Scottish Marine and Freshwater Science Volume 4 Number 5: Modelling of Noise Effects of Operational Offshore Wind Turbines including noise transmission through various foundation types

This report presents modelling of the acoustic output of operational off-shore wind

turbines and its dependence on the type of foundation structure used.

6 Discussion

6.1 Assumptions made and their effects on results

6.1.1 Assumptions made in numerical modelling

The modelling presented here was based on a generic 6 MW wind turbine, with excitation force, geometries and material properties representative of parameters and designs currently in use. The use of a generic wind turbine was to allow the direct comparison of acoustic output of different foundation types; in doing so avoids the complication of different vibration characteristics being produced by different turbine makes and models. The models presented here are appropriate for comparison of foundation structures. To accurately model the magnitude of SPL the models need to be calibrated using either on tower vibration data, marine based sound measurements, or a combination of both.

Throughout the modelling process a number of assumptions were made. The key ones being:

1. Biofouling: Over time encrusted organisms can increase the level of damping of the structure by acting as a granular aggregate. The rate of biofouling is difficult to constrain and has been omitted for simplicity. The increase in damping related to biofouling would reduce the SPL from those presented here.

2. Cylindrical Spreading: Cylindrical boundary conditions are applied to allow the sound waves to propagate beyond the domain boundary. The boundary condition uses a simplification where the source of the cylindrical wave is taken to be co-linear with the vertical axis at the foundation centre. Given the vertical linear nature of the source of noise and the shallow water, the author feels that this assumption is justified.

3. Aerodynamic Noise: Aerodynamic noise travels from the blades and generator, through the surrounding air to the interface between the air and water. Due to the large impedance contrast between air and water any aerodynamic noise is almost entirely reflected. The marine environment is modelled without any contribution from aerodynamic noise.

4. Hub Rotational Speed: In the model the wind turbine rotational speed is kept constant when applying the excitation forces. In reality the rotation speed increases with wind speed. However, given the generic nature of the wind turbine in the model a constant rotation speed was used to allow direct comparison of SPL over a range of wind speeds.

5. Excitation Forces: The excitation forces are frequency dependent. The frequencies used for the gear box and generator are approximated from experience of Xi Engineering Consultants Ltd with similar sized machines, as are the magnitude of the forces which are related to the torque acting on the rotor. It is therefore assumed that the forces from the gear box and generator are proportional to torque which has a linear relationship with power and so can be calculated for different wind speeds.

6. Surface scattering: The far-field model assumed no scattering at the sea surface at all wind speeds. At higher wind speeds and related sea states it is likely that surface roughness would result in some scattering so that the SPL at large distances from the wind farm would be lower than those presented here. The models do not consider increases in the background noise level that may occur due to precipitation.

6.1.2 Assumptions affecting biological behaviour

Due to limited empirical data on the hearing sensitivity of the species of interest and their likely behavioural response to operational wind farm noise, it was necessary to make a number of assumptions and extrapolations to assess the likely effects of operating wind turbines on marine mammal species. These are outlined below ( Table 6-1).

Table 6-1 - Details of data quality and assumptions made in assessment of effects of operational wind turbines on marine mammals.

Subject Data Quality Comments
Sensation Levels Low / Medium Limited data available and for only three species - harbour seal, grey seal and harbour porpoise. Harbour porpoise thresholds measured at a higher frequency than the dominant frequencies in this study. It is unclear how suitable this metric is for minke whales and bottlenose dolphin. Sensation levels calculated from Kastelein, et al. 2005, Kastelein, et al 2006 & Götz and Janik, 2010.
Audiogram data Low / Medium Limited data available for species and audiograms are derived from a small number of tested individuals. For minke whales, the audiogram is predicted based on another baleen whale species (no empirical data exist). For the bottlenose dolphin composite audiogram it was necessary to use data collected from beluga whales to extrapolate the hearing to lower frequencies (as those generated by operational wind turbines).
Behavioural avoidance Low/Medium Limited empirical data available on behavioural response thresholds for species of interest. Audiogram and sensation level approach is not yet validated for all study species. Sensations levels also derived from a limited number of individuals in these studies.

6.2 Performance of models relative to previous studies

The study presented here used a generic wind turbine loosely based on a REPower 6 MW turbine with excitation forces and frequencies estimated from turbine measurements by Xi Engineering Consultants. It would be extremely beneficial to verify the model using field measurements specific to this scale of turbine and the foundation types investigated. The purpose of the study is to compare how different foundation types used to mount offshore wind turbines affect the noise level entering the marine environment. The study by its nature is therefore comparative; thus the magnitude of the SPL emitted by the wind turbines is of secondary importance. However, care has been taken to assign excitation forces and frequencies that are representative of multi-megawatt scale offshore turbines to allow the comparison of SPL to audiograms and behavioural parameters of marine species likely to interact with offshore wind farms in Scottish waters. Both the near- and far-field sound fields modelled appear to be consistent with previous sound measurements. The modelled sound field is dominated by tonal noise below 700 Hz consistent with the findings of Wahlberg and Westerberg (2005). The SPL modelled 30m from the turbine ( Table 3-7) and 100 m from the wind farm ( Figure 4-4) are consistent with those measured in previous studies ( Table 2-2 and Figure 2-1).

6.3 Comparison of foundation types

All three foundation types were modelled using identical wind turbines with the same excitation frequencies and forces. This results in near- and far-field spectra for the three foundations having peaks in similar one-third octave bands that relate to the gear-meshing and generator vibrations in the drive train (e.g. 12, 31, 80, 200 and 630 Hz, see Figure 4-4). However, the sound pressure levels of these peaks vary greatly between the foundations due to different geometries, construction materials and surface contact with the marine environment. Generally the monopile produces higher SPL at frequencies below 630 Hz, with peaks ~10 dB higher than those produced by gravity bases and ~50 dB higher than those produced by jackets ( Table 3-7 and Figure 4-4). The jackets may produce substantially less noise at low frequency due to having less surface area in contact with the marine environment. Gravity foundations and monopiles have large surfaces area; the gravity foundation has significantly more damping than a monopile resulting in the greater dissipation of vibration energy and the subsequent reduction in the amount of noise produced.

At frequencies greater than 500 Hz SPL produced by jackets become high relative to the monopiles and gravity bases ( Figure 3-15). The geometry of the jacket is dominated by steel cross bracing elements which are likely to resonate in the 100's and 1000's of Hz. The resonance of the cross bracing elements in the jacket may amplifying high frequency vibrations resulting in high noise emission.

6.4 Operational noise from wind farms and its effect on the behaviour of marine species

The modelled scenarios presented here indicate that there is the potential for operational wind turbines to increase the level of anthropogenic noise in the marine environment. It is likely that operational wind farms will be audible to marine mammals, with minke whales (low-frequency specialists) likely to detect them over ~18 km away.

A proportion of approximately 10% of minke whales and harbour porpoises encountering the sound field were predicted to exhibit behavioural response out to ranges up to ~18 km. It is, however, noteworthy that due to the low SPLs produced by the wind farms, the majority of animals (e.g. 50% or 90% of animals) would not show a behavioural response to these noise levels ( Table 5-2). This indicates that whilst there is potential for displacement to occur around operational wind farms this is most likely to be observed in less than 10 % of individuals. As noted above, behavioural response should be considered as probabilistic and is therefore best described with a dose-response relationship that describes the proportion of animals that may be expected to respond to a given sound level. In addition, for porpoises it was only using the M-weighting that resulted in predicted displacement and not using the sensation level or reverse-audiogram weighting approach. For harbour porpoises an M-weighted SPL of 90 dB was used as a threshold to predict the behavioural response of approximately 10 % of animals. Although harbour porpoise do appear to be relatively sensitive to these low noise levels (Southall et al. 2007), animals were only predicted to respond to SPLs of 90 dB when using the M-weighted behavioural response. Given that M-weightings are likely to be over-conservative, and do not fully account for the fact that porpoise are less sensitive at these low frequencies, this predicted displacement for 10% of animals can be thought of as precautionary. Given, however, that harbour porpoise, and other high-frequency cetaceans, appear to be more sensitive to lower received levels than other marine mammal hearing groups (Southall et al. 2007) it may be reasonable to again assume that an approximate 10 % proportion of individuals may show an avoidance response to even very low received levels of operational wind farm noise. The seal species (harbour and grey) and bottlenose dolphins were not considered to be at risk of displacement from the operational turbines given the results presented here.

There are limited available data on the response of marine mammals to operational wind farm noise. A study by Koschinski et al. (2003) played simulated operational turbine noise (from a 2MW turbine) to harbour porpoise and harbour seals and authors noted a reduction in sightings of porpoise and seals at a maximum range of 60m and 200m respectively. There are no studies on the impact of operational wind farm noise on baleen whales. However, Madsen et al. (2006) reported 10 km as a theoretical maximum for the zone at which baleen whales and pinnipeds could detect operational noise but emphasised that it is likely the actual range would be significantly less.

In another study, harbour porpoise presence was recorded at Horns Rev wind farm during the operational period (Teilmann, et al. 2006a) and telemetry studies have also shown harbour and grey seals transiting through the Nysted and Rødsand II wind farm areas (Teilmann, et al. 2006; McConnell, et al. 2012). As such, it does not appear animals are displaced from existing operational wind farms.

The predicted displacement of harbour porpoise at such large distances may warrant consideration of the population effects of such disturbance. The modelled outputs presented here do predict a 10 % proportion of harbour porpoise to respond as far as 18km away, however, a recent study by Nabe-Nielsen et al. (2011) modelled the effects of operational wind farm noise and shipping on the harbour porpoise population in the Kattegat. Changes in the animal's energy budgets were observed, but the results indicated that those operating wind farms and shipping in the region did not affect the size of the porpoise population nor its long-term survival (Nabe-Nielsen et al. 2011). It is important to note that each assessment is site-specific and therefore the scale, nature and extent of the disturbance should be monitored and the consequences assessed at specific locations in the UK in light of the results presented here.

It is also important to recognise that whilst the results presented here indicate that some species of marine mammals may be impacted upon, these outputs will change considerably depending on the sound characteristics of the local environment, particularly in areas of high vessel traffic. Low-frequency shipping noise appears to have higher source levels than the R1 and R2 wind farms measured in Madsen et al. (2006). It is likely that large amounts of shipping noise, if present in the vicinity of the wind farm, would mask any operational wind farm noise. This is, however, likely to be a function of distance and if animals are close to the windfarm then the operational noise may still be detected.

In addition, it is unlikely that these very localised high SPL levels would be spatially or temporally stable around a real-world operational wind turbine (e.g. temporal fluctuations in wind speed, turbulence etc. would results in the localised volumes of high SPL moving about). Therefore some of the predicted responses and consequences of exposure presented here are considered precautionary.

The fish examined tend to be sensitive to low frequency noise ( Figure 5-8) and are therefore more easily able to detect monopiles than gravity bases or jackets. Eels are able to hear monopiles up to at least 18 km away operating in 5 ms -1 wind speeds, whereas they need to be within 9 km of a gravity base to be able to sense it and they cannot sense a jacket at all in the far-field ( Figure 5-9). Similarly, salmon can detect monopiles at least 18 km away, gravity bases 13 km away, but cannot sense jackets in the far-field domain ( Figure 5-10). Sea trout are more sensitive to higher frequency noise (~500 Hz) making them able to detect gravity bases and jackets 4 -5 km away while being insensitive to the presence of monopiles in the far field.


Back to top