Applied Ocean Research 57 (2016) 8–18

Contents lists available at ScienceDirect

Applied Ocean Research

journal homepage: www.elsevier.com/locate/apor

Model test and simulation of a ship with wavefoils

Eirik Bøckmann ∗ , Sverre Steen
Department of Marine Technology, Norwegian University of Science and Technology, Norway

a r t i c l e i n f o a b s t r a c t

Article history: The paper describes model tests of a tanker with two fixed bow-mounted foils (wavefoils), for resis-
Received 16 September 2015 tance and motion reduction in waves. Measured ship resistance, wavefoil thrust and ship motions were
Received in revised form compared with time-domain simulations. The wavefoil forces were calculated using a slightly modi-
17 December 2015
fied Leishman–Beddoes dynamic stall model, with a two-way coupling to the ship motions. In typical
Accepted 5 February 2016
sea states in head seas, employing the wavefoils reduced ship resistance by 9–17%, according to scaled
Available online 4 March 2016
model test resistance. Heave and pitch were reduced by −11% to 32% and 11% to 25%, respectively. The
foils affect the flow around the hull. This should be considered when selecting the wavefoil location in
Keywords:
Ship resistance
the design process.
Foil © 2016 Elsevier Ltd. All rights reserved.
Model test
Simulation

1. Introduction wave-powered boats, both experimental and theoretical, is given

To achieve stabilization of the global temperature at two degrees
above pre-industrial levels, global anthropogenic greenhouse gas
emissions must be reduced by 40–70% by 2050, compared to 2010
[1]. If all sectors accept the same percentage reductions and global
maritime trade doubles from 2010 to 2050, which is in accordance
with conservative predictions [2], greenhouse gas emissions from
shipping per tonne nautical mile must be reduced by 70–85% by
2050, compared to 2010. Faced with this substantial challenge, har-
nessing the non-polluting and free resource of wind and waves for
ship propulsion seems unavoidable.
Using waves to propel a boat was, to the authors’ knowledge,
first proposed in 1858 [3] and first successfully done in practice
in the 1890s [4]. The simplest and most common method of wave
propulsion is outfitting the ship with foils, so-called wavefoils, that
convert the vertical motion in waves into propulsive thrust. In
addition to producing thrust, the foils generally dampen the ship
motions, causing reduced added resistance in waves. Jakobsen [5]
introduced a foil with spring-loaded pitch to avoid stall. Active
pitch control has also been proposed [6,7] and tested in practice
[8,9]. In the present work, however, the foil pitch was fixed at zero
to validate the simulation model for the simplest wavefoil possi-
ble. A more detailed summary of previous work on partly and fully
∗ Corresponding author. Tel.: +47 73598893.
E-mail addresses: eirik.bockmann@ntnu.no (E. Bøckmann),
sverre.steen@ntnu.no (S. Steen).
by Bøckmann [9].
In theory, the motions and resistance in waves of a ship with
foils as appendages can be found by using a Navier–Stokes solver,
but this is still computationally very expensive. A faster approach is
to model the ship with foils in a panel code. Panel codes neglect vis-
cous effects, which is not applicable when flow separation is likely
to occur on the foils. A simple and common approach is to obtain the
hydrodynamic force coefficients for the bare hull using a computer
program based on 3D panels with a distribution of potential flow
singularities [10] or strip theory [11], add the separately calculated
foil forces, and solve the equations of motions (see e.g. [6,7]).
The equations of motions can be solved both in the time domain
and in the frequency domain. When adding foil forces to the
equations of motion and solving these equations in the frequency
domain, one must assume that the foil forces oscillate with the
frequency of encounter and are linearly proportional to the wave
amplitude. This implies that one must neglect drag and stall and
assume that the foil lift acts vertically relative to the ship in its
calm-water position, which rules out studying cases where the foil
has a large angle of attack. In time-domain simulations there is no
inherent limit to the angle of attack since the foil forces can vary
arbitrarily with time.
Solving the equations of motions in the time domain requires
the calculation of retardation functions which account for memory-
effects – i.e., the ship motions at a given time depend on the
ship motion at previous time steps. Calculating these retardation
functions from frequency-domain coefficients is not numerically
straight forward although the expression is simple (see Eq. (13)).
The ship seakeeping and manoeuvring simulator VeSim [12],

http://dx.doi.org/10.1016/j.apor.2016.02.002
0141-1187/© 2016 Elsevier Ltd. All rights reserved.
 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18 9

developed by MARINTEK and previously used by the first author

to study a ship with wavefoils [9], is an example of a computer
program where the equations of motion are solved in the time
domain. Before the ship motions are found by stepping the solu-
tion forward in time in VeSim, retardation functions are calculated
from frequency-domain hydrodynamic coefficients obtained from
the strip-theory program VERES [13].
In the present paper, VeSim was used to simulate a ship
with wavefoils, and simulation results are compared with model
test results. A slightly modified version of the Leishman–Beddoes
dynamic stall model [14] was used to calculate the foil forces with
a two-way coupling between the ship motions and foil forces.

2. Ship and wavefoils

Experiments were conducted with a 1:16.57 scale model of

an 8000 DWT tanker known as R&D 8000, designed by Rolls- Fig. 2. Front view of the ship. Illustration courtesy of Rolls-Royce plc.

Royce Marine. This ship was model-tested at the MARINTEK towing

tank in Trondheim, Norway, in 2010 with three different foreships
and three different aftships. In the experiments and simulations
described in this paper, the foreship (MARINTEK model M2946A)
had a bulbous bow, vertical stem with very little flare (see Fig. 1),
and the aftship (MARINTEK model 2944A) had a twin skeg design
with Rolls-Royce Promas rudders and a dummy propeller boss (see
Fig. 5). The main particulars of the ship and wavefoils in full and
model scales are given in Tables 1 and 2, respectively.
The wavefoils were mounted to the ship with a fixed roll angle of
10◦ (see Fig. 2) and a fixed pitch angle of 0◦ . The horizontal distance
from the fore perpendicular to the trailing edges of the foils was
16.926 m in full scale.
3. Model test
3.1. Test set-up
The present combination of fore- and aftship was tested in the
seakeeping carriage of the MARINTEK towing tank in March 2015.
The ship model was towed from wires attached to an aluminium
beam mounted perpendicular to the ship, allowing the ship to move
freely in heave and pitch (see Fig. 3). A spring system allowed the
ship to move relatively freely in surge. The towing force was mea-
sured at both port and starboard side of the towing boom and
summed to yield the total towing force. The ship motions were
measured by an optical motion capture system from Qualisys.
The wavefoils went through foil-shaped openings in the hull,
just larger than the foils, and were then mounted to a vertical circu-
lar beam. This circular beam went through a circular pipe, creating
a circular partially filled moonpool in the hull. Atop the beam car-
rying the wavefoils were mounted three vertical force transducers
(one in front of the other two), and a horizontal force transducer

Fig. 1. Side view of the bow with port wavefoil shown.

Table 1
Main particulars of the Rolls-Royce R&D 8000 tanker in full and model (1:16.57)
scales.

Main particulars Full scale Model scale

Waterline length (LWL ) 117.297 m 7.079 m

Length between perpendiculars (LPP ) 113.200 m 6.832 m
Breadth at waterline 19.000 m 1.147 m
Draft 7.200 m 0.435 m
Wetted surface area 3387.816 m2 12.339 m2
Displaced volume 11,616.770 m3 2.553 m3

Table 2
Main particulars of the wavefoils in full and model (1:16.57) scales.

Main particulars Full scale Model scale

Profile NACA 0015 NACA 0015

Span (tip to root) 10.8 m 0.652 m
Inner chord 3.2 m 0.193 m
Outer chord 2.2 m 0.133 m
Foil area 29.2 m2 0.106 m2
Planform Tapered Tapered
Fig. 3. Model test set-up.
10 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18

Fig. 4. Connection between wavefoils and ship model.

Fig. 5. Aftship with aluminium skeg.

was mounted atop the vertical force transducers (see Fig. 4). The
whole wavefoil-carrying rig was connected to the ship via a steel 3.3.2. Standard ship resistance scaling
beam mounted perpendicular to the centreline of the ship model. The total resistance coefficient in model scale, CTm , is given as

3.2. Test conditions CTm = CFm (1 + k) + CR , (1)

where CFm is the model-scale frictional resistance coefficient, k is

The specified and actual wave conditions in the model test are
the form factor, and CR the residual resistance coefficient, due to
given in Tables 3 and 4. In the simulations in Section 5, the actual
wave making. Subscript m implies model-scale values. The fric-
wave conditions were used.
tional resistance coefficient is given as

3.3. Scaling of model test results 0.075

CF = , (2)
(log(Re) − 2)2
3.3.1. Subtracting aluminium skeg resistance
All the model scale results were obtained with an aluminium where Re is the Reynolds number. Since the Reynolds number is
plate as an extra skeg mounted to the transom of the model (see larger in full scale than in model scale, CF is smaller in full scale
Fig. 5). After the tests with wavefoils were performed, the alu- than in model scale. The form factor for the ship model-tested was
minium skeg was found to be unnecessary for the ship to sail calculated using MARINTEK’s formula for the form factor,
straight with wavefoils, although a small yaw angle was observed
k = 0.6 + 753 , (3)
at 15 knots full-scale speed in calm water. Therefore, the resistance
of the skeg was subtracted from the model scale resistance before where
CB 
scaling the resistance to full scale, for the model with and with-
out wavefoils. The resistance of the aluminium skeg was found by = (TAP + TFP )B, (4)
LWL
assuming that the only resistance component is the frictional resis-
tance coefficient, see Section 3.3.2. In reality, the wavefoils will which gives k = 0.1139 for the ship model-tested. CB is the block
impede the ship’s manoeuvrability somewhat. This may be com- coefficient, LWL is the length at waterline, TAP is the draught at the
pensated by active use of the rudder, giving an unknown additional aft perpendicular, TFP is the draught at the fore perpendicular, and
resistance component, which is assumed to be small. B is the beam.

Table 3
Conditions in regular waves.

Set-up Specified speed Actual speed Specified wave Actual wave Specified wave Actual wave
[knots] [knots] height [m] height [m] period [s] period [s]

Wavefoils 11.00 10.95 3.00 2.91 7.00 7.01

Wavefoils 11.00 10.95 3.00 2.69 9.00 8.99
Wavefoils 11.00 10.95 3.00 3.00 11.00 11.06
Wavefoils 11.00 10.95 3.00 2.63 13.00 13.26
Wavefoils 13.00 12.94 3.00 2.96 7.00 7.04
Wavefoils 13.00 12.94 3.00 2.55 9.00 9.08
Wavefoils 13.00 12.94 3.00 3.03 11.00 11.01
Wavefoils 13.00 12.94 3.00 2.81 13.00 13.02
Bare hull 11.00 10.95 3.00 3.04 7.00 7.01
Bare hull 11.00 10.95 3.00 2.72 9.00 8.99
Bare hull 11.00 10.95 3.00 3.00 11.00 11.06
Bare hull 11.00 10.95 3.00 2.59 13.00 13.26
Bare hull 13.00 12.94 3.00 2.78 7.00 7.04
Bare hull 13.00 12.94 3.00 2.64 9.00 9.08
Bare hull 13.00 12.94 3.00 3.11 11.00 11.01
Bare hull 13.00 12.94 3.00 2.83 13.00 13.02
 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18 11

Table 4
Conditions in irregular waves based on measurements of approximately 30 min of irregular waves of the JONSWAP wave spectrum. The post-processed ship motions and
forces are based on three runs down the tank at each sea state, giving a total time series of 3 min and 30 s in model scale (14 min and 15 s in full scale) at each sea state.

Set-up Specified speed Actual speed Specified Actual significant Specified spectral Actual spectral
[knots] [knots] significant wave wave height [m] peak wave period peak wave period
height [m] [s] [s]

Wavefoils 13.00 12.94 3.00 2.70 7.00 7.17

Wavefoils 13.00 12.94 3.00 2.78 9.00 8.96
Wavefoils 13.00 12.94 3.00 2.95 11.00 11.02
Wavefoils 13.00 12.94 3.00 2.96 13.00 12.79
Bare hull 13.00 12.94 3.00 2.70 7.00 7.17
Bare hull 13.00 12.94 3.00 2.78 9.00 8.96
Bare hull 13.00 12.94 3.00 2.95 11.00 11.02
Bare hull 13.00 12.94 3.00 2.96 13.00 12.79

In full scale, there is an additional frictional resistance coefficient 1.5

due to roughness of the ship hull, CF , so that the total resistance Model scale, 11 knots
Model scale, 11 knots modified
coefficient in full scale, CTs , becomes Model scale, 13 knots
Full scale, 11 knots
CTs = (CFs + CF )(1 + k) + CR + CA , (5) Full scale, 13 knots
1
where CA is the correlation coefficient, which is an empirical coeffi-
cient varying from towing tank to towing tank. Subscript s implies

foil
T
C
full-scale values. The submerged transom area was very small with-
out forward speed and was assumed to be zero for the forward 0.5
speeds studied here. CF is given as

CF = [110.31 · (H · Vs )0.21 − 403.33] · CFs

2
, (6)

where H is the hull surface roughness in ␮m (10−3 mm), and Vs 0

is the ship speed in m/s. H = 100 ␮m was used here, and CF was 0 10 20 30 40
taken to be zero if Eq. (6) gave CF < 0. Based on MARINTEK data, α [deg]
CA =−0.2 · 10−3 .
Fig. 6. CTfoil vs. ˛ for 11 and 13 knots.

3.3.3. Reynolds-scaling the foil thrust

For completely attached flow, implying low angles of attack,
0.8
the lift coefficient of a foil is independent of the Reynolds num-
ber. For higher angles of attack, there may be separated flow for a 0.7
low Reynolds number and attached flow for a high Reynolds num-
ber. A higher Reynolds number also implies a lower drag coefficient. 0.6
Both of these effects will increase the foil thrust relative to a simply
Froude-scaled foil thrust. 0.5
In dynamic conditions, the lift coefficient can be much higher
0.4
Tfoil

than in static conditions. To be able to quantify the Reynolds scale

Model scale, 5 knots
C

effect in a simple manner, we considered the Reynolds scale effect 0.3 Model scale, 5 knots modified
on a two-dimensional foil, using a simplified approach. The lift coef- Model scale, 7 knots
ficient was assumed to be linear, whereas the drag coefficient was 0.2 Model scale, 9 knots
Model scale, 9 knots modified
taken to be the quasi-steady value, which is dependent on angle of
Model scale, 15 knots
attack and Reynolds number. We then get the following expression 0.1
Full scale, 5 knots
for the foil thrust coefficient, CTfoil : Full scale, 7 knots
0 Full scale, 9 knots
Full scale, 15 knots
CTfoil (˛, Re) = CL˛ ˛ sin ˛ − CD (˛, Re) cos ˛. (7)
−0.1
0 5 10 15 20 25 30
CL˛ is the lift coefficient slope and taken to be 6.3025 for a NACA α [deg]
0015 airfoil [15]. CD (˛, Re) is also taken from Sheldahl and Klimas
[15]. Fig. 6 shows the thrust coefficient, CTfoil , of a NACA 0015 foil Fig. 7. CTfoil vs. ˛ for 5, 7, 9, and 15 knots.

as a function of angle of attack for the full-scale speeds 11 and 13

knots, in model and full scales. There is not a one-to-one relation
between CTfoil,m and angle of attack for a full-scale speed of 11 knots
and angles of attack between 13◦ and 16◦ , which makes it trickier
to relate CTfoil,m to the angle of attack. To overcome this, the CTfoil,m
graph for 11 knots was modified in this angle of attack range. Mod-
ifying the CTfoil,m graph in this manner is conservative with respect
to the foil thrust scale factor, which is defined as CTfoil,s /CTfoil,m . Fig. 7
shows CTfoil vs. ˛ for full-scale speeds of 5, 7, 9, and 15 knots, with
similarly modified CTfoil,m graphs for full-scale speeds of 5 and 9
knots.
With the modified CTfoil,m graph for 11 knots, there is a one-to-
one relation between CTfoil,m and the angle of attack and hence also
between CTfoil,m and the foil thrust scale factor, as shown in Fig. 8.
Similarly, Fig. 9 shows the foil thrust scale factor vs. CTfoil,m for full-
scale speeds of 5, 7, 9, and 15 knots. If CTfoil,m from the model test
was lower than the lowest CTfoil,m in Figs. 6–9, the foil thrust scale
factor was set to be 1.
When post-processing the model test data, the foil thrust coeffi-
cient in model scale was multiplied by the corresponding foil thrust
12 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18

2 calculating the precision limit of a parameter, for example the foil

11 knots thrust, PTfoil , as
11 knots modified
13 knots
1.5 PTfoil = tTfoil , (9)
Foil thrust scale factor

1 where t is the t-value in the Student’s t-distribution, and  is the

standard deviation for the repeated condition. The standard devi-
ation of the mean value for the repeated condition is calculated
0.5 as
Tfoil
T ¯ = √ , (10)
0 foil
N

where N is the number of repeated tests. The precision limit for the
−0.5 mean is then
0 0.5 1 1.5
CT
foil,m PT ¯ = tT ¯ . (11)
foil foil

Fig. 8. Foil thrust scale factor vs. CTfoil,m for 11 and 13 knots.
This gives a smaller error bar for the test condition that was
repeated.
6 The error bars for the repeated Vs = 11 knots tests were applied
5 knots for all tests with Vs = 11 knots, and similarly, the error bars for the
5 knots modified
5 repeated Vs = 13 knots tests were applied for all tests with Vs = 13
7 knots
9 knots knots, including the tests with irregular waves. Since only runs
Foil thrust scale factor

4 9 knots modified with waves and without wavefoils were repeated, no error bars
15 knots are calculated for the calm-water and foil-thrust graphs.
3

2 3.5. Other remarks

1 Since the wavefoils and supporting rig were suspended from

0 gravity influenced the force measurement when the ship was pitch-
−1
0 0.2 0.4
CT
foil,m
Fig. 9. Foil thrust scale factor vs. CTfoil,m for 5, 7, 9, and 15 knots.
scale factor in Figs. 8 and 9 to obtain the foil thrust coefficient in
full scale. The foil thrust in full scale, Tfoil,s , was then obtained as
1
Tfoil,s = CTfoil,s s Vs2 Sfoil ,
2
where s is the mass density of water in full scale (seawater), and
Sfoil is the foil planform area. The effect of the Reynolds-scaled foil
thrust on the ship resistance with wavefoils was taken care of by
adding the Froude-scaled foil thrust and subtracting the Reynolds-
scaled foil thrust to the ship resistance.
3.4. Uncertainty analysis
To quantify the uncertainty of the test results, some of the runs
were repeated. The ship without wavefoils in regular waves, with
a full-scale speed (Vs ) of 13 knots, full-scale wave height (H) of
approximately 3 m, and full-scale wave period (T) of approximately
13 s was tested six times. The same configuration but with Vs = 11
knots was tested five times. Error bars were produced based on
the repeated runs. The error bars for the heave and pitch response
amplitude operators (RAOs) and standard deviations at 13 knots
are based on only four repeated tests, however, due to too short
time series to perform a high quality Fourier analysis (see Section
3.5).
The error bars in the figures in Section 5 mark the 95% con-
fidence interval and were obtained by repeating these tests and
the force transducer measuring the total horizontal wavefoil force,
ing. This contribution to the force was removed from the foil-thrust
graphs. The effect of buoyancy changes due to changing water level
0.6 0.8
above the foil was accounted for by using the water elevation rel-
ative to the hull measured a few centimetres in front of the foil.
Since RAOs are linear by definition, Fourier analyses were per-
formed to find the heave and pitch amplitude used in the RAOs as
the dominating harmonic component of the measured time series
of heave and pitch.
4. Simulation model
(8)
4.1. Equations of motion
The text in this subsection is intended as a brief introduction to
the theory behind VeSim, the combined seakeeping and manoeu-
vring simulator used in the present work, and is mostly taken from
Bøckmann [9].
We will distinguish between three reference frames: the North-
East-Down reference frame (n-frame), the body-fixed reference
frame (b-frame), and the seakeeping frame (s-frame). The n-frame
is fixed to the Earth’s reference ellipsoid [16], the b-frame is fixed
to the vessel (see Fig. 10), and the s-frame is travelling with the ship
but its alignment is fixed so that it coincides with the b-frame in
absence of waves. The n-frame positions and Euler angles are given
by the vector , the b-frame linear velocities and angular velocities
are given by the vector , and the s-frame perturbation vector is .
The time-domain seakeeping equations of motion in the s-frame
for a vessel with no forward speed can be written, thanks to the
works of Cummins [17] and Ogilvie [18], as [19]
 t
[MRB + A(∞)]¨ + Btotal (∞)˙ + ˙
K(t − )() d + G =  FK+d ,
0
(12)
 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18
xb
surge
roll
sway
yb
heave pitch
yaw
zb
Fig. 10. Body-fixed reference frame in VeSim [9].
Fig. 11. Body-fixed vs. seakeeping reference frames.
where the matrix of retardation functions, K(t), is given as
 ∞
2 leading edge, and it introduces empirical time lags to model the
K(t) = [Btotal (ω) − Btotal (∞)] cos(ωt) dω.
 0
MRB is the rigid body mass matrix, A(∞) is the added mass matrix for
infinite frequency, Btotal is a damping matrix including both poten-
tial and viscous damping, G is the restoring force matrix, and  FK+d
is a vector containing Froude–Krylov and diffraction forces. A dot
over the symbol denotes time differentiation, and two dots imply
that time differentiation is done twice.
The s-frame assumes that the vessel travels straight with con-
stant speed, which is not the case when manoeuvring. VeSim needs
all external forces to be given in the b-frame. Therefore, the equa-
tions of motion, Eq. (12), should be transformed from the s-frame
to the b-frame. When transforming the equations of motion to the
rotating b-frame, Coriolis forces between the s-frame and the b-
frame appear when the ship has forward speed. The reason for
this is that the velocities and accelerations in the s-frame can be
expressed using velocities and accelerations in the b-frame, in Eq.
(12). This is illustrated in Fig. 11, where ıu and ıw are the wave-
induced vessel velocities in xb - and zb -direction, respectively. If 5 is
small and the origins of the b-frame and s-frame coincide (assumed
in this example, for simplicity),
3 ≈ ıw − Vs 5 ,
where we have assumed that Vs  ıu.
The equations of motion in the b-frame then become [19]
[MRB + AVs (∞, Vs )]˙ + C∗RB  + Btot,Vs (∞, Vs )r
 t
+ KVs (t − , Vs )[() − Vs e1 ] d + G =  FK+d ,
0 5. Model test and simulation results
where C∗RB  are Coriolis forces and moments due to the rotation of
the b-frame about the s-frame, as derived in a linearized version in
[20], and e1 is the unit vector in xb -direction. Subscript Vs denotes
speed-dependence. r includes ocean currents. Other forces that are
not necessarily harmonically varying can be added to the right side
of Eq. (15), when solving Eq. (15) in the time domain. The speed-
dependent added mass and damping coefficients are calculated
13
using the strip theory by Salvesen et al. [11] and transformed to
the b-frame. This transformation removes speed-dependent terms
with the encounter frequency, ωe , in the denominator, which is
undefined when ωe = 0. The retardation functions are now calcu-
lated using the speed-dependent damping matrices, as
 ∞
2
KVs (t, Vs ) = [Btotal,Vs (ω, Vs ) − Btotal,Vs (∞, Vs )] cos(ωt) dω.
 0
(16)
VeSim first solves Eq. (15) for the vessel accelerations in the b-
frame. Then, the accelerations are integrated to obtain the vessel
velocities in the b-frame. The vessel velocities are then transformed
into the n-frame and integrated once again to obtain the vessel
positions in the n-frame [12].
4.2. Dynamic stall model for wavefoil forces
A slightly modified version of the Leishman–Beddoes dynamic
stall model [14] for the wavefoil forces, as described in [9], was
implemented in VeSim. This dynamic stall model combines clas-
sical unsteady foil theory with an expression for the separation
point position. For fully attached flow, the lift is expressed through
an effective angle of attack, which can be expressed through the
quasi-steady angle of attack at previous time instants, as explained
in [21]. The separation point position is obtained from tables of
static lift and drag coefficients, as shown in [9]. The dynamic stall
model also accounts for the lift due to a vortex developing near the
(13)
dynamics properly.
A few differences between the model implemented in the
present paper and the model implemented in [9] exist:
1. The nondimensional time constant for leading-edge pressure
delay was taken to be 3 in the present model.
2. The nondimensional time constant for boundary layer delay was
taken to be 7 in the present model.
3. The nondimensional time constant for vortex delay was taken to
be 6 in the present model.
4. The angle of attack where the leading-edge vortex shedding
begins (as used in the dynamic stall model of Larsen et al. [22]),
˛v , was taken to be 12.5◦ in the present model.
5. The criteria for when the leading edge vortex begins to travel
downstream was taken to be ˛f > ˛v in the present paper,
instead of ˛ > ˛v as in [9].
˛f is a modified version of the quasi-steady angle of attack, ˛, as
described in [14]. ˛v can be found from the static moment coef-
ficient curve where there is a sharp change in the pitch moment.
Corresponding to ˛v is a critical normal force coefficient, which
(14) is used as the leading-edge vortex separation criteria in [14]. The
changes in the time constants relative to [9] were made because the
new values gave better fits to the experiment data in [9]. ˛v = 12.5◦
appeared to be more correct than ˛v = 11◦ , which was used in [9],
based on the moment coefficient curve for the NACA 0015 foil from
Sheldahl and Klimas [15] for the appropriate Reynolds number.
(15)
5.1. Calm-water resistance
5.1.1. Model scale
Fig. 12 shows the calm-water resistance of the model with
and without wavefoils. The corresponding wavefoil thrust is also
plotted. Positive wavefoil thrust implies that the foils exert a
14 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18

120
Resistance with wavefoils, exp.
Resistance without wavefoils, exp.
100 Foil thrust, exp.

80
Force [N]

60

40

20

0
0.6 0.8 1 1.2 1.4 1.6 1.8 2
Vm [m/s]

Fig. 12. Calm-water resistance and foil thrust, model scale.

forward-pulling force on the hull. Paint tests on the bare hull with
the same bow model shows that the flow at the hull near the wave-
foil location is directed downwards at an angle of approximately
21◦ to the horizontal when the full-scale ship speed is 14 knots
[23]. When considering this, the fact that the foils produce thrust
and not drag is not surprising.
Despite the forward pull from the wavefoils, however, we see
that the resistance in calm water increases when employing the
wavefoils. This resistance increase, we believe, is caused by the
flow interaction between the bulbous bow, the hull shape and the
wavefoils. Fig. 13a shows that there is a breaking wave located
approximately midway between stations 19 and 19.5 when the full-
scale speed is 9 knots. When employing the wavefoils, this wave
steepens (see Fig. 13b), which implies increasing wave-making
resistance.
The downward-and-forward-lifting wavefoil appears to gener-
ate a wave crest that travels forward with increasing ship speed.
For a ship speed of 13 knots, the wave patterns with and without
Fig. 14. (a) Bow wave pattern from bare model in calm water at a speed of 1.64 m/s
(13 knots in full scale). (b) Bow wave pattern from model with wavefoils in calm
water at a speed of 1.64 m/s.
wavefoils are fairly similar, but when looking closely at Fig. 14 we
see that the water level is slightly higher above the bulb with foil.
For a ship speed of 15 knots, the wavefoil-generated wave crest
amplifies the wave crest above the bulb further, as can be seen by
comparing Figs. 15a and b.
In Fig. 16, the frictional resistance coefficient in model scale, CFm ,
is plotted against model speed. The calm-water foil thrust in model
scale, Tfoil,m , is made nondimensional by dividing by 12 m Vm 2S ,
m
where Sm is the wetted surface area of the model, and plotted in
the same figure for comparison. Fig. 16 also shows residual resis-
tance coefficients with and without wavefoils, denoted CR,foil and
CR , respectively, calculated from Eq. (1). Note that CR,foil here not

Fig. 13. (a) Bow wave pattern from bare model in calm water at a speed of 1.13 m/s Fig. 15. (a) Bow wave pattern from bare model in calm water at a speed of 1.89 m/s
(9 knots in full scale). (b) Bow wave pattern from model with wavefoils in calm (15 knots in full scale). (b) Bow wave pattern from model with wavefoils in calm
water at a speed of 1.13 m/s. water at a speed of 1.89 m/s.
 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18 15

−3 5
x 10 x 10
5 CR,foil, exp. 4
Resistance with wavefoils, exp. (scaled)
CR, exp. Resistance without wavefoils, exp. (scaled)
3.5
C , exp. Foil thrust, exp. (scaled)
4 Fm

Tfoil,m/(1/2 ρmV2mSm), exp.

3
Force coefficient

3 2.5

Force [N]
2
2
1.5

1
1
0.5

0 0
0.6 0.8 1 1.2 1.4 1.6 1.8 2 4 5 6 7 8 9 10 11 12 13 14 15 16
Vm [m/s] V [knots]
s

Fig. 16. Calm-water force coefficients, model scale. Fig. 17. Calm-water resistance and foil thrust, full scale.

only includes the wave-making resistance of the hull with wave-

foils but also the wavefoil thrust. In other words, if the wavefoils
were at the same location but not connected to the hull (which is of
course not possible in reality), CR,foil calculated from Eq. (1) would
Tfoil,m
have become CR,foil + 1  V2 S
2 m m m
the wave-making resistance when employing the wavefoils at the
present location. We see that CR,foil remains fairly constant when
the model speed increases, compared with CR .
5.1.2. Full scale
Fig. 17 shows the calm-water resistance in full scale, with
and without wavefoils, when scaling the model scale resistance
according to the procedure described in Section 3.3. Since the fric-
tional resistance coefficient is less in full scale than in model scale,
whereas the residual resistance coefficient are equal in the two
scales, the relative difference between the calm-water resistance
graphs with and without wavefoils is larger in full scale.
. This implies a significant increase in
5.2. Regular waves, 11 knots speed in full scale
Fig. 18 shows full-scale resistance and wavefoil thrust from scal-
ing the experiment results and from VeSim in addition to RAOs for
heave and pitch. The lack of agreement in foil thrust is caused by
the fact that the wavefoils in VeSim are not affected by the flow
around the ship hull. Hence, in calm water there is a small drag
force on the wavefoil in VeSim – no thrust force as in the experi-
ment. We see that employing the wavefoils reduces the resistance

5 4
x 10 x 10

6
15
Resistance [N]

Foil thrust [N]

4
10

2 5

0 0
6 7 8 9 10 11 12 13 14 6 7 8 9 10 11 12 13 14
T [s] T [s]

1 1.5
Pitch RAO, ξ5A/(k ζA) [−]
Heave RAO, ξ3A/ζ [−]

0.8
1
A

0.6
0.5
With wavefoils, exp.
0.4
Without wavefoils, exp.
0 With wavefoils, sim.
0.2 Without wavefoils, sim.
With wavefoils, exp.+sim.
0 −0.5
6 7 8 9 10 11 12 13 14 6 7 8 9 10 11 12 13 14
T [s] T [s]

Fig. 18. Results for regular head-sea waves, 11 knots speed, full scale.
16 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18

5 5
x 10 x 10

8
2

Resistance [N]

Foil thrust [N]

6
1.5

4 1

2 0.5

0 0
6 7 8 9 10 11 12 13 14 6 7 8 9 10 11 12 13 14
T [s] T [s]

1 1.5

Pitch RAO, ξ5A/(k ζ ) [−]

Heave RAO, ξ3A/ζ [−]

0.8

A
1
A

0.6
0.5
With wavefoils, exp.
0.4
Without wavefoils, exp.
0 With wavefoils, sim.
0.2 Without wavefoils, sim.
With wavefoils, exp.+sim.
0 −0.5
6 7 8 9 10 11 12 13 14 6 7 8 9 10 11 12 13 14
T [s] T [s]

Fig. 19. Results for regular head-sea waves, 13 knots speed, full scale.

for the two shortest wave periods. Also shown in Fig. 18 are resis- this implies that the difference in calm-water resistance with and
tance and foil thrust graphs where the effect of the wavefoils in without wavefoils from the experiment is added to the resistance
calm water, found experimentally, is added to the VeSim results with wavefoils from VeSim. For the wavefoil thrust, the calm-water
(denoted “With wavefoils, exp. + sim”). For the resistance graph, wavefoil thrust is simply added to the wavefoil thrust from VeSim.

5 4
x 10 x 10

4
10
Resistance [N]

Foil thrust [N]

2 5

1
0
0
6 7 8 9 10 11 12 13 14 6 7 8 9 10 11 12 13 14
T [s] T [s]
p p

0.8 1.5

0.6
Pitch std [deg]
Heave std [m]

0.4
With wavefoils exp.
0.5 Without wavefoils exp.
0.2 With wavefoils sim.
Without wavefoils sim.
With wavefoils, exp.+sim.
0 0
6 7 8 9 10 11 12 13 14 6 7 8 9 10 11 12 13 14
Tp [s] Tp [s]

Fig. 20. Results for irregular head-sea waves, 13 knots speed, full scale.
 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18
Fig. 21. Wavefoil root barely submerged for Hs = 2.95 m, Tp = 11.02 s.
We see that this not surprisingly gives a better correlation with the
experiment graphs.
The pitch RAOs with and without wavefoils from VeSim are in
fairly good agreement with the pitch RAOs from the experiment.
The agreement between the heave RAOs with wavefoils from VeSim
and from the experiment is good except for a wave period of 9 s in
full scale. This is not surprising considering that the heave RAOs
from VeSim and from the experiment are in poor agreement for
this wave period without wavefoils.
For none of the wave periods tested in regular head-sea waves
with a speed of 11 knots in full scale did the wavefoils penetrate
the free surface.
5.3. Regular waves, 13 knots speed in full scale
Fig. 19 shows results for regular head-sea waves and a speed
of 13 knots in full scale. We see that the resistance in waves with-
out wavefoils from VeSim is in slightly worse agreement with the
experiment graph for a speed of 13 knots than for a speed of 11
knots.
The heave RAOs without wavefoils from VeSim and from the
experiment are not in very good agreement, which affects the
agreement between the heave RAOs with wavefoils from VeSim
and from the experiment.
For none of the wave periods tested in regular head-sea waves
with a speed of 13 knots in full scale did the wavefoils penetrate
the free surface.
5.4. Irregular waves, 13 knots speed in full scale
Fig. 20 shows results for irregular head-sea waves and a speed
of 13 knots. For the heave and pitch motions, the standard devia-
tion is given here instead of the RAOs. The same irregular waves as
measured in the model test were used in the simulations.
Model test and simulation results are in good agreement for
the resistance in waves and for the heave standard deviation. The
pitch motion is slightly underpredicted in the simulations. The foil
thrust is clearly underpredicted in the simulations, because VeSim
is not able to predict forward foil thrust in calm water. Adding the
measured foil thrust in calm water to the foil thrust obtained with
VeSim in waves brings the foil thrust curve from VeSim much closer
to the foil thrust curve from the model test.
For none of the irregular head-sea wave conditions tested with
a speed of 13 knots in full scale did the wavefoils penetrate the free
surface, although the foil roots came very close to doing so at a few
occasions (see Fig. 21).
17
6. Conclusions and future work
There was a significant wavefoil thrust in calm water, yet the
calm-water ship resistance increased with wavefoils. The foils were
able to produce thrust because the inflow to the foils had an angle
relative to the horizontal, caused by the hull shape. We believe the
reason why the ship resistance increased despite of the foils pro-
ducing forward thrust is that the foils modified the wave-making
resistance of the ship unfavourably compared to the wave-making
resistance of the bare hull. Photos of the wave pattern in the
bow region with and without wavefoils support this. Studying the
effect of the location of the wavefoils on calm-water resistance, for
instance by means of computational fluid dynamics (CFD), would
be interesting to pursue as future work.
In four sea states of the JONSWAP spectrum, with Hs slightly
less than 3 m and Tp values in the range 7–13 s, the resistance
reductions with wavefoils for the ship tested at 13 knots in head
seas were in the range 9–17% in full scale, based on scaling the
model test results. This resistance reduction does not include the
resistance contribution from possibly increased rudder usage when
employing the wavefoils, but this effect is assumed to be small. The
heave reductions with wavefoils for the same four sea states varied
between −11% and 32%, whereas the pitch reductions were in the
range 11–25%. The difference between the resistance curves with
and without foils cannot always be directly explained by the foil
thrust and reduction in added resistance in waves due to reduced
motions. We attribute this to the increased calm-water resistance
with wavefoils, which reduced the benefits of the wavefoils.
Simulated resistance, the shape of the foil thrust curves, and
ship motions are generally in fairly good agreement with the model
test results. The value of the simulated foil thrust deviates clearly
from the measured foil thrust since VeSim is not able to predict for-
ward foil thrust in calm water. Adding measured foil thrust in calm
water to the simulated foil thrust in waves gives better agreement
between simulated and measured foil thrust in waves.
In none of the runs with wavefoils described in this paper did
the wavefoils penetrate the free surface. If the significant wave
height had been higher, however, the foil roots would probably
have emerged from water.
Acknowledgements
The authors would like to acknowledge the funding of this test-
ing by Rolls-Royce and Innovate UK through the WAFT (Wave
Augmented Foil Technology) project (Project Reference: 101512)
and with the support of the Rolls-Royce University Technology Cen-
tre at NTNU. The WAFT project was a collaborative research project
between Rolls-Royce plc, Seaspeed Marine Consulting Ltd, Marine
One Stop Technologies Ltd, and the Norwegian University of Sci-
ence and Technology. The authors acknowledge the valuable inputs
from the project partners. Dariusz Fathi and Ole Hermundstad at
MARINTEK are thanked for help with the simulations, and Volker
Bertram at DNV GL is thanked for advice on the paper in general.
Last but not least, the authors would like to thank the MARINTEK
personnel involved in the model test for excellent work.
References
[1] Intergovernmental Panel on Climate Change. Climate change 2014 synthesis
report; 2014.
[2] International Maritime Organization. Second IMO GHG Study 2009; 2009.
[3] Vrooman D. Vibrating propeller. US Patent 22,097. Patented November 15,
1858.
[4] Burnett RF. Wave energy for propelling craft – nothing new. Naval Archit
1979;(November):239.
[5] Jakobsen E. The foilpropeller, wave power for propulsion. In: Papers presented
at the Second International Symposium on Wave & Tidal Energy. 1981. p. 363–9.
18 E. Bøckmann, S. Steen / Applied Ocean Research 57 (2016) 8–18

[6] Naito S, Isshiki H. Effect of bow wings on ship propulsion and motions. Appl of vertical axis wind turbines. In: SAND80-2114. Albuquerque, NM: Sandia
Mech Rev 2005;58(4):253–68. National Laboratories; 1981.
[7] Belibassakis KA, Politis GK. Hydrodynamic performance of flapping wings for [16] NIMA. Department of Defense World Geodetic System 1984. NIMA Technical
augmenting ship propulsion in waves. Ocean Eng 2013;72:227–40. Report TR8350.2; 1997.
[8] Bøckmann E, Steen S. Experiments with actively pitch-controlled and spring- [17] Cummins WE. The impulse response function and ship motions. Schiffstechnik
loaded oscillating foils. Appl Ocean Res 2014;48:227–35. 1962;9(47):101–9.
[9] Bøckmann E. Wave propulsion of ships. Norwegian University of Science and [18] Ogilvie TF. Recent progress toward the understanding and prediction of ship
Technology; 2015 (Ph.D. thesis). motions. In: 5th Symposium on Naval Hydrodynamics. 1964. p. 3–80.
[10] Kring DC. Time domain ship motions by a three-dimensional Rankine panel [19] Fossen TI. Handbook of marine craft hydrodynamics and motion control. 1st
method. Massachusetts Institute of Technology; 1994 (Ph.D. thesis). ed. John Wiley & Sons, Ltd.; 2011.
[11] Salvesen N, Tuck EO, Faltinsen OM. Ship motions and sea loads. Trans SNAME [20] Fossen TI. A nonlinear unified state-space model for ship maneuvering and
1970;78:250–87. control in a seaway. Int J Bifurc Chaos 2005;15(09):2717–46.
[12] Fathi D. MARINTEK Vessel Simulator (VeSim), Vessel Model. Rev. 5. MARINTEK; [21] Leishman JG. Principles of helicopter aerodynamics. Cambridge aerospace
2013. series; 2002.
[13] Fathi D, Hoff JR. ShipX Vessel Responses (VERES), Theory Manual. MARINTEK; [22] Larsen JW, Nielsen SRK, Krenk S. Dynamic stall model for wind turbine airfoils.
2014. J Fluid Struct 2007;23(7):959–82.
[14] Leishman JG, Beddoes TS. A semi-empirical model for dynamic stall. J Am Heli- [23] Alterskjær SA. R&D8000 Phase 1, Calm Water Tests with 3 Different Fore Ships.
copter Soc 1989;34(3):3–17. MARINTEK report MT53 F10-200 530635.00.03; 2010.
[15] Sheldahl RE, Klimas PC. Aerodynamic characteristics of seven symmetrical air-
foil sections through 180-degree angle of attack for use in aerodynamic analysis