Strategic modelling of industrial noise with limited input data Chris Skinner 1 AECOM Ltd. 12 Regan Way, Chilwell Nottingham, NG9 6RZ, UK Seckin Basturk 2 AECOM Ltd. 12 Regan Way, Chilwell Nottingham, NG9 6RZ, UK Hilary Notley 3 Department for Environment, Food and Rural Affairs Seacole Building, 2 Marsham Street London, SW1P 4DF, UK

ABSTRACT Historically, strategic modelling of industrial noise exposure on a national scale has taken a wide range of approaches. Where sound emission data of industrial sites are not available, numerous as- sumptions are often made. These include simplified modelling of industrial sites and adoption of generic sound power levels for different industry types. In most cases such assumptions significantly increase the uncertainty in strategic noise mapping outcomes. This paper presents an alternative approach to strategic modelling of industrial sound which is based on the environmental permitting practice of large industrial sites in England. Drawing on this approach, a cost-effective and scalable modelling procedure has been developed to estimate sound emission levels and radiation patterns of industrial activities accounting for the local context of the site. This paper details the alternative modelling approach along with example prediction outputs. Potential future refinements to improve the modelling accuracy and opportunities to enhance site-specific input data in collaboration with national authorities are also discussed.

1. INTRODUCTION

Strategic noise maps, as defined in the Environmental Noise Directive (END) [1] produced in the last 20 years employed a wide range of modelling approaches. The disparity was undoubtedly driven by the conditions and constraints faced in different countries or agglomerations; depending on the extent of the modelling task and the quality of the readily available input data, the most cost-effective approach will vary significantly between countries and agglomerations.

1 chris.skinner@aecom.com

2 seckin.basturk@aecom.com

3 noise@defra.gov.uk

worm 2022

The toolkits introduced by the Good Practice Guide for Strategic Noise Mapping and the Produc- tion of Associated Data (GPG) [2] have been helpful to deal with the shortfalls in data availability and quality for strategic noise modelling. Moreover, the common assessment methods [3] have been an important step towards a more consistent and comparable strategic noise mapping system. How- ever, strategic mapping of industrial sound in particular is still a process where a huge number of decisions should be made to make use of the limited information available for each industrial source considered, to obtain an acceptable level of quality whilst keeping the costs at a reasonable level.

If cost was to be removed as a factor from this process, the modelling methodologies could become expansive and time demanding. For example, the most accurate noise map could be produced with detailed analysis of each site, employing sound measurements to characterise sound sources at every industrial site in consideration. This would provide an unparalleled level of accuracy, with vast quan- tities of data, but is not usually feasible when considering the cost and time required to undertake such detailed analysis on industrial processes on a national scale.

This paper presents a cost-effective approach to strategic modelling of industrial sound which is based on the environmental permitting practice for large industrial sites in England. The proposed methods in this paper make use of the readily available datasets to estimate sound emissions of indi- vidual industrial sites. The paper also provides a discussion on the potential future steps to further improve the quality of the modelling input data to achieve more reliable cost-effective strategic noise maps for industrial noise.

2. PROPOSED MODELLING APPROACH

2.1. Mapping Area and Scoping Distance

Strategic noise mapping for industrial noise aims to provide results within designated agglomera- tions. In order to account for industrial noise produced outside of an agglomeration but affecting the mapped area (i.e. receptors within the agglomeration) it is recommended to extend the spatial scope of the study. It is anticipated that at distances beyond 1 km (from the industrial site boundary), the sound from most industrial activities would be attenuated to levels which can be considered negligible (i.e. not resulting in levels above the minimum mapped values of 55 dB L den and 50 dB L night ). There- fore, a scoping distance of 1 km from the agglomeration boundary is considered appropriate for the purposes of noise modelling.

2.2. Industrial Sites and Modelling Area

In England, permitted industrial activities are classified as Part A(1), Part A(2) or Part B activities, as described in Environment Agency’s (EA) Environmental Permitting Regulation (EPR) [4]. Part A(1) activities, which are considered to be the most polluting, are regulated by the EA. The remaining Part A(2) and Part B activities are regulated by relevant local authorities (LAs).

The Part A(1) activities, could be determined using the ‘permitted installations list’ published [5] and regularly updated by the EA. This dataset includes high level information including site address, grid reference and industry type. The local-authority regulated activities are usually recorded in da- tasets created and maintained by individual LAs. A desktop investigation considered the West Mid- lands Agglomeration and Birmingham City Council and identified that most LAs within this area publish datasets of relevant industrial sites. It has been assumed that this level of coverage is repli- cated across England. However, the file format, and formatting of activity details and location infor- mation vary significantly. Collation of these into a unified national list, similar to that of Part A(1) activities, would roughly triple the number of industrial activities considered and require significant time. Such an increase in scope is also estimated to triple the data processing costs which would hence significantly increase total noise mapping costs. Therefore, currently, it is considered reasonable to

worm 2022

limit the scope of strategic noise modelling to include only Part A(1) activities, consistent with the approach used for previous rounds of strategic mapping in England.

In order to determine the spatial scope of the study and inform detailed modelling, it is necessary to identify the site boundary of each industrial activity included in the EPR Part A(1) dataset. The publicly available EPR Landfill Sites [6] dataset has been used to identify the extent of landfill activ- ities included in EPR Part A(1) dataset. The open-source Land Registry Parcels [7] dataset has been used to identify the boundary of the remaining activities.

After identifying the site boundaries, all Part A(1) activity sites that fall fully or partially within the agglomeration boundary or 1 km (scoping distance) buffer have been considered to be within the scope of the mapping exercise. Part A(1) activities with a site extent entirely outside the agglomera- tion and the scoping buffer area have been scoped out.

A further 1 km buffer from the edge of the activities scoped in has been considered appropriate to allow for the sound power level determination process (see Section 2.3 for details) which requires the inclusion of nearest residential receptors in the modelling. The total modelling area has been formed by union of these additional buffer areas with the previously prepared 1 km buffered mapping area. Figure 1 illustrates the spatial scoping of Part A(1) activities along with formation of the modelling area for four hypothetical industrial sites.

worm 2022

Figure 1: Modelling area formation

2.3. Modelling of Industrial Noise Sources

Careful consideration is required when selecting an appropriate method for modelling noise sources within each industrial activity scoped in. In practice, a great number of sources may be present that contribute towards the overall sound output of any one industrial site. How to most accurately represent this across multiple sites, without specific knowledge of equipment and running times, pro- vides a significant limitation in the accuracy of any noise model produced.

For individual industrial sites, data sufficiently detailed for direct input to noise modelling is not readily available in a usable format. However, documents such as noise management plans and noise impact assessments are sometimes available online or might be directly available with the cooperation of the operator. Although the opportunity exists for a more accurate noise model making use of such detailed information, due to the absence of a unified database including all sites to be modelled, col- lation of such detailed information would require significant human resource, and is highly likely to be inconsistent across different sites and local areas depending on the level of detail available.

Therefore, a desk-based exercise was considered to be a more appropriate approach. The use of publicly available data such as land registry, Ordnance Survey mapping, aerial photography,

‘Aaplomeration boundary (rapping area) ne ilometre butr from the edge of scoped Indust in india ses reference point Indust sts that fll ‘onto utside tho GS aping tance a0 Indust ste spain ‘boundary Modeling area ‘Scoping distance ‘fone klometre

street/roadside imaging, can provide details relating to industrial site layouts. These data sources are suitable for automated processing within Geographic Information Systems (GIS) environment and can provide information on structures to aid modelling assumptions such as: • Location and height of noise sources: e.g. all buildings within an industrial site with a certain size and height can be assumed noise emitting structure; • Identification of structures / outlets which might be linked to noise sources: e.g. based on aer- ial/street imagery outdoor plant and/or building elements can be assumed as noise source; • Location of buildings; • Identifying the use of areas within the site which can be assumed to emit noise due to the operation of mobile plant. It is important to note that due to automated processing opportunities, desk-based exercises as described above require relatively less human resource and allow for a consistent approach for all industrial sites included in noise modelling.

Type of Source and Its Distribution The information derived from publicly available data sources mentioned above can be used to identify likely sound sources and their distribution within an industrial site. A number of modelling tests (not detailed in this paper) have been conducted to determine a set of source distribution as- sumptions that provide reasonably reliable prediction results at various distances and directions from the industrial site.

Considering accuracy and complexity involved in these test scenarios, a hybrid approach was cho- sen. In this approach, the industrial activity is represented by an area source covering the entire site and additional area sources covering site building footprints. The site area source is located at 1.5 m above ground. Remaining sources are located at roof height of the corresponding industrial buildings.

Determining Sound Power Level of the Source Collation of noise source information for each industrial site, from site operators or permitting authorities, is anticipated to be a very costly process due to manual investigation of documents with varying content and formatting, and therefore considered unfeasible.

Following the guidance provided in the GPG Toolkit 10.5, a reasonable approach to determine industrial source sound power level is using “nationally defined maximum permissible sound power levels per unit of surface area”. In England, there is no nationally defined permissible industrial sound power levels. Industrial noise is usually studied in detail for each industrial site individually and its impacts assessed in accordance with the BS 4142 [8] standard using differential noise indicators.

The following steps are proposed to derive maximum permissible sound power levels for each site to be used in industrial strategic noise mapping.

Determination of the broadband background sound le vel a t t h e nearest noi se-sensi t i ve r esi dential receptor: Strategic noise mapping results involving road and/or railway traffic can be used to obtain equivalent noise levels during day, evening and night at any given location. Then background sound levels (L A90 ) for each time period can be derived from these using conversion factors (such as those derived from readily available measurement databases, e.g. Focussed Noise Monitoring Study 2013 (FNM2013) [9]). From the background sound level, the maximum permissible sound power levels for the nearby industrial sites can be determined as below.

Determination of the maximum permissible broadban d speci fi c sound l evel a t t he n ear est noise- sensitive residential receptor: BS 4142 states that a difference (when background level is subtracted from the rating level) of around +5 dB is likely to be an indication of an adverse impact, depending on the context. This is usually accepted as an impact threshold and for most industrial sites this could

worm 2022

be considered to define a maximum permissible BS4142 rating level. Applicable character correc- tions may vary from 0 to +9 dB depending on a number of specific characteristics of the sound from the site. However, for the purposes of strategic noise mapping, a default correction of +3 dB could be assumed to identify the maximum permissible specific sound level of an industrial site at the near- est noise-sensitive residential receptor.

The assumptions suggested in this step could be further investigated and adjusted to location spe- cific conditions, where needed, but the above approach is considered a reasonable set of assumptions in the absence of such local information.

Deriving the maximum permissible broadband overal l i ndust r ia l si t e sound power l e vel : Depend- ing on the details of how the noise sources of an industrial site are modelled, the overall site sound power level can be adjusted to give an identified maximum permissible specific sound level at the nearest noise-sensitive residential receptor. This could also be back-calculated using a known noise limit value at the façade of the nearest noise sensitive receptor and using this to assume a worst case (but still acceptable) sound power level of the industrial site.

Determining Sound Power Level Spectra Comparing two different industrial sites, the sound power level spectra of industrial plant may vary significantly. Site operations and plant operating on site are likely to vary significantly between sites, and some noise sources may be mobile and/or moving. Such detailed information can only be gathered by detailed analysis of sites and usually by undertaking measurements of source levels which are not considered to be feasible due to disproportionately high cost.

Following the guidance provided in the GPG Toolkit 10.5, a reasonable approach to determine industrial source sound power level spectra is using “public databases” such as the IMAGINE [10] database created as a result of an EU funded research project to support strategic noise mapping in the EU. The IMAGINE database includes sound power level spectra for a large number of plant that are widely employed in various industry types across the UK and the EU.

This database can be used to derive typical sound power level spectra for each industry type. These then can be adjusted to the identified maximum permissible broadband sound power levels (discussed above) for each site.

The modelling of industrial sound sources discussed above is summarised in Figure 2. 3. DISCUSSION

The proposed modelling approach has been tested through a pilot study applied to West Midlands Agglomeration (WMA) in England.

The predicted levels have been analysed and it is noted that, around clusters of industrial sites, predicted overall industrial noise levels exceed the identified maximum permissible specific sound levels by up to 9 dB. This is an expected outcome as noise from multiple industrial sites positioned relatively close to each other, resulting in higher overall industrial noise levels at nearby receptors compared to noise from a single industrial site.

At such locations, it is possible that the actual permitted specific sound levels for individual sites are lower, i.e. the permitted rating level over background level is smaller than the +5 dB threshold value assumed in this pilot study, in order to limit the cumulative industrial noise impact experienced in the area. The suggested method in this pilot study could be modified to consider such clusters of industrial sites and apply a more stringent criteria (e.g. 0 or -10 dB impact threshold) to identify maximum permissible specific levels. However, such site-specific variations would not necessarily

worm 2022

reflect the actual conditions and balance of levels between the contributing sites. Hence the applied method is considered to be a reasonable approach for the majority of the cases and appropriate given the strategic nature of the mapping.

With regards to the overall performance of the suggested method, it defines an appropriate and scalable scope which can be applied on a national level where readily available input data is similar to the those discussed in the case of England. A few potential areas for future improvement have been identified; Through examination of the documentation regarding environmental permits, the follow- ing site-specific information could be incorporated into the proposed method: • Measured specific sound levels from the industrial site at known locations; • Measured background sound levels in the vicinity of industrial sites; • Applicable character corrections determined by noise assessments according to BS 4142; and • Applicable noise impact threshold. Consideration of the above would require significant additional work over the mostly automated approach presented in this paper, due to the collation and processing of site-specific information. It is highly likely that this would require a large degree of manual processing and would vary signifi- cantly depending on the quality and format of the available information on environmental permits of industrial sites. However, a data standard could be set for future documentation to be submitted to the EA and local authorities for permitting or compliance purposes. In such a case, the data standard should be determined considering its usability in strategic noise mapping for the future. The following options summarise how different data standards could inform noise modelling: • Sound power level spectra of each sound source: This is the most detailed option and should be accompanied with additional source related information, such as source type, coordinates, di- rectivity, etc., to allow for inclusion in noise modelling with no or minimal processing. This ap- proach would provide accurate modelling but might cause longer calculation times due to the large number of sources included in the model. • Specific sound pressure level spectra at site boundary: Measured or predicted specific sound pressure levels at site boundary, ideally at several directions, would provide sufficient infor- mation to strategic noise mapping to approximate off-site noise propagation as considered in a detailed assessment. This would require some data processing and additional modelling time to be able to create a model of the site capable of predicting same levels at site boundary. • Specific sound pressure level spectra and background noise level at nearby noise sensitive recep- tors: The required processing work and resulting accuracy of this approach is similar to the above. The accuracy of the model might be lower if all directions are not covered.

worm 2022

worm 2022

* Noise monitoring data available from Focussed Noise Monitoring Study 2013 (FNM2013) project has been used to obtain given equations. These reflect the average dB difference between measured L Aeq and L A90

levels. Figure 2: Determining maximum permissible specific level 6. REFERENCES

1. European Commission. Directive 2002/49/EC of the parliament and of the council of 25 June

2002 relating to the assessment and management of environmental noise. (2002) Available from: http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32002L0049 2. European Commission Working Group. Position Paper Final Draft, Assessment of Exposure to

Noise (WG-AEN). Version 2 (2007) 3. As outlined in Commission Directive (EU) 2015/996 which was transposed into national law by

the Environmental Noise (England)(Amendment) Regulations 2018.

Step 1: Identified industial ste ‘boundaries ‘Step 3.1: entity geometrical contre ofall site buildings using publly avaiable building dataset Step 32: Identify ‘geometrical centre of the industrial ste I ‘Step 4: Identity closest noise sensitive ‘receptors to representative Industrial source centre points sing publi avaiable buling (eosdental buldings ony) dataset I ‘Step 7: Derive Lao soe 294 Lise Step 6.1: Define butter areas (in al directions) trom industrial site boundaries, which extend upto 100 m beyond the ces eee eat ot Ena ca Fema Jaa, sree ae te 1 vento oe bagasse Shegiiy <2 iio T ee a otsinagonenton, pune sot vouch fomeietoae ot) J co 28 btn 259 isin om sca coe ala tine ‘each industrial activity: Rating Level = Las 5 for each industrial activity: Specific Level = Rating Level -3

4. The Environmental Permitting (England and Wales) Regulations 2016. UK Statutory Instrument

1154. 5. Environment Agency. Environmental Permitting Regulations - Industrial Sites Quarterly Sum-

mary Available from: https://data.gov.uk/dataset/99ebf94f-5069-4470-9d27-09fe2d 3a05c8/ envi- ronmental-permitting-regulations-industrial-sites-quarterly-summary 6. Environment Agency. Permitted Waste Sites - Authorised Landfill Site Boundaries. Available

from: https://data.gov.uk/dataset/ ad 695596-d71d-4cbb-8e32-99108371c0ee/permitted-waste- sites-authorised-landfill-site-boundaries 7. Open source HM Land Registry INSPIRE Index Polygons. Available from: https://www.gov.uk/guidance/inspire-index-polygons-spatial-data 8. BSI Group. British Standard 4142:2014 + A1:2019 ‘ Methods for rating and assessing indus-

trial and commercial sound’. (2019) 9. Department for Environment, Food and Rural Affairs. Focussed Noise Monitoring Study (2013).

Available from: http://sciencesearch.defra.gov.uk/Default.aspx?Menu=Menu& M od- ule=More&Location=None&Completed=0&ProjectID=18578 10. DeltaRail NL. IMAGINE - Improved Methods for the Assessment of the Generic Impact of Noise

in the Environment. (2006)

worm 2022