Content uploaded by Thomas P Lloyd
Author content
All content in this area was uploaded by Thomas P Lloyd on Mar 25, 2024
Content may be subject to copyright.
Eighth International Symposium on Marine Propulsors
smp’24, Berlin, Germany, March 2024
Ship URN mitigation by air injection: model-scale experiments and
application to full-scale measurement data
Thomas Lloyd1, Frans Hendrik Lafeber1, Johan Bosschers1
1Maritime Research Institute Netherlands (MARIN), Wageningen, the Netherlands
ABSTRACT
It is recognised that continuous underwater radiated noise
(URN) from shipping needs to be mitigated in order to
minimise impact on marine animals. To this end there are
numerous technical and operational URN mitigation mea-
sures, although extensive data concerning their effective-
ness is not always available in the open domain. Examples
of this include the so-called ‘Masker’ and ‘Prairie’ systems,
designed to reduce machinery and propeller cavitation
noise respectively by means of air injection. These types
of systems are being studied within the EU Horizon
2020-funded project SATURN.
In this paper, we report model-scale sound measurement
tests of both systems and demonstrate application of the
measured data to a ship-scale test case. Two dedicated
ship models were made based on a tanker hullform, with
custom air injection systems developed. An overview of
the model test campaign is reported, covering preparation,
measurements and data analysis. After this, results of the
full-scale application case are presented, including system
power requirement estimation alongside URN abatement
potential.
From the model tests URN reductions of up to about 22 dB
and 12 dB were found for the Masker and Prairie-like sys-
tems respectively, although for the ship-scale application
this is limited to about 10 dB at the vessel design speed.
Keywords
underwater radiated noise; cavitation; machinery; bubbles;
air injection
1 INTRODUCTION
Underwater radiated noise (URN) from ships is receiving
increasing attention for its adverse impacts on marine life
(Duarte et al. 2021). Continuous noise from shipping -
generated primarily by propeller cavitation and machinery -
contributes to ambient sound levels in the oceans at a broad
range of frequencies, with the potential to negatively affect
a large number of species. Although there is currently
no international regulation specifically addressing ship
URN, in the European Union (EU) Member States are
required to monitor ambient sound levels in the 63 Hz
and 125 Hz decidecade frequency bands (EU 2017), with
a procedure for determining threshold values for impact
recently proposed (Technical Group on Underwater Noise
2022). Noise abatement is currently being addressed
by multiple stakeholder groups, in anticipation of future
regulation (Cruz et al. 2022). At international level,
the International Maritime Organization (IMO) recently
published revised guidelines for the reduction of URN from
shipping (IMO 2023), while significant attention has been
paid to monitoring and mitigation of ship URN within
the Enhancing Cetecean Habitat and Observation (ECHO)
Program (see e.g., Joy et al. 2019). The EU Horizon 2020
SATURN project is one initiative, addressing the sources,
impacts and possible technical mitigation solutions. Within
this project, one of the measures being studied by MARIN
is air bubble injection for reduction of both propeller
cavitation and machinery URN.
Although air injection systems for reducing noise and
vibrations on board merchant ships have previously been
studied (Hadler et al. 1984; Krueger et al. 2004), very
little information is available in the open literature focusing
on URN mitigation for similar vessels. Systems for this
purpose have been developed for and used by a number
of navies, meaning that almost no performance data is
available in the public domain. Reduction of cavitation
URN by air bubble injection was studied in the EU project
SONIC however (Baudin & Mumm 2015). The aim of the
present work was therefore to quantify the noise reduction
potential of such air injection systems, in order to provide
performance data for use in the design of (newbuild and
retrofit) quiet ships.
For machinery noise abatement, the so-called ‘Masker’
system is used, which generates an air bubbles layer around
the hull, thereby isolating the machinery noise sources on
board the vessel from the surrounding water. In this case,
the URN reduction is achieved by creating an impedance
jump due to the bubbly mixture close to the hull. To reduce
propeller cavitation noise, the ‘Prairie’ type of system was
developed. This works by injecting air into the cavitation
through holes in the leading edges of the propeller blades.
The non-condensable gas dampens the collapse of the
cavity, which is the main noise generation mechanism. In
this work, we studied a solution whereby air is injected
from upstream of the propeller disc, which we refer to as a
‘Prairie-like’ system. Sketches of the two systems applied
to a cargo vessel are shown in Figure 1.
Figure 1: Schematic representations of Masker and Prairie-
like systems applied to a cargo vessel.
This paper reports the model test campaigns performed at
MARIN for studying the performance of the Masker and
Prairie-like systems, followed by an example application
of the processed measurement data to a full-scale test case.
Although several different types of measurements have
been performed for each system, we focus here on the
URN results. More details of the complete model test
programme are given in Lloyd et al. (2024).
2 MODEL TESTS
2.1 Ship geometry
A common hullform geometry was selected for the scale
model tests of both air injection systems. The chosen
ship was a 7000 DWT tanker, previously used in the EU
Streamline project (Dymarski et al. 2011) and not subject to
restrictions in sharing of the hull and propeller geometries
or associated results. She has a design speed of 14 knots
and is equipped with a single fixed-pitch propeller. Main
particulars of the hull and propeller are given in Table 1.
Table 1: Streamline tanker principal particulars.
Particular Symbol Value Unit
Length between perp. Lbp 94.00 m
Beam B15.42 m
Draught (forward/aft) T6.01 m
Block coefficient Cb0.762 -
Propeller diameter D3.85 m
Number of blades Z4 -
Pitch at 0.7R P0.7/D 1.00 -
Exp. area ratio AE/A00.579 -
Two dedicated scale models were developed for the present
work, due to the differing requirements of the measure-
ments for each air injection system. This also resulted
in two different scale factors (λ), with 12.00 used for the
Masker system and 12.83 for the Prairie-like system.
2.2 Experimental design and execution
2.2.1 Test facility
All tests were performed in MARIN’s Depressurised Wave
Basin, which measures 240 m x 18 m x 8 m. The ambient
air pressure in the facility can be reduced in order to obtain
cavitation number similarity in combination with Froude
scaling of the model speed and propeller rotation rate. This
was applied when testing the Prairie-like system, but was
not required for the Masker system, with these tests per-
formed at atmospheric pressure conditions. Furthermore,
all tests were performed in calm water conditions with the
models towed straight ahead at even keel. A dedicated
silent towing carriage (Bosschers et al. 2013) was used
for the URN measurements, with all other measurements
made using the main towing carriage. Sound was measured
using hydrophones mounted on a mast at the centre of the
basin, which the models sail over. Results from a single
hydrophone - located on the tank centreline - are presented
in this work. More details of the test facility and setup for
URN measurements can be found in Lloyd et al. (2018).
2.2.2 Masker system
In order to imitate ship machinery noise, a simplified
scaled metal midship section was developed and attached
to wooden fore and aft sections to form the ship model.
The metal section was excited by a shaker which was
fed with white noise signals, with the aim of achieving a
broadband excitation of the ship-like structure. A low-pass
(LP) filtered signal was used for some tests with the aim
of increasing the response of the structure - and therefore
signal-to-noise ratio of the tests - at low frequencies.
No propeller was included since the focus was only on
machinery noise. Air was injected along the sides and keel
of the model through porous hoses recessed into the hull
just upstream of the metal midship section. The model was
painted blue in order to improve the quality of high-speed
camera observations of the bubble layer, which were made
in keel and beam aspect. Full details can be found in Lloyd
et al. (2023).
In addition to URN measurements, a custom bubble
measurement system was developed in-house. This system
was able to quantify the time-averaged normal-wall air void
fraction distribution, at discrete girthwise and streamwise
positions under the metal midship section. More informa-
tion on the development and application of this system can
be found in Klinkenberg et al. (2023). A photograph of the
Masker scale model, showing both the air injection system
and bubble measurement system, is provided in Figure 2.
Firstly, observations of the bubble layer were made in order
to select appropriate air flow rates at several model speeds.
Following this, the URN measurements were performed, in
which the model speed, air flow rate and shaker signal were
varied. Finally, for selected conditions, the characteristics
of the bubble layer were measured at three stations.
Figure 2: Air injection (Masker) system and bubble measure-
ment system developed for present model tests: perspective
view from port side below the keel.
2.2.3 Prairie-like system
A Prairie-like system was developed with air injected
through needles mounted in a duct located upstream
of the propeller disc. The duct was designed (using
computational fluid dynamics) to have a minimal effect on
propeller performance, and sized and oriented with the aim
of having bubbles enter the cavity close to the propeller tip.
The final design was manufactured using 3D printing, with
air supplied to the upper and lower halves independently
via two mass flow controllers.
The model was prepared for cavitation and URN measure-
ments following MARIN’s standard procedures, with the
propeller painted blue and sandgrain roughness applied to
the leading edges of the propeller blades. Figure 3 gives an
impression of the Prairie-like system.
Figure 3: Prairie-like system developed for present model
tests. Air is injected from needles mounted in the trailing edge
of the duct upstream of the propeller rather than from holes
in the propeller blade leading edges (as in the original Prairie
system).
Observations were performed first, using high-speed cam-
eras. Tests were performed for four different com-
binations of propeller thrust coefficient and cavitation
number, designed to cover low and high propeller loading
conditions, with varying forms and degrees of cavitation
present. This was done intially without air injection for
baseline performance assessment, after which a range of
air flow rates were used. Propeller thrust and torque,
and hull pressure fluctuations, were also measured. The
test conditions were subsequently repeated to measure
URN. Lastly, shadowgraphy measurements were made for
selected test conditions (air flow rates and model speeds),
using a dedicated high-speed camera setup from inside
the model, with the aim of quantifying the bubble size
distribution generated by the injectors. Further details can
be found in Lloyd et al. (2024).
2.3 Data analysis
2.3.1 Source levels
Since the measurements of the Prairie-like system closely
resembled other typical URN setups tested in the DWB,
standard data processing procedures could be used to
estimate the source levels (SLs). This involves windowing
the measurement such that only data points within the
reverberation radius of the facility are included, after which
this data is segmented and corrected for propagation loss
(including the Lloyd’s mirror effect), before correcting for
background noise and averaging over all segments to obtain
the final (model-scale) SL. Scaling follows International
Towing Tank Conference (ITTC) Recommended Proce-
dures (ITTC 2017). Further details can be found in Lloyd
et al. (2018).
In the case of the Masker system, a modified procedure
was adopted as the noise source being studied was quite
different from propeller cavitation noise. Since machinery
noise is not monopole in nature, no Lloyd’s mirror
correction was applied. In addition, no scaling of the
derived source levels was carried out, assuming that the
measured sound level reduction was primarily a function
of the bubble layer void fraction and that the sound was
generated by a generic representative broadband source
(Lloyd et al. 2023).
2.3.2 Air injection system performance
The performance of the air injection systems in terms
of change in sound levels is computed as the difference
between the predicted source levels without and with the
system turned on. For the Masker system, this is termed
the insertion loss (IL), since the mitigation device is located
between source and receiver, and the source itself is not
reduced in magnitude. For the Prairie-like system, injecting
air directly into the cavity means that the source strength
does reduce, hence we refer to this as the source level
attenuation (SLA). These two quantities are simply defined
as the difference in SL between the test condition with the
system switched off and switched on.
The difference between IL and SLA may appear to be
largely semantic, with both quantities derived in the
same way, even if the URN reduction mechanisms are
different. Despite this, the distinction in terminology
is important since the ‘source level’ at one metre when
the Masker system is switched on does not represent the
strength of the sound source; it can be thought of as
a distance-corrected sound pressure level. This is not
relevant for the URN performance quantification however,
which for both systems is based purely on the difference in
the processed sound levels between tests with and without
air injection.
2.4 Exemplar model test results
This section contains a brief overview of the URN results
and related bubble observations. For full results the reader
is referred to Lloyd et al. (2024).
2.4.1 Masker system
A high-speed camera observation of the Masker air bubble
layer in keel aspect is shown in Figure 4. The model speed
(Vm) is 2.13 m/s - equivalent to the ship service speed -
and the air flow rate (Qm) was 3.75×10−3m3/s. Overall a
rather uniform bubble layer can be seen. The bubbles were
also observed to convect reasonably evenly over the sides
of the hull for this condition.
Figure 4: Example bubble layer observation for Masker
system. Camera mounted in keel aspect on hydrophone mast,
with model sailing from right to left.
A corresponding SL spectra comparison is presented in
Figure 5, with the associated IL shown in Figure 6. There
is a clear SL reduction across most of the frequency range,
resulting in an IL of up to 21 dB for this case. Despite
this, at certain frequencies the measured mitigated source
level is equal to the source level with bubble injection
only. The derived IL is limited at these frequencies,
primarily due to the lower response of the metal midship
section. The blue spectrum in Figure 5 exhibits several
distinct peaks, between which the SL is up to 25 dB lower,
thereby reducing the signal-to-noise (SNR) ratio at these
frequencies. This artefact is denoted by dashed faded lines
in Figure 6, indicating that the true IL may be higher.
Another point of note is the small increase in SL (negative
IL) which was obtained across multiple test conditions for
a limited frequency range (between about 300 Hz to 400
Hz in this case). It is hypothesised that this may be related
to bubble resonance (Lloyd et al. 2023).
Figure 5: Example processed decidecade bandwidth source
level spectra for Masker system.
Figure 6: Example insertion loss spectrum for Masker system.
Dashed lines indicate frequencies for which the predicted
insertion loss is limited by the signal-to-noise ratio of the
measurements.
2.4.2 Prairie-like system
The test conditions for the Prairie-like system are defined
in terms of the propeller thrust coefficient and cavitation
number:
KT=T
ρn2D4(1)
and
σn=2(patm +ρghshaft −pv)
ρn2D2,(2)
as well as the air flow rate. In Eq. 1, Tis the mean propeller
thrust, ρthe water density and nthe propeller rotation rate
in Hertz. In Eq. 2, patm and pvare the atmospheric and
vapour pressures respectively, gacceleration due to gravity
and hshaft the (reference) propeller shaft immersion.
An example cavitation and bubble observation for the
Prairie-like system can be seen in Figure 7, for which
KT=0.274, σn=0.259 and Qm(×106) = 25.0 m3/s.
Here (as for most of the tests performed) air was only
injected from the upper half of the duct, with the aim of
bubbles entering the cavity close to the top position, where
the cavity was at its largest, and just before the primary
collapse occurred.
Figure 7: Example cavitation and bubble observation for
Prairie-like system.
Source level spectra for this test condition are shown in
Figure 8, where (in contrast to the Masker system) the
background noise spectrum is the same both with and
without air injection, as a much smaller amount of air is
injected. The corresponding SLA is plotted in Figure 9.
Although the SNR was not sufficient at some frequencies
- resulting in data points being omitted - modest to large
reductions in source level are seen at most of the remaining
frequencies. The largest reduction in all cases is found for
the cavitation ‘hump’ (here centred at 40 Hz), indicating
that the non-condensable gas is able to dampen the main
cavity collapse. At lower frequencies (below 30 Hz here),
the SL increases when air injection is applied, which could
be due to the increased size (thickness) of the cavity when
it contains air.
Figure 8: Example processed full-scale decidecade bandwidth
source level spectra for Prairie-like system.
Figure 9: Example full-scale source level attenuation spec-
trum for Prairie-like system.
3 APPLICATION TO FULL-SCALE MEASUREMENT
DATA
We now present the application of results such as those
presented in Section 2 to a case of measured ship URN.
3.1 Test case description
A test case from the open literature was selected (Arveson
& Vendittis 2000). It concerns the bulk cargo ship M/V
Overseas Harriette, built in 1978, with an overall length
of 173 metres. The vessel is propelled by a 4.9 metre
diameter fixed-pitch propeller directly driven by a 8.4 MW
two-stroke diesel engine. Auxiliary power is provided by a
diesel generator. Decidecade radiated noise level (RNL)**
spectra measured in keel aspect are available for five test
conditions. Three of these were selected in the present
work, corresponding to the lowest, middle and highest ship
speeds tested. The RNL spectra for these speeds are shown
in Figure 10.
Figure 10: Radiated noise level spectra for full-scale appli-
cation case. Data taken from Arveson & Vendittis (2000).
Legend indicates vessel operating condition in terms of ship
speed and propeller rotation rate.
The three conditions show clear differences in terms of
spectral characteristics, which can be interpreted as being
due to the increasing contribution of propeller cavitation
noise with vessel speed. At 8 knots, machinery tonals
**Spherical spreading loss was applied as opposed to calculating the propagation loss.
dominate the spectrum at low frequencies (about 25 Hz)
while the hump centred between 300 Hz and 400 Hz may
be caused by the diesel generator. Note that the propeller
is not cavitating at this speed, with the cavitation inception
speed (CIS) reported to be 10 knots (Arveson & Vendittis
2000). For 12 knots, the machinery noise contributions at
the aforementioned frequencies are still visible, yet there is
a broadband increase in RNL, particularly centred between
50 Hz and 60 Hz, which can be attributed to the cavitating
tip vortex. At 16 knots, this characteristic hump has
increased in terms of both level and width, while cavitation
noise dominates the spectrum at all frequencies, due to this
source mechanism having a higher speed-dependency than
machinery noise.
Table 2 summarises the three test conditions and the model
test data selected for application to each. For the Masker
data, the highest measured insertion loss was used, which
was obtained for the highest model speed and air flow rate
tested. In this we assume that the power requirement of the
air injection system (at ship scale) is not a deciding factor
when selecting the air flow rate. Furthermore, part of the
variation in IL as a function of model speed was attributed
to scale effects in the model tests (Lloyd et al. 2024) and
we therefore do not try to match the (scaled) speed from
the model tests with that from the application case. On
the other hand, for the Praire-like system, source level
attenuation data were selected by taking the model-scale
test conditions having the closest correspondence (in terms
of ship speed and propeller rotation rate) with those of the
ship-scale case. The ship speeds match reasonably well
across both conditions. For the 16 knots case, although
the speed and rotation rate differ by about 14 %, the
propeller apparent advance ratios JV=Vs/(nsD)are very
similar (0.72 and 0.70 for the M/V Overseas Harriette
and Streamline tanker respectively). The main reason for
trying to match the test conditions is that the performance
of the Prairie-like system was observed to be somewhat
dependent on the propeller loading condition (Lloyd et al.
2024). No Prairie-like data is applied at 8 knots since the
propeller is not cavitating.
Table 2: Test conditions for application case and corresponding selected model test conditions.
Full scale Masker Prairie
Condition VsNsVsSignal type Qm(×103)VsNsQm(×106)
kn rpm kn m3/s kn rpm m3/s
Low speed 8 68 14 no filter 6.25 - - -
Middle speed 12 106 14 no filter 6.25 12 108 16.7
High speed 16 140 14 no filter 6.25 14 160 25.0
3.2 Data processing
The IL and SLA spectra used for the present analysis are
shown in Figure 11 and 12.
Figure 11: Masker system insertion loss for selected operating
condition. Comparison data for Big Bubble Curtain taken
from Bellmann (2014).
Prior to applying the data to the ship-scale measured RNL
some further processing was applied to the insertion loss
spectra. The IL obtained from the present measurements is
compared to data from literature for a ‘Big Bubble Curtain’
(BBC), used to mitigate noise from offshore pile-driving
(Bellmann 2014). Data for the BBC at the closest air
flow rate to the selected Masker test condition was chosen,
where the flow rate of the Masker data was scaled to the
same units as used by Bellmann (2014) for reporting the
BBC results; that is, the volumetric flow rate per metre
injector.
Figure 12: Prairie-like system source level attenuation for
two operating conditions. Dotted lines indicate data filling by
linear interpolation.
Overall, a good agreement is seen between the two
datasets, with a similar trend in terms of spectral shape and
level across much of the frequency range presented. As
discussed in Section 2, the magnitude of the IL obtained
from the present measurements was sometimes limited
by the achievable SNR. The comparison with the BBC
data seems to reinforce this observation, with a very
good agreement seen at frequencies for which a strong
resonant response of the scaled metal midship section was
measured. This motivated two further data processing steps
before applying the IL data to the ship RNLs. Firstly,
the missing IL data at low frequencies was extrapolated
using data from the BBC (dashed blue line in Figure
11). Following this, two data fits were made in order
to obtain a smoother IL spectrum: a linear regression in
three frequency ranges using all data points; and a linear
fit between the peaks of the Masker IL data. These two
data fits therefore represent ‘conservative’ and ‘optimistic’
estimates of the Masker system performance respectively
(see Figure 11).
Data fitting was not performed for the SLA data since
similar artefacts from the model tests were not present in
the results. However, in order to have continuous data
across all decidecade frequency bands, the missing data
points were approximated by linear interpolation. This is
indicated by dotted lines in Figure 12.
For the present application case, it is necessary to estimate
the contribution of propeller cavitation noise to the total
measured RNL spectrum. This can be done by:
LRN,cav (f) [dB] = 10 log10 10
LRN,total(f)
10
−10
LRN,mach(f)
10 ,(3)
where LRN,total is the RNL as reported in Arveson &
Vendittis (2000) and LRN,mach is the machinery noise,
taken as the 8 knots condition. Subsequently, the mitigated
RNL for each source mechanism can be computed as:
LRN,cav,air (f) [dB] = LRN,cav (f)−S LA (f)(4)
and
LRN,mach,air (f) [dB] = LRN,mach (f)−I L (f),(5)
following which the total mitigated RNL is obtained as:
LRN,total,air (f) [dB] = 10 log10 10
LRN,cav,air (f)
10
+10
LRN,mach,air (f)
10 .(6)
The overall change in RNL including the mitigation effect
of both Masker and Prairie-like systems is given by:
∆LRN,total (f) [dB] = LRN,total,air (f)−LRN,total (f).
(7)
3.3 URN mitigation results
Results are presented in terms of the reference and mit-
igated RNL, as well as the change in RNL, for each of
the three speeds considered. Figure 13 shows the RNL
spectra for the three ship speeds considered. For the
12 knots and 16 knots cases, the effect of the systems on the
cavitation and machinery source mechanisms separately is
also presented, in Figure 14.
Figure 13: Unmitigated and mitigated radiated noise level
combining propeller cavitation and machinery noise, for
three operating conditions: 8 knots (top); 12 knots (middle);
and 16 knots (bottom). Left-hand y-axis shows RNL and
right-hand y-axis gives the change in RNL. ‘Conservative’
data fit for Masker insertion loss used.
Radiated noise level reductions of almost 20 dB are found
for the non-cavitating condition reducing to about 10 dB
above CIS. For the lowest speed, this is centred at 1 kHz,
where the Masker system is most effective, with little
effect at low frequencies where machinery noise tonals
are present. For the other two speeds, the contribution
of machinery noise to the total RNL is smaller, meaning
the effect of the Masker system does not contribute as
much to the change in RNL. At 12 knots, the Prairie-like
system leads to increased RNL in the frequency range
where the Masker system works best, meaning that any
benefit is cancelled out. For 16 knots, cavitation noise
dominates at almost all frequencies, making the Masker
system ineffective. As expected, the Prairie-like system is
most effective at frequencies centred around the cavitation
hump peak frequency. The RNL increases at very low
frequencies (<20 Hz) while changes at higher frequencies
can be positive or negative depending on the case, but
mainly lie within about 3 dB.
Figure 14: Unmitigated and mitigated radiated noise level
when applying appropriate air injection system to propeller
cavitation and machinery noise source mechanisms sepa-
rately, for two operating conditions: 12 knots (top); and
16 knots (bottom). ‘Conservative’ data fit for Masker
insertion loss used.
A summary of the estimated changes in RNL for all three
ship speeds is shown in Figure 15, which includes results
for both ‘conservative’ and ‘optimistic’ performance of
the Masker system. The optimistic scenario results in an
increase of the peak change in RNL from 19 dB to 22 dB
for the 8 knots case, while the difference between the two
modelling approaches is much smaller when applied to the
two higher speeds, since the Masker system is less effective
in reducing the total RNL for these cases.
3.4 Air flow rate and power requirement
An estimate should also be made of the required air flow
rate when applying the air injection systems on board ships,
such that compressor power requirement can be specified.
Here we present arguments for scaling of the air flow
rates used during the model tests to full scale, for both
Masker and Prairie-like systems. Sketches of both systems
including the important parameters are shown in Figures 16
and 17.
Figure 15: Comparison of change in radiated noise level for
each of the three operating conditions considered: ‘conserva-
tive’ estimate (top); and ‘optimistic’ estimate (bottom).
Figure 16: Schematic representations of Masker system used
to derive scaling laws for air flow rate.
Figure 17: Schematic representations of Prairie-like system
used to derive scaling laws for air flow rate.
For the Masker system, the required air flow rate is
assumed to scale as:
Q∝V·b·d·α, (8)
while for the Prairie-like system this becomes:
Q∝V·Nin ·sin ·a. (9)
In Eq. 8, Vis the forward speed, bthe injector length in
the girthwise direction, dthe bubble layer thickness and α
the air void fraction. In Eq. 9, Nin and sin are the number
of injectors and spacing between them respectively, while
ais the bubble radius. These variables are also indicated
in the schematics shown in Figures 16 and 17. In Eq.
9 we assume that it is possible to replace awith 3
p⟨Va⟩
where ⟨Va⟩is the mean air bubble volume and ⟨Va⟩ ∝ α.
Furthermore, we do not account for changes in cavitation
extents (propeller thrust coefficient or cavitation number)
on the required flow rate, although this should be studied
as part of future work. Applying Froude scaling in both
cases, the ratio of the required air flow rates on ship- and
model-scale becomes a function of the scale factor:
Qs
Qm
=(λ1.5,Masker
λ2.5,Prairie-like. (10)
It is also necessary to estimate the power requirement of
the systems so that a suitable compressor can be selected.
This is done here following M¨
akiharju et al. (2012), who
give the required compressor power as:
Pc=Qpatmγ
ηc(γ−1) "pin
patm γ−1
γ
−1#,(11)
where pin =patm +ρgTin is the static pressure at the
injector and Tin is the draught at the injector(s). Table
3 summarises the results obtained using Eqs. 10 and 11.
Overall, the power requirement is modest compared to
other power demands on board (when considering typical
installed auxiliary engine power for cargo vessels). Of
course, the Prairie-like system requires much less air than
the Masker system, since it is targeted at only part of the
propeller disc compared to a relatively large section of the
hull in the latter case.
Table 3: Air flow rates and required compressor power for
ship scale application of Masker and Prairie-like systems,
estimated from model-scale air flow rates.
System Qm/(m3/s)Qs/(m3/s)Pc/ kW
(×103) (×106) (×103)
Masker 6.25 260 24.8
Prairie-like 16.3 10 0.7
25.0 15 1.0
4 DISCUSSION
Several points should be noted when interpreting the results
presented, which also require further study.
One important aspect is how to apply the SLA data. This
is due to the fact that the effect of the Prairie-like system
is mainly to reduce the magnitude of the spectral hump
caused by tip vortex cavitation, while in general also
leading to an increase in source levels at (harmonics of)
the blade passing frequency (BPF). Since the frequency
of the hump peak is a function of the propeller design
and loading condition (Bosschers 2018), the measured
SLA spectrum is characteristic to the ship and propeller
geometry used for the model tests, as well as the selected
test conditions. Therefore the SLA data should not be
applied to measurement data from other cases without
shifting the maximum SLA to the correct peak frequency
of the cavitation spectra hump, and the low-frequency
negative SLA to the BPF, for the case being considered.
Following Bosschers (2018), the peak frequency fpcan be
estimated as:
fp
fbp ∝1
τ KT
Z√σn
√σn
Z=σn
τKT
,(12)
where fbp =nZ is the BPF, τis a propeller maximum
tip loading parameter, which models the effect of propeller
design and hull form, and all other symbols have been
previously defined. Applying Eq. 12 for the present case,
a good agreement between the propeller rotation rate and
loading condition was found between the M/V Overseas
Harriette and the Streamline tanker, with the hump peak
frequencies for the highest speed case being approximately
equal to 50 Hz for both vessels. Based on this, no further
adjustment of the SLA spectra was performed for the
results shown here. However, it is expected that this will
be necessary for other cases; for example, when a vessel
has a much higher propeller rotation rate.
Another consideration relating to the Praire-like system
is that the air injection can lead to a degradation in the
propulsive performance of the propeller. This means
that the model tests were not always performed at the
self-propulsion condition. One way to correct for this effect
is to estimate a new self-propulsion condition when the
system is switched on, resulting in a change in rotation
rate, as well as associated propeller efficiency and URN
source levels. For the results presented here, the maximum
relative changes in rotation rate and propeller efficiency
were estimated to be about +4 % and -5 % respectively.
The accompanying change in SL can be predicted by
rudimentary scaling as a function of propeller tip speed
(e.g. see ITTC 2017) and was found to be about +1.5 dB.
This serves to increase the expected absolute mitigated
source levels in practice and has not been included in the
analysis presented in Section 3. We expect that this effect
can be minimised by reducing the amount of air injected
through optimisation of the injector design to focus on
the wake peak rather than the complete upper half of the
duct. In addition, it may be possible to offset any change
in propeller efficiency and/or source level by re-desiging
the propeller using optimisation techniques, taking the
application of a Prairie-like system into account.
One final point is the differences between the model
tests and full-scale application in terms of ship motions,
something which may affect the performance of both
systems. Since the model tests were performed in ideal
conditions, without any seaway modelled, no data is
available on the relationship between ship motions and
URN reduction. However, we might expect that (roll)
motion would result in air escaping more quickly towards
the free surface in the case of the Masker system, while for
the Prairie-like system (time-varying) oblique inflow to the
propeller may reduce the amount of air entering the cavity.
This motivates additional model- or full-scale tests to study
these effects.
5 CONCLUDING REMARKS
Exploratory model-scale tests of two air injection systems
for mitigation of merchant ship URN were performed, with
the aim of providing useful data for the design of quiet
vessels. A brief overview of the model tests was presented,
focusing in particular on how the URN measurements were
analysed to quantify noise abatement potential. These
results were then applied to measurement data for a real
ship to demonstrate how they can be used in practice and
what the expected reductions in URN could be. A method
for estimating the power requirement of both systems was
provided, since this is an additional design consideration.
For the Masker system an insertion loss of up to 22 dB
was measured. This was obtained for the highest model
speed and air flow rate used during the tests. In general,
the IL showed a strong dependency on frequency. In the
case of the Prairie-like system the maximum source level
attenuation was about 12 dB, which occured around the
centre frequency of the spectral hump caused by tip vortex
cavitation. When combining these data in the ship-scale
application case, the maximum change in radiated noise
level was limited to around 10 dB, due to the relative
contributions of the two source mechanisms.
The effect of limited or insufficient signal-to-noise ratio on
the measured changes in source level in the model tests was
discussed. Data fitting was required when addressing the
ship-scale application case.
The procedure for applying the model test results to
measurement data requires further development and for-
malisation; for example in how the SLA results should be
applied on a case-by-case basis. This will be the subject of
future work, which will also consider application to other
cases, for different ship types than used here.
ACKNOWLEDGEMENTS
This work was carried out as part of the ‘Developing Solu-
tions for Underwater Radiated Noise’ (SATURN) project:
https://www.saturnh2020.eu/. SATURN has
received funding from the European Union’s Horizon 2020
research and innovation programme under grant agreement
no. 101006443. We would like to thank our colleagues at
MARIN who were involved in preparing, performing and
analysing the model tests.
REFERENCES
Arveson, P. T. & Vendittis, D. J., 2000. ‘Radiated noise
characteristics of a modern cargo ship’. The Journal
of the Acoustical Society of America, 107(1), pp.
118–129.
Baudin, E. & Mumm, H., 2015. ‘Guidelines for regulation
on UW noise from commericial shipping (projects
AQUO-SONIC)’. Tech. rep., Bureau Veritas and DNV
GL. Publication Title: SONIC deliverable 5.4.
Bellmann, M. A., 2014. ‘Overview of existing Noise
Mitigation Systems for reducing Pile-Driving Noise’.
Proceedings of Inter-noise 2014. 16th-19th November,
Melbourne, Australia.
Bosschers, J., 2018. ‘A semi-empirical prediction method
for broadband hull-pressure fluctuations and underwa-
ter radiated noise by propeller tip vortex cavitation’.
Journal of Marine Science and Engineering, 6(2), p. 49.
Bosschers, J., Lafeber, F. H., de Boer, J., Bosman,
R. & Bouvy, A., 2013. ‘Underwater radiated
noise measurements with a silent towing carriage
in the depressurized wave basin’. Proceedings of
3rd International Conference on Advanced Model
Measurement Technology. 17th-18th September,
Gdansk.
Cruz, E., Lloyd, T., Lafeber, F., Bosschers, J., Vaz, G. &
Djavidnia, S., 2022. ‘The SOUNDS project: towards
effective mitigation of underwater noise from shipping
in Europe’. Proceedings of the Institute of Acoustics,
44.
Duarte, C., Chapuis, L., Collin, S., Costa, D., Eguiluz,
V., Erbe, C., Halpern, B., Havlik, M., Gordon, T.,
Merchant, N., Meekan, M., Miksis-Olds, J., Parsons,
M., Predragovic, M., Radford, A., Radford, C., Simp-
son, S., Slabbekoorn, H., Staaterman, E., Opzeeland,
I. V., Winderen, J., Zhang, X. & Juanes, F., 2021.
‘The soundscape of the anthropocene ocean’. Science,
371(583).
Dymarski, P., Salvatore, F., Pereira, F., Franchi, S. &
Falchi, M., 2011. ‘Results of model tests on reference
configurations.’ Tech. Rep. STREAMLINE-INS-DEL-
D21.1, INSEAN, Rome, Italy.
EU, 2017. ‘Commission decision (EU) 2017/848 of 17
May 2017 laying down criteria and methodological
standards on good environmental status of marine
waters and specifications and standardised methods
for monitoring and assessment, and repealing Decision
2010/477/EU’. Official Journal of the European Union,
125(May), pp. 43–74.
Hadler, J. B., English, J. W. & Gupta, S. K., 1984. ‘Program
to minimize propeller-induced vibration on converted
Maersk “E”-class ships’. SNAME, New York, USA.
IMO, 2023. ‘Revised guidelines for the reduction of
underwater radiated noise from shipping to address
adverse impacts on marine life (MEPC.1/Circ.906)’.
Tech. rep.
ITTC, 2017. ‘Recommended procedures and guidelines.
Model-scale propeller cavitation noise measurements’.
Tech. rep., International Towing Tank Conference,
Zurich, Switzerland. Publication Title: 7.5-02-01-05.
Joy, R., Tollit, D., Wood, J., MacGillivray, A., Li, Z.,
Trounce, K. & Robinson, O., 2019. ‘Potential benefits
of vessel slowdowns on endangered southern resident
killer whales’. Frontiers in Marine Science, 6, pp. 1–20.
Klinkenberg, Y., Bloemhof, F., Kamphof, H. J. & van
Rijsbergen, M., 2023. ‘Measuring air bubble layer char-
acteristics around the hull of a ship model’. Proceedings
of the 7th International Conference on Advanced Model
Measurement Technology for the Maritime Industry.
24th-26th October, Istanbul, Turkey.
Krueger, S., Friesch, J. & Stoye, T., 2004. ‘Reduction
of propeller induced pressure fluctuations by system-
atic air injection’. Proceedings of 9th International
Symposium on Practical Design of Ships and other
Floating Structures. Luebeck-Travemuende, Germany.
Lloyd, T., Foeth, E.-J., Lafeber, F. H., Bosschers, J.,
Klinkenberg, Y., Birvalski, M., Kaydihan, L. & Lidtke,
A., 2024. ‘SATURN Deliverable 4.3. Impact of mitiga-
tion sources on source level of a single-screw vessel’.
Technical report, Wageningen, The Netherlands.
Lloyd, T., Lafeber, F. H. & Bosschers, J., 2018. ‘In-
vestigation and validation of procedures for cavi-
tation noise prediction from model-scale measure-
ments’. Proceedings of 32nd Symposium on Naval
Hydrodynamics. 5th-10th August, Hamburg, Germany.
Lloyd, T., Lafeber, F. H., Bosschers, J., Kaydihan, L.
& Boerrigter, B., 2023. ‘Scale model measurements
of ship machinery noise mitigation by air injec-
tion’. Proceedings of 7th International Conference
on Advanced Model Measurement Technology for the
Maritime Industry. Istanbul, Turkey.
M¨
akiharju, S. A., Perlin, M. & Ceccio, S. L., 2012. ‘On
the energy economics of air lubrication drag reduction’.
International Journal of Naval Architecture and Ocean
Engineering, 4(4), pp. 412–422.
Technical Group on Underwater Noise, 2022. ‘Setting
of EU Threshold Values for continuous underwater
sound. Marine Strategy Framework Directive (MSFD)
Common Implementation Strategy’. Technical report.