Environmental Noise Monitoring and Assessment of Petrochemical Plants Richard Mackenzie 1 Robin Mackenzie Partnership Unit 1, 7Hills Business Park, 37 Bankhead Crossway South Edinburgh EH11 4EP Scott Tunnah 2 Edinburgh Napier University Unit 1, 7Hills Business Park, 37 Bankhead Crossway South Edinburgh EH11 4EP

ABSTRACT This paper will look at the Scottish regulatory guidance for the monitoring and assessment of envi- ronmental noise from petrochemical plants. The practical challenges of on-site measurement and long-term monitoring will be discussed. The use of BS 4142 and NANR 45 within this context and the appropriate data analysis to assess noise impact on residential locations. 1. INTRODUCTION

Scotland is home to multiple large industrial facilities that have a variety of uses. With Scotland’s proximity to the North Sea oil fields a significant amount of these facilities are linked to petrochem- ical production. Often these facilities are located adjacent to residential areas which can result in noise being a cause of complaints from local residents. SEPA (Scottish Environmental Protection Agency) regulate the pollution emissions from these facilities, including noise.

SEPA regulates these facilities using PPC permits (Pollution Prevention and Control) which are required for the facility to operate. These are legislated via the Pollution Prevention and Control (Scotland) Regulations 2012.[1]

Much of the data in this paper has been anonymized to protect client operational sensitivities. 2. REGULATORY REGIME

2.1. Guidance - Noise and vibration management: environmental permits

The main source of guidance for noise and vibration from industrial sources in Scotland can be found in the recently updated ‘Guidance - Noise and vibration management: environmental permits, 2021’[2] published jointly by The Environment Agency, Natural Resources Wales, the Scottish En-

1 Ri.Mackenzie@Napier.ac.uk

2 S.Tunnah@Napier.ac.uk

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

vironmental Protection Agency and the Northern Ireland Environment Agency. This covers the meth- ods of assessment of noise and vibration for PPC renewals and any assessments that may be required as a condition of the permit.

PPC permits are reviewed typically, every four years to ensure BAT (Best Available Techniques) are being implemented to reduce emissions. A Noise Impact Assessment is required as a part of this renewal with the aim of preventing noise creep and ensuring Best Available Techniques are used. More frequent NIAs can be requested as part of the PPC permit as a result of complaints from local receivers or as a response to changes in plant or emergency response plant. For example, a petro- chemical facility can have an NIA requirement if flaring is undertaken or a specific item of plant needs a more regular assessment.

The Relevent Noise and Vibration Guidance states the following with regards to noise; “You must use BS 4142 to quantify the level of environmental noise impact from industrial processes. In rare circumstances, other methods may also be appropriate (for example, NANR45 for assessing existing low frequency sound inside a residential property).” 3. GENERAL INDUSTRIAL NOISE

3.1 BS 4142:2014 +A1:2019

BS 4142:2014 +A1:2019 is a standard focused on assessing the noise impact of industrial noise sources. The objective assessment in its most basic form, is a comparison of the Rating level (L Ar,Tr noise generated by the industrial source at a Noise Sensitive Receiver plus any penalties applied) against the Background noise level (L A90 noise level at NSR when source is inoperative). The rating level can be altered using penalties which can be objectively or subjectively applied to compensate for factors that affect perceptibility such as Tonality, Impulsivity, Intermittency, and other sound characteristics that are readily distinctive against the residual acoustic environment. [3]

BS 4142 Specifically states that is ‘not applicable to the assessment of low frequency noise’ and refers to the NANR 45 ‘Procedure for the assessment of low frequency noise disturbance’ assessment methodology from the University of Salford. [4]

Applying BS4142 in the context of a petrochemical plant can provide many challenges. For ex- ample in some scenarios obtaining representative background or residual levels without the plant or item of plant being assessed operating may require surrogate positions which can introduce signifi- cant uncertainties that are difficult to quantify. 4. LOW FREQUENCY NOISE

4.1. NANR 45 ‘Procedure for the assessment of low frequency noise disturbance’

NANR 45 sets an internal noise limit curve similar to NR/NC curves but at the low frequency end of the spectra. Figure 1 below shows the limit criteria from NANR 45.

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 1: Proposed reference curve, NANR 45 5. PETROCHEMICAL PLANT MAJOR NOISE SOURCES

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

5.1. Steam

When in the vicinity of a petrochemical plant the most obvious noise sources are a number of high- pressure steam releases. These can be from leaks or pressure control valves on various larger items of equipment such as cracking towers. These dominate the mid to high frequency spectrum and are therefore very noticeable to humans. Petrochemical plants can have hundreds of simultaneous steam release locations and therefore prediction of the attenuation of these sources can be uncertain.

Decoking cracking towers with the use of steam can also generate high levels of steam noise. This is undertaken intermittently to remove the build-up of coke on equipment as a by-product of the cracking process.

5.2. Boilers

The boilers generate the steam used around the plant. Boilers generally have a low frequency rumble alongside the associated steam pipework and are located in central areas.

5.3. Condensers

Condensers are used to cool gasses to below boiling point. There can be a variety of different types of compressors in use at a petrochemical plant. Noise sources include pumps and fans.

5.4. Ground Flares

Ground flares are used to burn of excess hydrocarbon stock. Typically ground flares have a low frequency rumble from the combustion noise. As they are often quieter and less visually intrusive than elevated flares they do not attract as much attention.

5.5. Elevated Flares

Elevated Flares are often a source of complaints related to petrochemical operations. They com- bine a visible flame at hight with an elevated noise source. The noise profile of an elevated flare can vary significantly depending on the flare tip design and the hydrocarbon and steam flow rates but will typically contain a significant low frequency element.

aa a a ag cx cect | {5 E

6. LONG TERM NOISE MONITORING

6.1. Equipment

Long term noise monitors typically consist of a Class 1 Sound level meter, a mobile capable mo- dem, and a power supply system. They can also contain other instrumentations such as meteorological sensors and air quality monitors. Figure 2 below shows a permanent monitoring system provided by ANV Measurement Systems containing a RION NL-52 sound level meter.

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 2: ANV LivEnviro Systems on site

6.2. Benefits of Permanent Monitoring Locations

Permanent monitoring of noise provides two main benefits. The first is an in depth understanding of the noise environment and therefore a more detailed understating of background and residual noise for the purpose of a BS 4142 assessment. The second benefit is response time. For example, in the event of a process upset event resulting in elevated flaring, monitors are already collecting data and can capture the entire event. Hydrocarbon flow rate to an elevated flare is often at its maximum at the beginning of an event and therefore, it is important to measure any noise emission during this period. Permanent monitoring also allows for retrospective analysis of noise output to ensure that BAT is followed and that noise emissions do not ‘creep’.

6.3. Practical Implementation of a Permanent Monitor

Powering the monitor is the main practical hurdle to overcome. In Scotland during winter months solar energy is at a minimum, therefore large battery banks and multiple panels are required. Calibra- tion of the equipment should also be checked every month and any deviation noted.

Web based platforms such as LivEnviro also allow for multiple users to view and analyse data as well as remote altering of settings such as enabling audio recording. 7. DATA FILTERING AND ANALYSIS

7.1. Meteorological

When undertaking noise impact assessment data should be analysed in conjunction with synchro- nised meteorological data to disregard periods where weather has an undue influence on the results.

BS 4142 suggests that wind speeds above 5m/s should be excluded. RMP data suggests that if low frequency noise analysis is to be undertaken wind speeds above 3m/s should be excluded as aerody- namic noise across the microphone can generate increased low frequency levels.

7.2. Unwanted noise

During periods of interest audio recording should be enabled to allow for the identification and removal of extraneous noise sources such as sirens, road works or noise from animals or locals. RMP enable all monitors with 100ms logging to allow for the repossessing of data in software such as RION AS60. This allows for periods to be removed and reprocessed while still providing ‘true’ per- centile values.

7.3. Synchronized Clocks

Ensuring all measurement devices have synchronized clocks is important for analysis of data over short time periods. For example, two synchronized sound level meters can give real time level differ- ence data analysis which is helpful for estimating the propagation of the source under consideration. Measurement equipment can also be synchronized to the clock of the facility being monitored allow- ing for analysis of flow rate vs noise data. 8. EXAMPLES OF DATA ANALYSIS

8.1. Wind Speed affect on noise measurements

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 3 below shows the noise level Vs wind speed for a position exposed to the wind.

Figure 3: Broadband SPL Vs Wind Speed

Figure 3 shows that broadband noise levels increase as wind speed increases and that low frequency noise is more affected by this than broadband. Even below 5m/s as recommended in BS 4142 wind has significant impact on LF noise and this should be considered if analysing for an NANR 45 assessment. Figure 3 demonstrates wind speed should also be taken into account when considering representative background and residual noise levels for a BS 4142 assessment and for calculating internal levels in line with NANR 45.

8.2. Wind Direction Effect on Residual Noise Such as Road Traffic

Figure 4 below shows the noise level Vs wind direction for a monitoring location approximately 420m South of the road traffic noise source.

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 4: Wind Direction Vs L A90 , Road to south of measurement position, wind speed below 5m/s

Figure 4 shows that L A90 can vary up to 10dB depending on the wind direction. Wind direction should be taken into account when considering representative background and residual noise levels for a BS4142 assessment.

8.3. Hydrocarbon and Steam Flow rate analysis for flaring

Figure 5 below shows the noise levels verses steam and gas flow rate in the vicinity of an elevated flare.

ween, He

Figure 5: Elevated steam and hydrocarbon flow Vs noise

Figure 5 shows correlation between LF and broadband noise and flaring activity. This is useful for determining noise levels near the source and looking for correlations at NSRs. It also allows for anal- ysis of steam and hydrocarbon flow rates against the noise measurements to optimise the ratio to minimise noise.

8.4. Calibration of Acoustic Models

By using simultaneous measurements at source and receiver an understanding of the attenuation profile of a source can be used to calibrate the factors affecting the attenuation simulated in acoustic models such as CADNA or SoundPlan.

8.5. Shadow Zone and Nearfield Propagation Issues

Calibrated software models combined with a detailed knowledge of an elevated flare tip geometry can allow for a more detailed understanding of shadowing and therefore can inform the positioning of a monitor to accurately measure noise from a flare.

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 6: Example of a flare tip model with a shadow zone

Figure 6 above shows a hypothetical flare tip where shadowing occurs. This model has been cali- brated using on site measurements and can now be used to measure flare noise in a location where it can be distance corrected to other receivers with a reduced level of uncertainty. 9. ASSESSMENT METHODS AND SOURCES OF INNACURACY

9.1. Residual and Background Noise

From long term monitor analysis RMP has found that there is significant variance in residual and background noise over time. After 6 months of monitoring at NSR’s where residual and background noise is dominated by constant road traffic flow data ranges can vary on average by 17dB within the same weekday hour. The selection of representative residual and background levels is therefor subject to large uncertainties which feed through to the assessment rating level.

As BS 4142 is a comparative assessment of up to 3 measured broadband factors (ambient, residual and background) it can be prone to erroneous results from inaccurate measurements or misinterpreted data. For example, adjacent to B roads with intermittent traffic the difference between residual and background noise can be ±20dB. In this scenario if there is limited specific noise measured at the NSR but the perceived ambient is only slightly above the assumed residual a large BS 4142 rating level can be produced.

9.2. Low Frequency Noise Sources in BS 4142

BS 4142 Specifically states that is ‘not applicable to the assessment of low frequency noise’ and refers to the NANR 45 ‘Procedure for the assessment of low frequency noise disturbance’ assessment methodology from the University of Salford however regulators may still require a BS4142 assess- ment to be undertaken which can be affected by the significant uncertainties described above if the source does not have a significant broadband content at the receiver.

9.3. NANR 45 Internal Measurements

The criteria shown in Figure 1 is an internal criteria. In the context of petrochemical plant noise monitoring where some low frequency noise sources such as flaring only occur as an emergency response it is difficult to access dwellings with short notice. Therefore, façade insulation has to be estimated which introduces further uncertainty. Internal NANR 45 measurements can also be affected by room modes and other room acoustics phenomena. Low frequency sound insulation estimates are often difficult to undertake or find data on. 10. CONCLUSIONS

This paper has covered the Scottish regulatory guidance for the monitoring and assessment of environmental noise from petrochemical plants as well as a discussion on the practical challenges of on-site measurement, long-term monitoring, data analysis and noise impact assessment.

As the topic of this paper is broad and data is being continually gathered there is a large potential for future studies from many of the topics touched on in this paper such as the effect of wind direction on measured noise, the impact of wind speed on low frequency noise, methodologies for determining representative background and residual noise periods from long term measurements to reduce assess- ment uncertainty. 11. ACKNOWLEDGEMENTS

I would like to thank my colleagues at Robin Mackenzie Partnership and Edinburgh Napier Univer- sity for helping us to gather and process the data found in this paper. 12. REFERENCES

1. Scottish Environment Protection Agency Website, 2022, https://www.sepa.org.uk/regula-

tions/pollution-prevention-and-control/ 2. Guidance - Noise and Vibration Management: Environmental Permits , The Environment Agency, Natural Resources Wales, Scottish Environmental Protection Agency, Northern Ireland Environment Agency, 2021 3. BSI Standards, BS 4142:2014+A1:2019, Methods for rating and assessing industrial and com- mercial sound, 2019 4. Moorhouse, AT, Waddington, DC and Adams, MD, Procedure for the assessment of low fre- quency noise disturbance , Department for Environment Food and Rural Affairs, University of Salford, Manchester, 2005

‘inter.a 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW