, and C. M. Wang (Eds.),
also result in distortions of the power grid if the drive uses Proceedings of the 2nd World Conference on Floating Solutions 2020,
WCFS 2020: Paving the Waves, pp. 340-352. Blue21.
too much power than the grid can handle. One should look
into realistic offshore vessel designs which meet the needs of
the current market. For example, the offshore wind turbines
are becoming larger and therefore bigger offshore (support)
vessels are needed. This also results in the need of actuators
which provide enough power for maintaining these structures.
There is however an optimal combination of offshore crane
vessels with a corresponding AC drive. This is an aspect one
should look into more extensively. There is also unknown
territory when using accurate wave predictions in combination
with MPC and AC drives. Further research is required to obtain
more in-depth knowledge of this subject. On top of this, a
physical model should be made in combination with a towing
tank. Both regular and irregular waves can be used as input
in order to achieve a variety of results.

R EFERENCES
[1] M. A. Majid, ”Wind energy programme in India: Emerging energy
alternatives for sustainable growth.” Energy and Environment 30.7, 2019.
[2] T. Worzyk, ”Submarine power cables: design, installation, repair, envi-
ronmental aspects,” Springer Science and Business Media, 2009.
[3] W. Musial and B. Ram, ”Large-scale offshore wind power in the United
States: Assessment of opportunities and barriers,” Tech. rep. National
Renewable Energy Lab.(NREL), Golden, CO (United States), 2010.
[4] B. Maples, G. Saur, M. Hand, R. Van De Pietermen, and T. Ob-
dam, ”Installation, operation, and maintenance strategies to reduce the
cost of offshore wind energy,” Tech. rep. National Renewable Energy
Lab.(NREL), Golden, CO (United States), 2013.
[5] E. E. Halvorsen-Weare, C.Gundegjerde, I. B. Halvorsen, L. Magnus
Hvattum, L. Magne Nonås, “Vessel fleet analysis for maintenance
operations at offshore wind farms,” In Energy Procedia 35, 2013, pp.
167–176.
[6] V. D. Angelis, “Comparison study of electric, electro-hydraulic, and
hydraulic drive science winches,” In 11th European Research Vessel
Operators Meeting. Vol. 2009, 2009.

1094

Authorized licensed use limited to: TU Delft Library. Downloaded on September 24,2021 at 11:43:27 UTC from IEEE Xplore. Restrictions apply.