In the underwater environment, it is all but silent. In most parts of the oceans, sunlight is barely available and thus marine animals have evolved to rely on sound for navigation, foraging and communication. Marine animals are not the only sources of underwater sound. Other natural sources such as earthquakes, waves and rain cause ambient noise, but also loud impulses. Furthermore, human activities consisting of dredging, surveying, construction and shipping cause loud noise in a wide frequency range. Underwater radiated noise (URN) from shipping has severe negative consequences on marine animals. The negative effects include auditory masking, stress, and behavioural and acoustic responses, possibly leading to collision with ships. It is evident that the mitigation of the underwater radiated noise of a ship is advantageous and worth researching.

In the past, plenty of research has been done into the modelling of the acoustic characteristics of vessels. A research gap was identified in the modelling of mitigation methods of the underwater radiated noise from ships with a focus on marine mammals. Up until now, most research has focused on the attenuation of vibrations on board a ship or radiated noise due to the propellers. At all times, the focus was either on human comfort or the radiated noise in general, however, for marine mammals certain frequency bands are of greater importance. It is valuable to assess the URN of a ship in the design phase, such that adjustments can be made to decrease the URN without excessive costs. The research goal of this thesis is:
Predict and mitigate the structure-borne underwater radiated noise of a ship in the design phase caused by onboard machinery.

The research goal and corresponding research questions are answered by first setting up the framework for the models. This framework sets the frequency analysis range to 20 - 200 Hz, formulates a reference ship case for validation of the data and gives the inputs and boundary conditions for the models. The acoustic metric of interest is set to the source level (SL) in dB re 1 μPa2m2.

Secondly, a simplified model is setup in Ansys 2021R2. This model is a 3D solid element model shaped like a beam. The equivalent beam (EB) model has similar global properties as the reference ship case. Around the EB model an acoustic domain is located that is modelled to represent an infinite domain. No physical boundary effects are included as the source level is per definition not dependent on this.
Subsequently, mitigation methods for machinery URN are researched and the resilient mount was found to be the most promising. A resilient mount is applied to the EB model in Ansys 2021R2 and source level spectra for different mount parameters are investigated.
Lastly, a two degree of freedom (2-DOF) schematisation is made that incorporates the Ansys model using a dynamic stiffness. The 2-DOF schematisation allows for faster computations of the complete model and thus a more extensive parameter study of the resilient mount is possible.

Over the frequency analysis range of 20 - 200 Hz, the results show that machinery structure-borne URN can be reduced by 45 - 65 dB re 1 μPa2m2. The reduction oscillates over the frequencies at lower frequencies. A linear SL reduction was observed from 60 Hz and above, which gradually lessened for higher frequencies. The SL reduction increased from 45 dB re 1 μPa2m2 to 60 dB re 1 μPa2m2 when the resilient mount damping ratio was changed from 0.18 to 0.02. In addition, the normalised resilient mount parameter study showed the system's parameter sensitivities and responses. It became clear that the resilient mount does not respond to a change in resilient mount damping as expected. The accuracy of the absolute results is subject to assumptions and limitations, which introduce uncertainties.

The absolute decrease in URN with the resilient mount was computed using acceleration input rather than force input. The acceleration input was found to have overestimated the 'no-mount' case, which was used to compare cases with the resilient mount. The total URN reduction with the resilient mount could thus have been overestimated. Furthermore, an effect of the model boundaries and the model domain size on the results was present in the models. The magnitudes of the results were influenced by this effect, which could not be eliminated due to the computational limitations reached. Finally, there was a scarcity of model input data and reference data. The magnitudes of the results were obtained and compared to limited data in order to determine the accuracy of the results.
Taking these limitations into consideration, the findings of this study should be interpreted with caution. The findings support literature claims that a resilient mount can reduce structure-borne machinery URN by 20 - 40 dB re 1 μPa2m2, with more reduction at higher frequencies.

The effect of the application of the resilient mount on marine mammals was hard to quantify. The structure-borne machinery URN is a part of the total URN of a ship. Due to the logarithmic relation of the noise, the reduction of one part could have very limited effects. Furthermore, the total soundscape in the ocean is formed by the combined noise of many ships. Moreover, the relation between the perceived nuisance of marine mammals and the URN levels is hard to indicate. The effect is undoubtedly positive but could be negligible in the bigger picture. At low speeds and close distances, the machinery URN is governing and the influence of URN from other ships is reduced. In those cases, the reduction of structure-borne machinery URN with resilient mounts could be expected to be the most positive for marine mammals.