Underwater radiated noise from marine vessels: A review of noise reduction methods and technology

1. Introduction

Underwater Radiated Noise (URN) from shipping has grown significantly over the past few decades due to increases in both the size and number of commercial vessels. In the past, URN has primarily been a concern for naval vessels, particularly those engaged in anti-submarine warfare or mine counter-measure vessels, and also some research vessels. However, as the detrimental impacts on marine life become clearer, there is a need to understand and reduce the noise produced by commercial and recreational vessels as well. While marine anthropogenic noise has long been recognised as a problem to marine life, early work focussed on the problems of high-intensity, short duration sounds such as those from sonars or pile driving, and much research has gone into understanding the impact of this on marine life (Weilgart, 2007). There is now an extensive body of research clearly showing the negative impacts that anthropogenic noise is having on marine life. This research has accelerated in recent years alongside calls to impose limits on the noise from human activityin the oceans (Williams et al., 2015). Commercial shipping is a major contributor to the rising levels of noise in the oceans, particularly at low frequencies (Hildebrand, 2009), and this has been found to have a detrimental impact on a wide range of marine life including mammals (Richardson et al., 1995, Nowacek et al., 2007), fish (Simpson et al., 2016, Mickle and Higgs, 2018) and invertebrates (Wale et al., 2013, Murchy et al., 2019). These studies highlight the urgent need to reduce noise from shipping to prevent further damage to marine life.

This is now widely recognised, as noted by Simmonds et al. (2014), but only limited action has been taken to-date. It is also recognised that in the absence of a concerted effort by governments, regulators, and shipbuilders, shipping noise is expected to increase in the coming decades (Kaplan and Solomon, 2016). As well as large commercial vessels, the role of smaller vessels and ferries is being increasingly scrutinised, with studies finding that the noise from smaller recreational vessels and ferries can dominate the soundscape in shallow coastal waters (Hermannsen et al., 2019, Cope et al., 2021).

As a result, a wide range of policy options are being proposed and considered for reducing commercial shipping noise (Markus and Sánchez, 2018, Merchant, 2019, Vakili et al., 2020). Changes to vessel routing, speed limits, and changes to vessel design are all being assessed and reviewed at national and international levels. Chou et al. (2021) compile a wealth of reports, journal publications and other documents from governments, inter-governmental bodies, and academic and research institutions related to underwater noise and its mitigation. It is highlighted here that underwater noise from marine vessels is not regulated appropriately at present, but that proper regulation combined with better technology could have a significant positive impact on the environment. There are several important international guidelines that support the reduction in underwater noise from shipping. The “EU Marine Strategic Framework Directive 2008/56/EC” of 2008 requires Member States to achieve a “good environmental status” including the reduction of URN, although the practical implications of this are not clear. Additionally, in 2014 the International Maritime Organisation(IMO) published MEPC.1/Circ.833 “Guidelines for the reduction of underwater noise from commercial shipping to address adverse impacts on marine life”. This document provides some guidance and good practice but does not propose any hard limits or regulations. Despite efforts to classify underwater radiated noise as a form of pollution, it is yet to be formally outlined in the IMO’s International Convention for the Prevention of Pollution from Ships (MARPOL). Since 2019 there has been a call from a number of member states to revisit the IMO guidelines for underwater radiated noise. This was initiated in January 2019, where a technical workshop brought together over 140 participants from around the world. The workshop pulled together best practice from industry and outlined mitigation techniques in the VARD report (Vard, 2019). The work from the event contributed to several papers submitted to IMO’s Marine Environment Protection Committee (MEPC) in the following years. Following MEPC76 in June 2021, the IMO formally accepted the proposal from Australia, Canada, and the United States to review the 2014 guidelines and work is now underway to update them.

The lack of overarching international regulation has not prevented some local schemes from being developed. Vessels entering the Canadian ports of Vancouver and Prince Rupert are eligible for reduced harbour fees if they meet certain noise and pollution criteria. Whilst schemes such as this are not widely implemented at the time of writing, many other governments and ports are reviewing their local environments. This includes creating underwater soundscapes to develop a clearer understanding of the local acoustic environment, such as that conducted at the Port of Gothenburg (Lalander et al., 2021).

The internal noise and vibration within a ship is highly regulated due to their health and safety implications, and so standardised definitions and measurement procedures exist for this. The standardisation of external radiated noise is, on the other hand, far less developed. The main areas of standardisation so far have been focused on the consistent measurement of underwater noise including ISO 17208 “Underwater acoustics — Quantities and procedures for description and measurement of underwater sound from ships” and ANSI/ASA S12.64 “Standard Quantities and Procedures for Description and Measurement of Underwater Sound from Ships”. An important area of future development is the standardisation of URN measurement in shallow water. Measurements in shallow water are difficult to perform because of the complex interactions between the acoustic waves and the seabed and sea surface. Work is ongoing to develop a Section 3 of the existing ISO 17208 standard to look at shallow water measurements (Ainslie et al., 2021).

One of the main challenges to bringing in regulations such as mandatory noise limits is the lack of data. Regulations analogous to the Energy-Efficiency Design Index (EEDI) are often discussed, but this would require a much better understanding of the noise levels produced by both new and in-service ships. Acoustic trials are not common and seldom carried out for commercial vessels, meaning we have only a limited idea of how much noise most vessels actually produce. A number of research programs have sought to address this issue, including the EU-funded AQUO project, Hallander et al. (2015) and Audoly et al. (2015). This work has given rise to parametric models for the level of noise produced by a given ship, but further data is needed to improve and test these models across a broader range of vessels. There is also significant uncertainty in the efficiency and cost effectiveness of the mitigation measures as well as a lack of a harmonised and standardised methodology for underwater radiation noisemeasurement (Vakili et al., 2020).

Most classification societies now offer a Quiet Ship Notation to help guide and verify the low signature credentials of a vessel. Certification is normally outlined through 3rd octave target sound pressure levels and tested through the use of full-scale trials, although it is acknowledged that this measurement process is fraught with challenges of consistency and standardisation. Efforts are now being made to compare the different notations across the different class societies and also to harmonise some definitions and procedures (Ainslie et al., 2022). All of the notation targets broadly follow the same pattern for noise reduction linked to scientific research on marine life hearing ranges: the target curves will typically create a line or “V” shape curve between 10 Hz and 1000 Hz (recognising the critical hearing ranges for marine life), before tapering off after 1000 Hz in recognition of higher dissipation rates at higher frequencies. However, it should be noted that some disagreements remain about how to classify vessels and what the exact targets should be for each vessel type.

As well as the design of a vessel and the technologies used, operational conditions and speed are important determining factors for radiated noise levels. Recently, the Vancouver Fraser Port Authority published results of a study where vessels were asked to reduce speed to 11 knots (MacGillivray et al., 2019), with a view to reducing to levels of noise produced. This showed significant reductions in the levels of URN recorded, showing that it could be an effective measure. However, a more recent study has shown that some vessels actually produce more noise at lower speeds (McIntyre et al., 2021). Despite the general trend supporting the idea that slower vessels produce less noise, this study highlights the need to understand the physics of noise generation from different vessels, and the importance of developing better designs and technology to limit the noise produced. The use of speed reduction as a means of reducing noise was also explored by Tani et al. (2015). In this work, two mechanisms used for speed reduction were compared: reducing propeller revolutions per minute (RPM) at constant pitch, and reducing pitch at a constant RPM. The study showed that reducing speed via an RPM reduction was indeed highly effective at reducing the levels of URN, but the same was not true for a pitch-reduction at constant RPM. In fact, reducing the pitch was shown to cause pressure-side cavitation, which in some cases led to an increase in the overall levels of noise. This again highlights the complex relationship between the operating condition of a vessel and the radiated noise levels, particularly when considering the different propulsion architectures used.

URN from a ship arises from multiple sources, leading to a complex signature covering a wide range of frequencies. Descriptions of these sources can be found in many works, for example Abrahamsen (2012). A number of research programs have greatly enhanced our understanding of the make-up of ship radiated noise, including the large collaborative EU projects (SONIC, SILENV, AQUO). Such programs have improved our understanding of cavitation noise, machinery noise, and other components, but they have also highlighted the large differences that occur between vessel types and operating conditions. This provides further weight to the argument that more data is needed across a broader range of ship classes. The signature is often broken down into machinery noise, flow noise, and propeller noise which includes cavitation noise (Rizzuto et al., 2015). There are numerous published works of full-scale trials and other experimental studies that demonstrate the levels of noise produced by different vessels, and these also provide details on some of the dominant components (Arveson and Vendittis, 2000, McKenna et al., 2012, Li et al., 2018). These studies tend to show that low frequency noise (<125 Hz) is dominated by tonal components associated with the main engine firing rates and the propeller blade-rate harmonics. Higher frequency noise is made up of machinery noise, propeller and flow noise, and cavitation, but breaking this down into its constituent components is challenging. This leads to uncertainty as to the relative importance of the different components and further research is needed to better understand this. However, by targeting each of the individual sources, significant reductions in the overall noise produced should be achievable. Furthermore, by considering the sources individually and examining their physical origins, designs and technology can be developed and applied more successfully to reducing the overall radiated noise levels.

This paper provides a review of the sources of underwater radiated noise from marine vessels, together with a wide range of design methods and technologies that might be applied to reduce it. As well as reviewing existing technologies and design techniques, this work also includes an extensive review of new and novel technologies from a range of industries to see how these might be applied to reducing underwater radiated noise. This includes technology and design innovations for reducing propeller cavitation, passive design modifications to reduce leading and trailing edge noise, as well as meta-materials that could be used to reduce acoustic propagation across a range of applications. The technology considered here is drawn from a range of industries and different areas of research, with a view to assessing how they could successfully be applied to reducing URN from marine vessels. Not all the technology considered has been developed for noise reduction, but for efficiency gains or flow control. However, by understanding the underlying physics of how such devices work, their application to reducing particular sources of noise becomes clear. Some of this technology is in its infancy and would require extensive research and development before it could be applied at scale on a ship. However, much of the technology discussed has been shown to be effective by multiple experimental and numerical studies, and some is now being used within industry, although not necessarily on marine vessels. The maturity of each of the technologies and devices as applied to a marine vessel is discussed, together with the research and development needed to progress the maturity of the technology.

Beginning with a review of noise sources from a marine vessel in Section 2, a review of design methods and technologies that can be adopted to reduce noise is presented. Reducing propeller cavitation, non-cavitating propeller noise, and flow noise is considered in Section 3, and reducing machinery noise is considered in Section 4. Conclusions are presented in Section 5, along with recommendations for future work, with a particular focus on the research and development needed to increase the technology readiness levels of the technologies discussed.

2. Source of URN from marine vessels

The total URN of a vessel is often split up into machinery noise, propeller and flow noise, and cavitation (see Fig. 1). The contribution of each component will differ from one vessel to the next and is also strongly dependent on the vessel’s speed. Studies have shown that the signature tends to be dominated by machinery noise at low speed, with cavitation becoming the dominant source at higher speeds (Moreno et al., 2015).

Machinery noise contributes to the URN of a vessel by inducing vibration of the hull plating and by transmitting sound through openings that are directly connected to the sea. There are multiple paths by which the sound and vibration from machinery radiates into the surrounding water. The first of these is where vibrations transmit through the machinery mounts into the hull structure and then into the shell plating. The airborne transmission of sound from the machinery inside the ship also induces vibrations of the structure, which transmits to the shell plating. A description of this can be found in Spence and Fischer (2016). Low speed diesel engines produce low frequency tonal noise associated with the engine firing rates and this can be readily identified during acoustic trials, see Arveson and Vendittis (2000) for example. Other machinery including generators, pumps, and heating, ventilation, and air conditioning (HVAC) systems will all contribute to the overall sound radiated into the surrounding water. The level of radiated noise resulting from onboard machinery is therefore very specific to the type and size of the vessel, as well as the mounts used for individual engines, generators, etc.

Propeller noise is often split into cavitating and non-cavitating, with cavitation noise tending to dominate at higher speeds (Salinas and Moreno, 2014). For a non-cavitating propeller, studies have shown that tip vortex noise is the dominant source if the propeller is operating in a uniform flow (Ianniello et al., 2013, Keller et al., 2018). Propellers also produce trailing edge and leading edge noise, as well as noise from the hub. Trailing edge noise results from the turbulent boundary layer convecting over the trailing edge, where the fluctuations scatter as acoustic waves. This can be broadband or narrowband, depending on factors such as the Reynolds number and the thickness of the boundary layer relative to the edge thickness. This can also lead to propeller singing, where a fluid–structure interaction leads to a high-pitched tonal sound being produced. This occurs due to a matching of the dominant vortex sheddingfrequency and the natural frequency of the blade, and details of the physical mechanisms of this can be found in Blake (2017) and Fischer (2008). A detailed explanation of trailing edge noise together with a range of experimental data and numerical methods can be found in Blake (2017) and Brooks et al. (1989). Leading edge noise results from unsteady and turbulent flow interacting with the leading edge of a lifting surface, which induces pressure fluctuations and hence noise.

Low frequency tonal noise occurs as a result of the propeller rotating through a spatially-varying wake field which can lead to cavitation as well as non-cavitating noise. A typical wake is shown in Fig. 2 together with the axial velocity at a non-dimensional radius of $r / R = 0.7$ . The variations in the axial velocity cause the local flow velocity and angle of attack to vary as the blades rotate, leading to a fluctuating pressure field. This induces dipole-type noise and, if cavitation is present, fluctuating cavity volumes on the surface of the blades.

At higher speeds, propeller cavitation is one of the main sources of URN and research into reducing it has been ongoing for many decades. For commercial vessels, the cavitation inception speed can be as low as 8 knots (Leaper et al., 2014) and so it is likely that most commercial vessels will experience cavitation. Cavitation noise is complex and arises in many different forms depending on the propeller and hull design, as well as the operating conditions. Low frequency cavitation noise tends to be due to attached cavitation bubblesoscillating as the propeller rotates through a spatially varying wake field, whereas detached cavitation and vortex cavitation tend to increase noise levels at higher frequencies. Vortex cavitation can occur at the tip and the hub. Tip vortex cavitation is often the first to occur and arises due to the low pressure in the tip vortex core. This type of cavitation is very sensitive to the particular dynamics of the tip flow and also the water quality, which refers to the size and distribution of nuclei in the water (Arndt et al., 1991, Chen et al., 2019). Therefore, both propeller design and operating conditions are important determining factors here. Hub vortex cavitation occurs when the pressure in the hub vortex drops below a certain point, which is again dependent on the water quality as well as the dynamics of the vortex itself.

Sheet and bubble cavitation can occur when the pressure drops below the cavitation inception point at, or close to, the propeller blade. This will typically occur on the suction side, but can also occur on the pressure side in off-design conditions or for controllable-pitch propellers operating at reduced pitch. Bubble cavities originate at the blades but travel with the flow, and have historically been associated with high noise levels and erosion. However, research in the 1990s showed that, at full-scale, bubble cavitation was likely less of an issue than previously thought, and it was suggested that sheet and cloud cavitation should be given more attention (Kuiper, 1998). Unstable sheet cavitation leads to cloud cavitation, which tends to lead to a large increase in the radiated noise levels. The stability of sheet cavitation and the dynamics that lead to cloud cavitation have been studied in many works, see Pham et al. (1999) for example. Sheet cavitation is found to increase the noise levels at blade rate harmonics and also at higher frequencies, where the increase in broadband noise is due to splitting and collapse of bubbles (Salinas and Moreno, 2014). Noise is produced by growth, oscillation and collapse of cavitation bubbles, as well as by splitting and coalescence. It has been shown that cavitation bubble collapse can produce noise several orders of magnitude higher than the noise associated with growth, oscillation, splitting (Choi and Ceccio, 2007). More detailed expositions of the physics and types of cavitation, as well as cavitation noise can be found in Arndt, 2002, Franc and Michel, 2006, Foeth et al., 2008 and Blake (2017).

As well as noise from the propeller, the flow over the hull and appendages will also produce noise, and this generally becomes more prominent at higher speeds. Both leading and trailing edge noise will be produced by appendages such as the rudder, A-brackets and stabiliser fins. Interaction noise will also be important here. This arises when turbulence from the flow over one structure interacts with another. For example, the unsteady turbulent flow from a propeller will interact with the rudder, inducing pressure fluctuations on the rudder surface. Hydrofoils will produce trailing edge noise, tip vortex noise, and may also cavitate. The dynamics of hydrofoil cavitation are discussed in a great many works, for example Blake et al., 1977, Huang et al., 2013 and Blake (2017).

Noise from openings arises primarily from two sources: pumps used within systems directly connected to the sea, and from the flow over the openings and sea chests. Noise and vibration induced by flow over sea chests has been noted as a problem on several vessels (Carlton and Vlasic, 2005), and is most likely to affect those operating at higher speeds. Lateral thrusters may also be categorised in this way, with the thruster producing noise when in operation as well as noise from the flow over the opening. The issue of lateral thruster noise has received more attention recently, both due to onboard noise and URN (Reinikainen, 2021).

3. Reducing propeller cavitation and non-cavitating propeller noise

In this section, technologies and design methods for reducing propeller cavitation, non-cavitating propeller noise, and flow noise are considered. The origins of many sources of non-cavitating propeller noise share much in common with the cavitating sources, and so similar technologies, devices and design methods are often used to reduce both. For example, devices that reduce the onset of tip vortex cavitation do so by reducing the vortex strength, which should also reduce the noise produced by the non-cavitating vortex. Similarly, devices that improve wake uniformity will potentially reduce sheet cavitation(if present) as well as the non-cavitating component of the blade-rate noise.

3.1. Sheet, bubble, and cloud cavitation

Whilst sheet, bubble, and cloud cavitation are different phenomena, there are commonalities in their origins and so it is helpful to consider them together when reviewing designs and technology for reducing their presence. Efforts to reduce these types of cavitation have focussed on the propeller blade area ratioand the section design. Increasing the blade area ratio allows for the propeller blades to have a reduced load per unit area while maintaining the overall thrust. This, in turn, results in a higher minimum pressure on the suction side, thus reducing the propensity for cavitation. This has been understood for many years, following studies by Burrill and Emerson (1963) and others.

For section design, it is preferable from a cavitation perspective to design sections that have a more even pressure distribution along the chord length, such as a round-back section. Wing section profiles such as the NACA 4-digit series have a very favourable lift-to-drag ratio, but they also have a very uneven pressure distribution, with most of the lift being generated shortly downstream of the leading edge. By designing sections with a more even pressure distribution, it is possible to generate the same amount of lift along the chord length but have a higher minimum pressure, which can alleviate or at least reduce the levels of cavitation for a given speed. This is now common design practice for modern propellers. The development of sections with improved cavitation performance can be found in a number of works, for example Dang (2004) and Zeng and Kuiper (2012), and this has been effective at increasing the cavitation inception speed for many vessels.

As well as the propeller design, one should also consider the local flow field that the propeller is operating in, as this can have a large influence on the presence and dynamics of cavitation. As discussed in Section 2, the spatial variations in the nominal wake field lead to pressure fluctuations on the propeller blades, which in turn can lead to oscillating cavity volumes. Therefore, improving the level of wake uniformity will reduce the magnitude of these oscillations, and potentially delay the onset of this type of cavitation. The design of the hullform is crucial here, and efforts should be made from an early stage in the design to ensure as uniform a flow field as possible in the propeller plane. Improving wake uniformity can also improve propulsive efficiency, and a wealth of research has gone into this over the past 20 years. One such device that can improve the wake uniformity is the vortex generator. These devices have been applied widely for controlling boundary layer separation (Lin, 2002), reducing vortex-induced vibration (Xin et al., 2018), and also improving the uniformity of the flow into a ship propeller (Kim et al., 2015). Depending on the application, vortex generators can be used to promote local mixing, prevent separation, or to re-structure the flow field (Jirasek, 2005).

Applied to flow control at the stern of a ship, vortex generators can be used to transfer momentum towards the top of the propeller plane, increasing the axial velocity and reducing flow separation. This was demonstrated by Li et al. (2021), who showed that placing a vortex generator on the hull upstream of the propeller reduced flow separation at the top-dead-centre location, which led to a decrease in the amplitude of the pressure fluctuations at the blade-rate harmonics. Combined with a high-skew propeller, see Ji et al. (2014) for example, such technology has the potential to dramatically reduce low frequency propeller noise as well as cavitation. As well as being a mature technology within the aerospace industry, vortex generators have now been successfully trialled on full-scale ships, indicating a high level of maturity. Saydam et al. (2018) used CFD to design vortex generators which were then retro-fitted to a tanker (see Fig. 3). Analysis carried out on the tanker indicated a significant reduction in vibration at the blade-rate harmonics, indicating that the vortex generators were successful in improving the wake uniformity on a full-scale ship.

3.2. Tip vortex cavitation

Tip vortex cavitation (TVC) has received much attention as it is often the first type of cavitation to appear on a modern propeller. This is due to improved propeller section design over past decades which has led to the inception of sheet cavitation being delayed, and it is now more common for tip vortex cavitation to appear first (Kuiper et al., 2006, van Terwisga et al., 2007). As a result, extensive research has gone into understanding, predicting, and suppressing tip vortex cavitation. Common design methods to delay the onset of tip vortex cavitation are often based on tip off-loading (Molland et al., 2017, Park et al., 2014), which reduces the pressure differential between the two sides of the propeller close to the tip, thus reducing the tip vortex strength. This tends to reduce the efficiency of the propeller, as the lift-to-drag ratio of the propeller surface close to the tip is decreased. Novel propeller designs, such as the contracted and loaded tip (CLT) propeller (Gaggero et al., 2016) have been shown to reduce the presence of tip vortex cavitation and these have now been installed on a range of vessels (Ebrahimi et al., 2019).

A number of active and passive techniques have also been proposed, either for suppressing TVC or reducing the noise produced. One such active method is to inject either water or a polymer solution into the vortex core, which increases the core radius and hence increases the pressure in the vortex core (Chang et al., 2011, Park et al., 2014). The use of polymer injection for foils and propellers was first used in an attempt to reduce drag (Brennen, 1970) but it was subsequently found that it might be successful in delaying the onset of TVC. Chahine et al. (1993) showed that injecting a polymer solution at the tip delayed the onset of cavitation compared to injecting water although a more recent study by Lee et al. (2018) showed that water injection could be effective. Studies have shown that injection of polymer solutions and dilute solutions can both suppress tip vortex cavitation, but that the mechanisms are different for the two cases. In one such study (Fruman and Aflalo, 1989), it was found that a homogeneous polymer solution inhibited TVC by reducing the pressure differential on the two sides of a foil, thus reducing the circulation at the tip. Injection of a dilute solution led to a much smaller change in the lift, but altered the tangential velocity field in the core, which led to a delay in cavitation inception.

An alternative but related method is to attach a flexible fibre to the tip which is drawn into the vortex core. Studies on water and polymer injection have found that the results are highly dependent on the location of the injection port. This leads to issues around repeatability and potentially limits the conditions under which the technique would be effective. Park et al. (2014) instead propose attaching a flexible thread to the tip of the propeller blades, which is sucked into the vortex core. This was found to be effective in suppressing TVC and may be a more robust approach than water or polymer injection. Fig. 4 shows how effective this can be, with the thread completely suppressing the tip vortex cavitation in certain conditions. This idea is further investigated by Amini et al. (2019b). Here, a flexible thread is shown to be most effective in suppressing TVC when its diameter is similar to that of the viscous vortex core. The thread is periodically pulled into the vortex core where it oscillates, thickening the vortex core and suppressing the cavitation. These studies also suggest a negligible impact of the thread on the propeller efficiency.

Whilst these techniques have been shown to be effective in laboratory experiments, they are not yet used on full-scale vessels and further research is needed to demonstrate their effectiveness in realistic environments. One of the main appeals of using a thread instead of injection of water or a polymer into the flow is the ease of implementation on a full-scale vessel. Active methods such as fluid injection increase cost and complexity, and this may limit their feasibility compared to a more passive technology such as a flexible thread. Furthermore, depending on the particular molecular structures used, injection of a polymer may be detrimental to the environment.

As well as active methods, passive methods have also been developed for reducing or suppressing TVC. It has been shown both experimentally and numerically that surface roughness can be used to suppress TVC (Asnaghi et al., 2020, Svennberg et al., 2020, Krüger et al., 2016). These studies showed that roughness modifies the vortex roll-up process leading to a weaker vortex forming, with a higher core pressure than is observed for a smooth blade. Clearly, there is a trade-off here between vortex suppression and the efficiency of the propeller which is degraded by the presence of surface roughness. Asnaghi et al. (2021) found that by positioning the roughness only at the tip and a small area on the leading edge, TVC mitigation was achieved with a modest 1.8% reduction in propeller efficiency.

One measure that has its origins in aerospace engineering is the use of wingletsto reduce the vortex strength. The addition of winglets to aircraft wings for reducing trailing vortices has been widely research, for example Lee and Su (2012) and Lee and Choi (2015). Recent studies have shown that such technology can be successfully applied to marine propellers (Amini et al., 2019a, Gao et al., 2019). Designed appropriately, the winglets have been shown to decease the core axial velocity and increase the core radius, thus increasing the core pressure and preventing, or at least delaying TVC. However, the winglets are likely to generate their own noise and further research is needed to understand how they would affect the overall sound pressure level at different operating points. There are clear similarities in the mechanisms these different techniques invoke. In particular, winglets, the injection of water or polymer solutions, and the use of flexible threads all act, at least in part, to increase the vortex core diameter and so increase the pressure.

An alternative approach is to place holes in the propeller tip to reduce the pressure differential on the two sides. Using both experimental and numerical methods, Aktas et al. (2020) showed that locating holes in specific places close to tip modified the nature of the cavitation and reduced the overall sound pressure level but with a loss of efficiency of 2%. One potential issue with technology of this nature comes from the environment in which it operates. Propellers are susceptible to the build-up of marine fouling, and this can occur quite rapidly. It is quite probable that this will block the pores used to alleviate the tip pressure differential and reduce the effectiveness of this over time. Further research could examine this to better understand the viability of this technology and also try to develop mitigation measures to ensure its continued effectiveness.

3.3. Hub vortex cavitation

Whilst not as large a contributor to the overall sound level as cloud or bubble cavitation, hub vortex cavitation may still be important, particularly if other forms of cavitation are not present. It is therefore worth exploring technologies that can reduce hub vortex strength and potentially suppress hub vortex cavitation.

Energy-saving devices have been developed that may also act to suppress, or at least reduce hub vortex cavitation and associated noise. Propeller boss-cap fins have been shown to be an effective energy-saving device that work by recovering energy from the hub vortex. This, in turn, reduces the strength of the vortex, which will reduce associated noise and help to reduce or remove hub vortex cavitation. These are a mature technology and have been applied to many in-service vessels (Ouchi et al., 1989, Hansen et al., 2011). The effect of a boss cap fin on hub vortex cavitation can be seen in Fig. 5, reproduced from Sun et al. (2016). This shows a propeller in a cavitation tunnel with and without a boss cap fin, showing the complete removal of the hub vortex cavitation. A wealth of literature is available on these devices, with both numerical and experimental methods being successfully used to design and optimise them (Kawamura et al., 2012, Seo et al., 2016, Mizzi et al., 2017). However, only a few studies quantitatively consider the impact of removing hub vortex cavitation on noise and so further research is needed to understand the overall impact. Tachikawa et al. (2019) do consider this, and find that removing hub vortex cavitation reduces the overall sound pressure level by 6 dB alongside a modest improvement in efficiency for the case considered. Several other devices have been developed to remove the hub vortex, often by integrating the propeller boss into the rudder (Okada et al., 2015, Su et al., 2020). Again, these devices have been developed for energy-efficiency, but it would be of interest to see if these devices can have an appreciable impact on the noise levels.

It should be noted that the noise associated with hub vortex cavitation is likely to be substantially less than that produced by cloud or bubble cavitation, if present. In this case, the effectiveness of removing the hub vortex cavitation on the overall noise levels is likely to be minimal. However, if other measures have been taken to limit these forms of cavitation, then eliminating a cavitating hub vortex is an important step to further reducing the overall noise levels.

3.4. Noise reduction through reduced propeller loading

In general, the more heavily loaded a propeller is, the more noise it will produce. A heavily loaded propeller is associated with lower pressure on the back of the blade, which increases the levels of cavitation. Higher propeller loading will also lead to a higher pressure differential between the pressure and suction sides, which increases the strength of the tip vortex. Dipole-type noise, which results from pressure fluctuations on the blades, scales according to $p_{a} \propto U_{r}^{6}$ , where $p_{a}$ denotes the acoustic pressure and $U_{r}$ is the resolved velocity in the co-ordinate system of the propeller blade section. Therefore, all else equal, a propeller rotating at a higher speed will produce more noise. Reducing the rotation speed of a propeller is seen as a practical means of reducing noise and so energy-saving devices (ESDs) that can provide either additional thrust or improve the propulsive efficiency should be considered as part of this. There is an important caveat here: if the propeller thrust is reduced by way of pitch reduction, as is common practice for a controllable-pitch propeller, then noise reduction may not be realised due to the propeller operating in an off-design condition. Controllable-pitch propellers are susceptible to pressure-side and root cavitation when operating at reduced pitch. The importance of this was highlighted by Tani et al. (2015), who found that a speed reduction achieved through a pitch reduction could lead to an increase in the radiated noise levels due to the presence of pressure-side cavitation. Therefore, from a noise perspective, energy-saving devices should be seen as a mechanism to reduce the propeller rotation rate, as this has the most potential to reduce the radiated noise levels.

A wide range of devices have been developed in order to improve the vessel propulsive efficiency. These devices have become more prominent in recent years following the introduction of the Energy Efficiency Design Index, and a wealth of research and development has gone into their design and implementation on a wide range of vessels. Whilst the correlation between energy efficiency and noise reduction is not always as strong as one would like (for example blade tip offloading reduces TVC but at a cost in terms of efficiency), there are Energy Saving Devices that may act to reduce the acoustic signature as well.

Many of the novel devices being developed and applied to commercial vessels, such as Flettner rotors, kites, and turbo-sails (Chen et al., 2010, Traut et al., 2014) provide thrust and so can be used to reduce the loading on the propeller. If this technology is used in conjunction with a reduction in propeller speed, it should reduce the levels of cavitation and hence reduce the noise. The extent to which noise levels will be reduced will depend on the particular vessel and the prevailing operating conditions, but if such devices can reduce the propeller loading such that it ceases cavitating, then substantial benefits may be realised in terms of URN.

Other devices designed to increase propulsive efficiency could also lead to a decrease in the noise produced by the propeller, particularly if the propeller is cavitating. Examples of such technology include pre-swirl stators (Zondervan et al., 2011, Park et al., 2015) and pre-swirl ducts (Shin et al., 2013) which modify the flow field into the propeller in order to improve the efficiency. However, many of these devices will also generate their own noise, and so their suitability as a tool for overall noise reduction should be assessed considering the whole system, not just the impact on the propeller.

3.5. Leading and trailing edge noise

When compared with the noise levels from cloud and bubble cavitation, leading edge and trailing edge noise from propeller blades and other lifting surfaces may not seem like a priority. However, there are instances where such components can be problematic and where efforts should be made to minimise them. Leading and trailing edge noise occur for all lifting bodies operating in turbulent flow. Within the context of a marine vessel, this includes the propeller blades, the rudder, A-brackets, stabiliser fins, as well as rotating machinery on board the ship such as pump impellers and fans in HVAC systems. For vessels that are not experiencing cavitation, either by design or speed, and for those with particularly strict noise requirements such as research or naval vessels, further reductions from fluid-induced can still be achieved by minimising the noise produced by lifting bodies.

One particularly troublesome problem related to the flow over the trailing edge is propeller singing. This can occur when the frequency of the vortex sheddingat the trailing edge is close to the resonant frequency of the blade (Park et al., 2005, Fischer, 2008). The phenomenon has been recognised for many decades, and early studies from the 1960s (Van de Voorde, 1960) considered the possible origins and solutions. The most common treatment for this problem is to use an anti-singing edge, which usually takes the form of a bevelled trailing edge. The effectiveness of this has been known for many decades, but the problem of propeller singing still occurs on many vessels, see Bonney and Bahtiarian (2006)and Fischer (2008) for example. As well as a bevelled edge, many of the developments discussed in this section, particularly the use of wavy and serrated edges, may also help to alleviate propeller singing.

Much inspiration for reducing noise from lifting surfaces can be found in the natural world, and a range of bio-mimetic designs have been developed and analysed numerically and experimentally. Designs have been developed to reduce both leading and trailing edge noise, which is produced by propeller blades and appendages on marine vessels. One such area where biologically-inspired designs have been applied is in the suppression of trailing edge instability noise. This occurs at moderate Reynolds numbers and afflicts wind turbines, gliders and other small aerial vehicles as well as some rotating machinery (Yakhina et al., 2015). Marine propellers and appendages typically operate at very high Reynolds numbers, where instability noise is unlikely, but there are important lessons that can be learnt from the research into instability noise. Specifically, one of the conditions required for this type of noise is a strongly span-wise coherent boundary layer where two-dimensional tubular vortices scatter at the trailing edge. Effective measures to reduce instability noise often work by breaking up these structures before they can scatter as acoustic waves. One of the most widely researched technologies for reducing this type of noise is the serrated trailing edge, which helps to break up these structures as they scatter (Jones and Sandberg, 2012, Chong and Joseph, 2013, Hu et al., 2022).

Despite being primarily designed to reduce instability noise, serrated edges can be applied to high Reynolds number marine applications including appendages and propeller blades. In such applications, the boundary layer will be fully turbulent and instability noise is highly unlikely to occur. However, as the Reynolds number increases, the boundary layer thickness reduces relative to the thickness of the trailing edge. Thus, beyond a certain Reynolds number, the edge becomes hydrodynamically blunt, and a narrowband component will re-emerge, associated with coherent shedding. Because serrated edges reduce the coherence of the scattered acoustic waves, they should also be effective here. The use of trailing edge serrations was considered for a turbulent boundary layer by Moreau and Doolan (2013). Here, experimental studies showed that the serrations reduced the noise across a wide range of frequencies but were most effective at the frequency range associated with the vortex shedding due to trailing edge bluntness. The relationship between the size of the serrations relative to the Strouhal number of the vortex shedding was found to be important for maximising the reduction in acoustic intensity.

We can therefore conclude that the physical mechanism by which the serrations reduce noise is similar, irrespective of the flow regime. That is, if designed appropriately, serrated edges act to prevent coherent turbulent structures from scattering as efficiently and so reduce the noise produced. There is likely to be an additional benefit to the use of serrations that can be seen when considering interaction noise. If they are used on an upstream geometry, they will break up the larger turbulent structures into smaller structures where the energy is distributed over a wider range of frequencies. Thus, the turbulent flow that interacts with the downstream geometry will be less coherent with many different scales and phase relationships, leading to less noise being produced. Furthermore, the break-up of larger structures into smaller structures accelerates the process of turbulent decay, as smaller structures are more affected by viscous dissipation.

A recent study used CFD to assess how serrated edges could be applied to a duct to reduce pump-jet noise (Qin et al., 2019). This was partly inspired by research chevron nozzles that have been shown to reduce jet noise for some types of turbofan engines (Callender et al., 2005, Bastos et al., 2017). Jet noise is not prevalent for marine propulsion systems as the velocity gradients are too small, but the study showed that a “sawtooth” duct resulted in a modest noise reduction of between 2.5 and 5 dB compared to a conventional duct. Two mechanisms were identified that led to this reduction. Firstly, the turbulent kinetic energy was found to be lower in the wake for the pump-jet incorporating the sawtooth duct. Secondly, the turbulent structures in the wake were found to be smaller and broke down more rapidly. This second mechanism is the same as that identified from the studies of trailing edge instability noise in that the sawtooth duct acts to break up larger coherent turbulent structures into smaller, less coherent structures.

Serrated edges have also been combined with other modifications. Wang et al. (2017) used large eddy simulations together with a Ffowcs-Willams and Hawkings model to assess the noise reduction from a foil with surface ridges and trailing edge serrations that are based on structures found in the natural world. A decrease in both tonal and broadband noise is seen compared to a normal foil geometry with minimal change in the drag. The mechanism for tonal noise reduction appears to be similar to that seen when using serrations. That is, the surface ridges and trailing edge serrations break-up the two-dimensional tube vortices into three-dimensional vortices.

Much of the research into biomimetic structures for noise reduction is inspired by the structure of owl feathers, which has been a source of inspiration for noise reduction since the 1930s, when Graham (1934) discussed the possibility of using owl feathers as a guide to designing quieter aircraft. A study by Chen et al. (2012) found that the leading edge serrations and trailing edge fringes on owl feathers can reduce the pressure fluctuations, and hence the noise, over the wing. Furthermore, it was also suggested that the porous grid structure over the feather also contributes to sound suppression. Further analysis of the structure of owl feathers and how they suppress sound can be found in Bachmann and Wagner (2011) and Wagner et al. (2017).

This work has inspired both experimental (Clark et al., 2016) and numerical (Wang et al., 2021) studies where airfoil designs are modified to try and mimic some of the feather’s features in order to reduce the trailing edge noise. Fig. 6, Fig. 7 show a detailed view of some of the features of the feather that are thought to suppress noise and Fig. 8 shows some of the designs that have been developed to mimic certain features. These designs have all led to modest reductions in airfoil self-noise and could potentially be applied to the reduction of propeller blade trailing edge noise and noise from other rotating machinery such as within HVAC systems.

Another approach, inspired in part by the porous structure of an owl feather, is the use of porous materials over part or all of the geometry (Geyer et al., 2010a, Geyer et al., 2010b, Carpio et al., 2019). Porous materials have also been considered computationally, for example Ananthan et al. (2020) where a hybrid LES/APE approach is used to look at trailing edge noise reduction by way of a porous trailing edge. This study identified two mechanisms by which the porous trailing edge reduced the sound pressure level: firstly by reducing the span-wise coherence of the turbulent structures and secondly by reducing the convection velocity of the eddies.