Marine piling - energy conversion factors in underwater radiated sound: review

Executive Summary

This project was commissioned by Marine Scotland, with the aim to improve the understanding of the Energy Conversion Factor (ECF) method, and to make recommendations regarding the modelling approaches for impact piling as used in environmental impact assessments (EIA) in Scottish Waters.

It is necessary to distinguish the genuine ECF, which is the ratio of the total sound energy in the water column to the hammer input energy, from what is termed the point-source equivalent ECF. The point-source equivalent ECF method involves calculating a broadband source level for a pile based only on the hammer strike energy and a value for the conversion factor. Sound source energy is then distributed over frequency bands based on a typical piling spectrum, and propagation losses calculated such that the sound field can be generated.

The key finding from this work is that, while the standard use of ECFs for piling is entirely valid, the process of generating a source level and propagating using point source models as used by the point-source equivalent ECF method reproduces the sound field from piling poorly. While there are concerns raised about the selection of a suitable value of ECF, greater errors are likely to arise in the choice of propagation models.

Values of the ECF in research show a range from 0.17 % to 1.56 % (9.6 dB). In EIAs where the method is used, a value of 0.5 % or 1.0 % has been arbitrarily chosen. Whilst these values sit well within the range of expected ECF values, there has been no regard for aspects of the operations that would affect this value. The difference between using an ECF of 0.5 % and 1.56 % is 4.9 dB, indicating that equivalent levels may be underpredicted by that much. There are few reported values of the ECF from measurements and modelling, and those that do exist are in shallow waters. Where point-source equivalent ECFs have been estimated by back-propagated point source model results, a much wider range of values has been generated (up to 13%).

In terms of sound propagation, the nature of a point source is to transition from a spherical spreading regime (∝20log10(r)) to a cylindrical spreading regime (∝10log10(r)) to a mode-stripping (or intermediate) regime (∝15log10(r)). The sound field from impact piling, however, features a Mach cone wavefront for which the propagation within the first few kilometres is described by the damped cylindrical model (DCS) (∝10log10(r)+ar). These differences result in very different rates of energy loss and can lead to over or underestimates in the region of 10 dB within 5 km of the pile.

Combining the effect of predicting the source level using a point-source equivalent ECF of 0.5 % and using point-source propagation for one presented example yielded underestimates of the per-pulse sound exposure levels between 100 and 1000 m from the pile of between 9.5 and 12.1 dB. Using the point-source equivalent ECF method for a benchmark case scenario similarly showed underestimates between 6.3 and 10.2 dB for receivers at 250, 750, and 1500 m from the pile.

The recommendation is that the point-source equivalent ECF method, i.e., the combination of source level prediction and point-source propagation modelling, should not be accepted in EIAs due to the evident errors in its predictions. Whilst there may be potential for ECFs to be used in the prediction of noise from offshore piling, it would require a significant amount of development due to the many factors controlling the ECF and few studies that report it. Additionally, such an approach would need the source output to be coupled to an appropriate model. Sound propagation calculations using a point source model to represent a pile, as used in the reviewed reports, can result in substantial errors with examples showing deviations of around 10 dB within 5 km Alternatives are suggested depending on the availability of data and the level of detail required. In cases where measurements are available for the operation, sound field estimates can be reasonably well retrospectively calculated using the DCS model. In cases where measurements for similar operations exist there has been some success in applying scaling laws to account for changes in input parameters to generate sound field predictions. Additionally, sensitivity studies have been performed for operations in the US to generate recommendations as to when existing model data can be reapplied to a new prediction or if new modelling is required. In cases where some aspect of the pile, the hammer, or the environment of the scenario to predict is sufficiently different from those available in previous works numerical modelling using models that are designed specifically for piling is recommended. Furthermore, if complex environments, underwater piling, raked piles, or mitigation techniques such as bubble curtains are present, numerical modelling is recommended.