Noise from Pubs & Clubs - Phase II

NOISE FROM PUBS AND CLUBS

(PHASE II)

FINAL REPORT

MAY 2006

Contract No. NANR 163

NOISE FROM PUBS AND CLUBS

(PHASE II)

FINAL REPORT

MAY 2006

Contract No. NANR 163

Document Control

Version

Issued

Date

Document Reference

Status

Version 2.0

Final Report

11 May 06

Noise from Pubs and Clubs Final Report.doc

Final

Defra

Noise from Pubs and Clubs (Phase II)

Contract No. NANR 163

Final Report

CONTENTS

41executive summary

72introduction

93legislation

144phase one

165Laboratory Testing

356Field trials

447Stakeholder Involvement

458assessment laboratory and field testing

499conclusions

5110recommendations

53Appendix a - GLOSSARY OF ACOUSTIC AND STATISTICAL TERMS

56Appendix B – Briefing and order of presentation

59Appendix C - QUESTIONNAIRES

76Appendix D – noise measurement data from laboratory tests

100Appendix E – descriptive statistics from lab testing

107Appendix F – Results from test questionnaire, lab testing

114APPENDIX G - Field questionnaire for EHPs

116APPENDIX H - Descriptive analysis of field test data

124APPENDIX I – Comments from field test questionnaires

126APPENDIX J – BRIEFING DOCUMENTS FROM FIELD TRIALS

Acronyms & Abbreviations

CSL

Capita Symonds Ltd

BRE

Building Research Establishment Ltd

Defra

Department for the Environment, Food and Rural Affairs

EHP

Environmental Health Practitioner

FPN

Fixed Penalty Notice

DPS

Designated Premises Supervisor

Drawing List

List of Figures

Figure No.

Title

Page No.

1

Test houses used for laboratory testing

16

2

Typical bedroom in test house

17

3

LAeq and LCeq indicators for tests with structure-borne transmission

24

4

LAeq and LCeq indicators for tests with airborne transmission

24

5

Responses on acceptability of the overall noise level by test condition

26

6

Acceptability ratings in the laboratory for Absolute LAeq

31

7

Acceptability ratings in the laboratory by LA90 minus LA90 (no music)

33

8

Scatter plot of field assessments of acceptability as a one-off event (inside ratings)

43

9

Scatter plot of field data of acceptability as a one-off event vs. LA90 minus LA90 (no music)

43

List of Tables

Table No.

Title

Page No.

1

Median tests on noise test levels

27

2

Kruskal-Wallis test on differences between rooms

28

3

Spearman’s rho coefficients for acceptability of level of sound and noise metrics

29

4

Semantic descriptor and associated value of acceptability

34

5

Summary of Field Trial Venues

35

6

Acceptability ratings of entertainment noise by EHPs judged as one-off and regular events

39

7

Spearman’s rho coefficients for acceptability of level of sound and measures of sound, for entertainment noise judged as a one-off event

40

8

Spearman’s rho coefficients for acceptability of level of sound and measures of sound, for entertainment noise judged as a regular event

41

9

Summary of performance of each noise metric, lab and field tests

45

1 executive summary

Bringing licensed premises within the scope of the Noise Act 1996 is intended to add to and complement existing powers. It will provide a relatively easy to use mechanism that can be fully implemented in response to a complaint on the night that any problem arises. The aim is to fill any gap in existing legislation and reduce the time to provide effective enforcement. Such a measure is required to help counter the potential for increased noise disturbance due to the liberalisation of the licensing regime brought about by implementation of the Licensing Act 2003 in late 2005.

Changes to the licensing laws in 2005 mean that licensed premises can be open later and for longer hours. Any such changes need to be supported by enforcement powers so that anti-social behaviour, where it occurs, is not tolerated. Extending the provisions of the Noise Act 1996 from dwellings to other premises was also one of the recommendations of the Environmental Audit Committee. The new powers are intended to provide a rapid reaction to problems when they first arise with the penalties aimed at discouraging further repetition of the problem.

Amendments to the Noise Act 1996 mean it is no longer adoptive and local authorities no longer have to respond to all complaints between 11pm and 7am every night. Instead local authorities now have the discretion to provide a response to complaints by targeting particular sources of noise or types of complaint, seasons of the year or nights of the week and any period between 11pm and 7am as they deem appropriate to local circumstances.

From October 2006, extending the provisions of the Noise Act 1996 will include licensed premises, including temporary licenses. This will enable a local authority to serve warning notices where they suspect that noise from licensed premises exceeds the specified permitted noise levels. If after a short warning period the noise continues to exceed the permitted levels, the Council will be able use their discretion to decide whether to prosecute in the Magistrates court, where fines of up to £5000 may be levied, or to serve fixed penalty notices of £500. The new powers also mean that the revenue from the fixed penalty notices will be retained by the local authorities to help fund the service, rather than passed directly to the Treasury, as is the current situation.

Fixed penalty notices for night noise from licensed premises will be a useful, additional tool for dealing quickly with temporary non-persistent noise problems that, whilst disturbing to nearby residents in the short term, are not of sufficiently negative impact to warrant the use of statutory nuisance under the Environmental Protection Act 1990 or the powers to close licensed premises temporarily under the Anti-Social Behaviour Act 2003. This measure will give local authorities an extra option to take a phased enforcement approach to dealing with night noise from licensed premises, and tailor enforcement to the severity and impact of a noise disturbance.

The Noise Act 1996 uses a noise protocol for calculating the accepted level. It is an offence to cause a night noise above the permitted level once a warning has been issued. The current noise protocol was developed for domestic night noise. Noise from licensed premises may be of a different nature, so a different noise protocol might be more appropriate. This project informs the development of an appropriate noise measurement protocol for licensed premises and Defra will be consulting on the protocol in summer 2006, before the measure is implemented.

At present, the Noise Act 1996 only applies to noise from dwellings and there are concerns that its existing noise level measurement protocol and criteria might not be well suited to entertainment noise from licensed premises. Consequently, Defra commissioned Capita Symonds Ltd and BRE to jointly study methods and criteria for assessment of entertainment noise from licensed premises.

This study comprised extensive laboratory testing of the correlation of 18 variations of 9 different noise measurement methodologies and criteria, with the subjective response of a representative group of ordinary members of the public; field testing of the practicability of EHPs using these methodologies and criteria for the assessment of entertainment noise from licensed premises. The laboratory experiments deliberately constrain some independent and confounding variables in order to test the parameters of interest to the experiment. The conclusions should always be viewed with the understanding that controlled experimental testing cannot, by its nature, model all combinations of variables that exist in the field.

Whilst the primary objective of the study has been to identify which of the methodologies and criteria tested were best suited for assessment of entertainment noise from pubs and clubs late at night, of equal importance is the requirement that they are practicable for EHPs to enforce and are fair and realistic for licensees to comply with.

The outcomes of the study have been as follows:

A. The majority of the members of the public reported the ability to tolerate a modest degree of intrusive audible entertainment noise in their home late at night for a “one-off” occurrence (i.e. occurring at intervals of less than six months), and that the onset of audibility of the entertainment noise did not equate to a threshold of acceptability for intrusive entertainment noise.

B. The majority of EHPs also reported that a modest degree of intrusive entertainment noise from a “one-off” occurrence was acceptable, and that the onset of audibility of the entertainment noise did not equate to a threshold of acceptability for intrusive entertainment noise in such circumstances. EHPs also reported that a lesser degree of intrusive entertainment noise was acceptable for more regular occurrences (i.e. once a week), and that for either scenario the onset of audibility of the entertainment noise did not equate to a threshold for enforcement action for intrusive entertainment noise in such circumstances.

C. The results of the laboratory testing identified several methodologies and criteria, which gave reasonably good correlation with subjective response.

D. The noise metric that provided the best overall prediction of subjective ratings of all the entertainment noise types tested by ordinary members of the public was the Absolute LAeq.

E. However, during the field testing it was apparent that the “highest performers” from the laboratory testing all had clear disadvantages in use under real world conditions, so there is no clear best option for recommendation which combines optimum correlation with subjective response with ease and rapidity of use. The following options are considered the best of the available options, in descending order of correlation with subjective response, each raising different issues regarding practicability of use by EHPs.

· Absolute LAeq – That is an LAeq,5min noise level value set at a single action level. However an intrusive entertainment noise criteria based on Absolute LAeq, would be difficult to use where the existing ambient noise level without the entertainment noise was close to, equal to or above the action level. Therefore, we would recommend an action level Absolute LAeq, with an additional subjective requirement that the entertainment noise itself has a clearly audible (to an otologically normal listener) contribution to the overall noise e.g. the songs/tracks would be recognisable to a listener familiar with the music and any words intelligible. In terms of an action level, a table in this report is provided showing various levels of entertainment noise used in the laboratory testing and the responses of test subject’s. In the context of this study’s objective to determine criteria that represents a clearly unacceptable situation, the noise levels at which test subjects felt the noise was “just unacceptable” for a one off event within a habitable room with windows closed was at 34 dB LAeq,5 minute. The range for the first two scores of unacceptability was LAeq,5 minute 34 to 37 dB. Analysis of data from the 2000/2001 National Noise Incidence Study (NNIS) indicates that with windows closed, only a small percentage of the UK population (5.5%) are estimated to have internal ambient noise levels above LAeq,8 hour 34 dB and just 2.1% above LAeq,8 hour 37 dB.

· LA90 – LA90 (no music) – That is the difference between the LA90,5 min noise level with the intrusive entertainment noise and the equivalent LA90,5 min with no intrusive entertainment noise. This allows consideration of the background level, but requires a measurement without intrusive entertainment noise that may not be possible on the night of a complaint. This in itself may be problem enough to make the metric unusable for “one-off” events or as a quick response to a problem.

· LAeq – LA99.95 or existing Noise Act methodology (LAeq – LA99.8). These metrics include some consideration of the underlying noise level at the same time as any offending noise level is measured, without requiring a separate “no music” measurement to be made. The former is slightly more effective in prediction of subjective response than the latter, but not substantially so, and using the latter has logistical advantages. The performance of both these noise metrics was less good than the previous two options, but they also avoid the practical disadvantages highlighted above.

2 introduction

Existing Scope and Specification

The original purpose of this project was to scrutinise 9 different methods for assessing the impact of entertainment noise from pubs and clubs at night and to develop an appropriate rating method to complement the application of the provisions of the Noise Act 1996 to licensed premises. In the course of the study the number of variations of different assessment methods scrutinised increased to 19.

This study only relates to entertainment noise including amplified music, singing and speech and “beam back” of sports TV broadcasts sourced from inside and within the curtilage of pub and club type licensed premises.

This project focussed on the assessment of noise from infrequent and one-off entertainment activities operating between 2300 and 0700 hours, with a view towards determining which methods are best suited to gauging the impact of such noise on persons trying to sleep or trying to get to sleep.

Any methodology developed will be mindful of the alternative legislation already available for use by EHPs e.g. The Licensing Act 2003, Anti-social Behaviour Act 1998 as amended, and the statutory nuisance provisions of the Environmental Protection Act 1990.

The original scope to this study identified that whilst the preferred outcome of this project was a single methodology and criterion, this might not be practicable in all situations and it was possible that a matrix of methodologies and criteria would cover a wider range of circumstances more effectively than a single methodology and criterion.

Project Objectives

The primary objective is to provide a set of criteria and methodologies best suited for assessment of entertainment noise from pubs and clubs at night, to complement the application of the provisions of the Noise Act 1996 to licensed premises.

However, of equal importance to the primary objective is the requirement that any final recommended criteria and methodologies are practicable for EHPs to enforce and are fair and realistic for licensees to comply with.

Any recommended methodology must clearly establish and minimise the uncertainties with noise measuring; by quantifying the method, clearly identifying the assessment process, provide guidelines on the utilisation of equipment and highlight the applicability of the method to appropriate circumstances.

It would also be ideal if the final recommended methodology and criteria were able to settle or at the very least establish the limits of the long on-going subjective versus objective assessment of the music noise debate, and streamline the enforcement procedures available to EHPs.

Laboratory and Field testing

To achieve the project objectives, entertainment noise was assessed in laboratory tests at the Building Research Establishment during January 2006 with a total of 60 experimental subjects, selected from a cross section of the public. The tests were undertaken in houses rather than listening rooms, partly because it enabled testing of noise sources from within the building and from outside the building, and also because it created a more realistic environment for subjects. The laboratory test methodology and assessment is detailed within section 5.

The primary objective of the field tests was to assess the practicability of using the assessment methods identified in the lab-testing phase as having the best correlation with subjective response to entertainment noise, in real world conditions. The field tests aimed to cover variables such as urban and rural environments with a good geographical spread within the UK, with significantly different background noise climates. Furthermore, the venues were selected such that they included locations where the music was a recognised noise problem, borderline and was considered to provide acceptable noise levels. The field trials were undertaken at 10 venues around the UK and these provided a good range of venues (pubs, clubs and town halls), locations (urban, towns and rural) and types of music. At each venue there were at least 2 EHPs that completed the questionnaires regarding their perception of the noise both inside and immediately outside each selected residential location. The field test methodology is detailed within section 6.

A glossary of acoustic and statistical terms that have been used within this report are shown in Appendix A.

3 legislation

This contract represents Phase II of the project to define measurement methods and criteria to support the implementation of section 84 and Schedule 1 of the Clean Neighbourhoods Act 2005 which apply the provisions of the Noise Act 1996 to licensed premises. Phase 1 of the project is discussed in the next section of this report.

This report seeks to satisfy the contract requirement to provide a set of criteria and methodologies for assessment of entertainment noise from pubs and clubs to complement the application of the provisions of the Noise Act 1996 to licensed premises. Currently the Noise Act only applies to noise from domestic premises between 23:00 and 07:00 hrs and the existing noise level measurement protocol was developed with domestic noise sources in mind. Given the potential difference in noise sources and intensity of noise from licensed premises compared to domestic premises it was thought that the existing Noise Act noise level measurement protocol may not match particularly well with the subjective impact of noise from licensed premises. Consequently, this research contract was let to test the correlation of the subjective response to noise from licensed premises against a range of noise level measurement criteria and methodologies (including the existing Noise Act measurement protocol). This was completed under laboratory and field conditions in order to determine a set of practicable measures, which could be applied to noise from licensed premises.

The current Noise Act 1996 measurement methodology is not intended to provide a universal fine tuned benchmark of the acceptability of noise from domestic premises or for the assessment of statutory nuisance (see sections 7 and 8 of defra circular NN/31/03/2004). Instead the current Noise Act 1996 measurement methodology and criteria are aimed at providing a robust and sustainable threshold, which if exceeded indicates a significant unacceptable noise problem. It is also possible that noises that are not a Noise Act 1996 offence may nevertheless be prejudicial to health or a nuisance and so be a statutory nuisance. This study intends to continue this approach such that the final methodologies and criteria recommended represent a degree of intrusive noise which is clearly unacceptable in most circumstances and is not intended to represent a threshold of acceptability in all circumstances.

In commenting on the new powers in the April 2006 edition of Environmental Health Practitioner, the defra Local Environment Minister Ben Bradshaw made the point that existing powers were not being replaced or supplanted by the new powers and the new powers were merely extending the range of enforcement options available to local authorities. He also made the following comments in regard to the extension of the Noise Act 1996 to licensed premises in particular:

“Changes to the the licensing laws in 2005 mean that licensed premises can be open later and for longer hours. Any such changes need to be supported by enforcement powers so that anti- social behaviour, where it occurs, is not tolerated.”, and;

“Currently, noise can be addressed through statutory nuisance under the Environmental Protection Act, and, for night noise from domestic dwellings between the hours of 2300 and 0700, under the Noise Act 1996; this provides for noise levels to be set and it is an offence for the occupier of a domestic dwelling to exceed these once he has been warned. However, from October 2006 the night-noise provisions of the Noise Act 1996 will be extended to include licensed premises, including temporary licenses. This means that once the person responsible for licensed premises has been warned about excessive noise, it will be an offence, punishable by a fine of up to £5,000 or a fixed penalty of £500, to exceed the set noise level.”, and:

“Section 82 of the Clean Neighbourhoods and Environment Act 2005 makes it an offence for a person committing a Noise Act offence to fail to give his name or address when asked do so, or to give a false or inaccurate response. There is a fine on summary conviction for this offence of up to £1000.”

It could be argued that existing legislation is adequate to deal with noise from licensed premises. The short answer to this argument is yes in most circumstances, but there is a small gap regarding noise from licensed premises when there is clearly a problem but which occurs infrequently and for relatively short periods. In order to illustrate this gap it is necessary to understand the range of powers already available to deal with noise from licensed premises and their advantages and disadvantages.

Statutory Nuisance: Environmental Protection Act 1990 – This allows a Local Authority to serve an abatement notice where noise emitted from premises constitutes a statutory nuisance and prosecute the recipient and /or seize noise making equipment if the nuisance continues after the expiry of the time for compliance given on the notice. These powers can be used for “one-off” type noise problems that arise late at night (See East Northamptonshire District Council -v- Brian Fossett [1994] Env LR 388), and can be used against both private and public nuisance (see below). The ability to prove nuisance using subjective descriptors can be useful in dealing with subtly intrusive low level noise that is difficult to measure or distinguish from extraneous noise levels, but which becomes a problem due to its regular occurrence and or extended occurrence at noise sensitive times. However, it can time to enforce an abatement notice, particularly if the recipient exercises their right to appeal the notice and/or contests any prosecution for non-compliance.

Licensing Act 2003 – Where noise from licensed premises is causing a “public nuisance” a Local Authority can call for the review of a premises license, and if public nuisance is proven they can condition, constrain or revoke a license. The potential neutering or loss of a premises license is a significant deterrent with serious commercial consequences. However, in the context of noise from licensed premises this process relies on the noise causing a “public nuisance” which has a specific legal meaning as articulated in the case of Attorney General v PYA Quarries Ltd. (1957) 2 QB 169. The issue in this case was whether quarrying activities that affected the neighbourhood was a private nuisance affecting some of the residents only (which would have been actionable at civil/common law by the individuals affected), but not a public nuisance affecting all Her Majesty's subjects living in the area. In his judgment Romer LJ. concluded “that any nuisance is 'public' which materially affects the reasonable comfort and convenience of life of a class of Her Majesty's subjects. The sphere of the nuisance may be described generally as 'the neighbourhood'; but the question whether the local community within that sphere comprises a sufficient number of persons to constitute a class of the public is a question of fact in every case. It is not necessary, in my judgment, to prove that every member of the class has been injuriously affected; it is sufficient to show that a representative cross-section of the class has been so affected for an injunction to issue." Denning LJ. agreed "that a public nuisance is a nuisance which is so widespread in its range or so indiscriminate in its effect that it would not be reasonable to expect one person to take proceedings on his own responsibility to put a stop to it, but that it should be taken on the responsibility of the community at large." All of which suggests that noise from licensed premises which adversely affects only one household is unlikely to be a public nuisance actionable under the Licensing Act 2003 review powers (although statutory nuisance powers would apply). Additionally the license review process is as yet largely untested and is likely to be a relatively complex process involving extensive paperwork, committee and appeal hearings and take several weeks if not months to complete.

The Anti-Social Behaviour Act 2003/Licensing Act 2003 – This allows a local authority (Anti-Social Behaviour Act) or the Police (the Licensing Act), although in practice this will almost certainly be a joint exercise between a local authority and the Police rather than action by one agency only, to shut for 24 hours a licensed premises which is emitting noise sufficient to constitute a public nuisance. Again the issue of public nuisance comes into play and the remedy requires intensive enforcement resource and is only likely to be justified in the context the Human Rights Act 1998 in cases of severe wide spread public nuisance.

Consequently, there is a need for a methodology for dealing with noise from licensed premises that is easily implemented, efficient and quick whilst remaining workable for EHPs and sits well within a phased and proportionate approach to enforcement, as required by the Human Rights Act 1998 in regard to the rights of both the noise victim and the noise maker i.e. it’s quick, effective and not too onerous.

The extension of the Noise Act 1996 to cover licensed premises is intended to fill any gap in the existing legislation whereby the enforcement of an Abatement Notice for statutory nuisance or the calling in a license for review can be complex and involve weeks, even months of processing and any resolution of the problem can be further delayed by appeals or elaborate legal manoeuvres. Additionally, the provisions of the Anti-social behaviour legislation to empower the Police and Local Authorities to immediately close licensed premises causing significant public nuisance will be overly draconian and dis-proportionate to the harm caused for most entertainment noise problems, which whilst significant for the sufferer tend to be transient and affect relatively few persons.

The application of the Noise Act 1996 procedures to licensed premises provides a rapid and relatively easily implemented enforcement option, which fits comfortably in a phased proportionate hierarchy of increasingly stringent enforcement actions that can be implemented in succession; or at the relevant point in the hierarchy of enforcement actions depending on how serious the noise problem is and /or how laissez faire the attitude of the noise maker.

A hypothetical example of how the new power could be used and how it complements existing powers is as follows:

A local authority receives a complaint of excessive noise from a licensed premise after 23:00 hrs.

Appropriately authorised local authority staff visit the complainant and determine that the noise is an offence under the Noise Act 1996, it is also a statutory nuisance and is likely to be causing disturbance to a number of different households.

The local authority staff issues a warning notice explaining that the noise from the licensed premises is exceeding the Noise Act 1996 limits.

Although statutory nuisance is being caused, the local authority staff decide to defer serving an abatement notice under the new powers to do so under the Clean Neighbourhood and Environment Act 2005, in order to allow the person holding the premises license or the designated premises supervisor (DPS) or in absence of the DPS, the person who appears to be in charge of the premises the reasonable but not extended time stated on the warning notice to resolve the problem. If there is evidence of the statutory nuisance having occurred intermittently but infrequently in the past or that the statutory nuisance is likely to recur in future the local authority could also decide to serve an abatement notice at the same time as issuing the Noise Act warning notice.

If after service of the warning notice the problem is resolved, the case can be closed.

However, if after expiry of the reasonable but not extended time period for compliance on the warning notice the noise is not reduced to less than the Noise Act limits for licensed premises, the local authority can either decide to prosecute the recipient of the warning notice in the Magistrates Court at a later date or to serve a fixed penalty notice (FPN) for £500 on the person holding the premises license or the designated premises supervisor (DPS) or in the absence of the DPS, the person who appears to be in charge of the premises at the time of the offence as a means of discharging their liability for prosecution.

If after service of the FPN the noise problem is quickly resolved and does not recur before 07:00 hrs on the same day as the FPN is served the matter can be closed.

However, if after service of the FPN the noise continues or recurs soon afterwards in a manner that is a statutory nuisance and or another Night Noise Act offence, the local authority can:

A. enter the premises and seize noise making equipment.

and/or

B. serve an abatement notice (or enforce any previously served abatement notice).

and/or

C. prosecute under the Noise Act 1996 or serve a new FPN if a new Noise Act offence is committed.

and/or

D. set in place steps to review the premises license.

4 phase one

As previously discussed, this study is phase 2 of the overall project. Phase 1 was entitled “Noise from Pubs and Clubs – Phase 1: NANR 92” and completed in November 2005. The full final report can be viewed at http://www.defra.gov.uk/environment/noise/research/pubs-clubs-phase1/pubsclubs-phase1.pdf . Phase 1 of the project can be summarised as a detailed literature review of research into noise from pubs and clubs and current custom and practice in assessing such noise across England and Wales. Phase 1 of the project did not recommend or comparatively assess the value of the methods or criteria for assessing noise from pubs and clubs that it identified, but it did make recommendations as to further “validation” of various methods and criteria for assessing noise from pubs and clubs.

An outcome from phase 1 of the project was to develop a table of candidate methodologies and criteria for assessment of noise from pubs and clubs for further testing and comparative assessment to determine their effectiveness for assessing noise from pubs and clubs. This table is reproduced below:

Name

Parameter

Type

IoA working group annex

LAeq vs. LA90 plus L10 vs. L90 in 40-160 Hz 1/3 octave bands

Relative

BS 4142/Noise Act 1996

LAeq vs. background (L A90, L A99, etc.)

Relative

Noise Rating curve

1/3 octave (Leq, L10 or Lmax) vs. NR curve

Absolute

Absolute LAeq

LAeq

Absolute

DIN 45680/Moorhouse

10-160 Hz 1/3 octave Leq vs. reference curve

Absolute

Inaudibility

Subjective

Relative

In this study the list of candidate methodologies and criteria for assessment of noise from pubs and clubs was significantly expanded to a total of 18 variations on 9 separate methods. These methodologies are detailed within section 5 of the report.

Phase 1 of the project commented that:

“The only sensible way to develop an optimal rating method is by constructing tests in which listeners are exposed to pub and club noise and are asked to subjectively rate the noise in some way (annoyance, loudness or audibility, for example). An audio recording of the noise can then be analysed for the kinds of physical features identified above. If a computer is used to do this, a large number of objective rating schemes (and variants of them) can be compared. The result of each objective rating scheme can then be compared to the subjective responses. The extent to which a particular rating scheme matches the subjective responses can be quantified with statistical methods and the best rating scheme identified.”

This has formed the basis for the laboratory and field research carried out in this study.

Phase 1 of the project also provided commentary on the use of laboratory and field testing of different assessment methods and criteria as follows:

“Laboratory testing produces more reliable judgements from subjects and gives more control over the sound fields being heard, but inevitably lacks some of the context in which the original sound might be heard by a particular listener. A proper investigation of the physical features of pub and club noise can be carried out much more easily in the laboratory, because sounds can be altered to vary one physical parameter at a time and parameter values can be set precisely. It is also the only way to accurately quantify the difference in response from a group of listeners exposed to exactly the same sound. (The variance across subjects gives a valuable indication of the error to be expected when using the final rating method to predict listener response to noise from a pub.) Finally, it is easier to build confidence in lab tests because they are more easily refined through pilot tests. Therefore, the main plank of the validation programme should be based on laboratory tests.”

Consequently, this study has primarily used laboratory testing to determine the correlation of the subjective response of a representative sample of the population with the methodologies and criteria described above; and then used field trials to determine how practical it is to use those criteria to assess noise from pubs and clubs in the real world.

5 Laboratory Testing

Methodology

BRE test Facility

Laboratory tests were carried out at BRE during January 2006.

The decision was made to use houses rather than listening rooms. This was partly because it enabled testing of noise sources from within the building and from outside the building, and also because it created a more realistic environment for subjects. Two identical houses were used, each of which have three upstairs bedrooms. One test subject was in each bedroom.

Figure 1 - Test houses used for laboratory testing

Each test bedroom contained a table and a chair, at which the subjects were instructed to sit, and a bed. In addition, a noise analyser (Norsonic 121) was used to monitor noise levels in each room for the duration of each test. The noise analysers continuously recorded a large range of noise indicators, including LAeq, LCeq and 1/3-octave Leq spectra each 125 ms. These short time interval measurements then allowed various noise indicators, including statistical indicators to be calculated for each noise condition.

The microphone for each noise analyser was positioned in the central area of the room away from the window and at least 1 m from all reflecting surfaces. Noise measurements were taken in each room of all noise indices, both with and without the entertainment noise.

Figure 2 - Typical bedroom in test house

Types of Music and Sound System

5.1 The sound system was chosen to be representative of a typical high performance system that would be installed in a club. The system comprised of full range JBL cabinets supplied by Harman Pro UK, who have extensive experience at supplying these types of systems in numerous clubs. The equipment comprised 2 JBL AM6215 full range units, 2 JBL ASB6128 bass units and 2 IT6000 amplifiers.

Four different noise types were used for the laboratory testing, namely:

A. Noise Type A, Guitar Orientated Rock – this style of music typically operates with peak low frequency noise levels around the 63 Hz to 125 Hz octave bands, and a developed and extended frequency spectra with additional peaks at mid to high frequency

B. Noise Type B, Modern Dance Music - "House" and “Drum & Bass” and other modern dance music types have a reputation for persistent virtually non-stop low frequency bass thump, often peaking around the 63 Hz octave band, sometimes with significant energy in the 40 Hz and 50 Hz 1/3 octave bands, and then a pronounced and steep drop off in levels at mid to high frequency in the spectrum.

C. Noise Type C, Non-music entertainment noise - Sport noise is not uncommon in pubs and bars, an example being a football match being shown on a large TV or video screen. The spectrum of this type of noise typically has a relatively flat profile with modest peaks in lower frequency octave bands.

D. Noise Type D, Karaoke - the vocal content (of varying subjective quality) is often emphasised over the backing music compared to other music types and this sort of entertainment can be played at relatively high levels. The vocal element tends to be significant and the frequency spectrum of this type of noise typically has peaks at 63 Hz or 125 Hz octave bands and also in the mid frequency range of 500 Hz to 2 KHz octave bands.

Each noise type was presented to the test subjects at five different levels. These were subjectively described as:

1. inaudible to an average listener,

2. just audible to an average listener,

3. a noise level which is plainly audible i.e. the content of the noise is communicated to the test subject so they can recognise its type (music or speech etc) but the content is not intelligible;

4. a noise level which is clearly audible i.e. the noise is communicated so that the content is intelligible to an average person and subjects can make out words and recognise songs/tracks etc;

5. a noise level that an average listener might describe as loud.

Five of the tests (total of 30 subjects) had the source of the noise in the ground floor of the test houses, to simulate structure-borne transmission from a noise source within the same building. In another five tests, the noise sources were outside the test houses, to simulate airborne transmission from a noise source outside of the building.

Recruitment of test subjects

Each test could be attended by six subjects – one in each of the six bedrooms of the two houses. In order to maximise the statistical analysis, a sample of 30 subjects was recruited to experience each noise stimulus. Having fewer subjects than this would mean working with small-sample statistics, and while valid, these may not have provided the necessary robustness for this project. Therefore, a total of 10 tests were carried out, with a total of 60 experimental subjects. As discussed, 30 of these experienced the internal-source transmission scenario, and 30 experienced the external-source transmission scenario.

Subjects were recruited from an existing BRE database of people who have previously expressed an interest in taking part in testing.

Subjects were selected according to a number of criteria, in order to cover a broad mixture of the population. The pre-selection questionnaire was designed to obtain information on the age and sex of volunteers, any level of hearing impairment, and a confirmation of their ability to hear and understand instructions, and to read and complete questionnaires

Age: subjects were assigned to three broad age bands: 18-34 years, 35-54 years and 55 years and over.

Hearing impairment: The objective was to recruit subjects with normal hearing. Therefore, the only people considered for testing were those with no hearing impairment or with a mild hearing impairment. Most of the applicants declared no hearing impairment.

For each test, then, a balanced group of subjects were selected.

E. One male aged 18-34

F. One female aged 18-34

G. One male aged 35-54

H. One female aged 35-54

I. One male aged 55 or over

J. One female aged 55 or over

Subjects received a telephone reminder on the day of their test. In addition, a list of available reserves was held for each age/sex category, so that if a subject dropped out for any reason, substitutes were available.

Subjects received a standard information sheet about the tests (this did not include a technical description of the testing being carried out) before the test, and a more detailed briefing once they arrived at the BRE site. The on-site briefing notes are shown in Appendix B.

In order to avoid any confounding effect between the rooms used and the age and sex of subjects, subjects were allocated to the six test rooms by a prepared plan. Each house and test room, therefore, had a balance of male and female, and a balance of age groups during the testing.

Questionnaires

Two questionnaires were used during the testing and these are discussed below:

The first was a single-page questionnaire that was used at the end of each noise segment, i.e. 20 times during the testing (4 types of entertainment and at 5 noise levels). This questionnaire is shown in Appendix B. It includes a series of questions on various aspects of environmental comfort, including noise. These are necessary so that any confounding effects of other comfort factors can be controlled for in the analysis. The remaining questions refer to the entertainment noise – have they heard any in the last few minutes, whether what they heard would affect their activities at home, and an overall acceptability rating for the noise. This last question is the primary target variable for the analysis.

The second questionnaire was presented after the noise segments were finished, and asked for some background information about the subjects, including their general attitudes to noise and entertainment noise, and some information about their normal exposure to entertainment noise. This questionnaire is also shown in Appendix C.

The questionnaires were drafted by BRE and finalised after comments from Defra and from CSL.

Experimental Protocol

One test per day was carried out on the dates 16-20 January and 23-27 January. The tests in the first week used internal noise sources i.e. the sound systems in the ground floor of each test house with transmission of entertainment noise through the floor/ceiling to the test bedrooms above. The tests in the second week used external noise sources i.e. the sound systems were placed externally and transmission of entertainment noise was through the façade with closed thermally insulated glazing.

Testing took place in the late evening. This was to make the tests as psychologically and physiologically realistic as possible. Previous work has shown differences in responses between daytime and evening testing, and as the main research questions in this project relate to the impact of entertainment noise at home during the evening and night, moving the testing into the evening seemed an obvious and necessary decision.

The approximate timescale was as follows.

20:30Latest time for subjects to arrive at BRE

20:30Subject briefing begins

20:45Subjects are escorted to the test houses and shown to their rooms.

20:55Testing begins

Thereafter, the testing follows a 5-minute cycle for each of 20 noises. In the last 90 seconds of each 5-minute cycle, subjects complete the Test Questionnaire. The questionnaires are collected just before the next noise segment begins.

22:35Noise segments are complete, post-test questionnaire is distributed

22:45Subjects finish and are escorted back to the main building and car park.

The experimental schedule was run by one researcher in each house using stopwatches synchronised with the PC controlling the noise tracks for each test.

All subjects experienced all noise types at all levels, which equated to a total of 20 segments (4 noise types x 5 noise levels). Each subject only experienced one transmission scenario (either internal source or external), as incorporating both scenarios would make testing very onerous for subjects and likely to prejudice the quality of the responses. The order of presentation for the noise types and levels was balanced as far as possible. During each test, the segments for each noise type were grouped together, so subjects would hear all five levels of each noise type before moving to a different one. The order of levels within each noise type were randomised and balanced throughout the five internal-source tests, and the five external-source tests. The complete order of presentation is shown in Appendix B.

Local residents

BRE was keen to ensure that the external-source tests did not produce an unacceptable level of noise at nearby housing (the nearest dwelling being around 100m from the test houses). During pre-testing, the noise level was recorded at the site boundary, and for one noise segment, dance music at the highest level, it was considered that the maximum noise level may cause concern to local residents.

As a result of the pre-testing the local Environmental Health department was informed of the testing and the time schedule, in case any problems were reported. A letter was also delivered to all residents close to the test site, informing them of the testing, and assuring them that all testing would be completed well before 11pm.

In the letter, residents were invited to contribute to the research by completing a questionnaire during the week if they were interested in doing so. Nine residents accepted this invitation, and they were provided with a diary questionnaire to complete if (and only if) they were aware of noise from the BRE site during the week. An extract of this questionnaire is shown in Appendix C.

A member of BRE staff, previously an Environmental Health Officer, monitored noise levels in nearby roads during the testing.

Results (Noise Measurements)

During each laboratory test noise levels were recorded continuously in each test room. This enabled checking for any anomalies in noise levels. The noise analysers continuously recorded a large range of noise indicators, including LAeq, LCeq and 1/3-octave Leq spectra each 125 ms. These short time interval measurements then allowed various noise indicators, including statistical indicators to be calculated for each noise condition.

Two further test runs were carried out (one for airborne and one for structure-borne transmission) with no subjects in the test rooms. Data from these tests were used to calculate a number of different noise indicators.

Each noise indicator was calculated separately for each room and each noise condition (combination of airborne/structure-borne transmission, noise type and level). This gave a total of 40 noise conditions for each of the six rooms.

Calculation of noise levels

The following noise indicators were calculated for each noise condition and room:

K. Institute of Acoustics Draft Good Practice Guide on the Control of Noise from Pubs and Clubs – Annex 1: Criteria and Measurement Guides

· A-Weighted – difference between LAeq of entertainment noise (logarithmic subtraction of LAeq with entertainment noise and LAeq without entertainment noise) and LA90 without entertainment noise

· 1/3rd Octave – difference between LA10 of entertainment noise (calculated as logarithmic subtraction of LA10 with entertainment noise and LA10 without entertainment noise) and LA90 without entertainment noise in 1/3rd-octave bands between 40 Hz and 160 Hz. The maximum value in any of these bands was then used for the analysis

· Maximum value of each of the above

L. Noise Act / BS4142

· Noise Act methodology: LAeq – LA99.8 with entertainment noise present for both. It should be noted that this was based on a total 5 minute measurement from which both the LAeq and LA99.8 measurements were extracted, rather than the 15 minute period allowed in the current measurement protocol.

· BS4142 methodology: LAeq – LA90 with entertainment noise not present for LA90 measurement

M. Noise Rating (NR) Curves

· NR based on Octave Band Leq measurements

· NR based on Octave Band L10 measurements

· NR based on Octave Band L90 measurements

· NR based on Octave Band Lmax measurements

N. Absolute LAeq

O. The Moorhouse modification of the DIN 45680 methodology for investigating low frequency noise

· Maximum exceedance of 1/3rd-octave band Leq measurements over reference curve in the range 12.5 Hz to 160 Hz

P. C-Weighted

· Absolute LCeq· LCeq – LC99.8 with entertainment noise present for both

· LCeq – LC90 with entertainment noise not present for LC90 measurement

Q. Comparative L90

· LA90 (with entertainment noise) – LA90 (without entertainment noise)

· LC90 (with entertainment noise) – LC90 (without entertainment noise)

R. Short temporal averaging (using Leq,125ms measurements to assess the quietest period with entertainment noise on – e.g. the quietest 125 ms being the L99.95 for a 5-minute measurement)

· LAeq – LA99.95· LCeq – LC99.95

S. Inaudibility (assessed directly from responses to questionnaires)

An Excel spreadsheet was developed for use in both laboratory and field trials to calculate all of the above 18 noise metrics (excluding inaudibility as this was assessed from the questionnaire responses) for each noise condition (and for each measurement/assessment in the field trials), to ensure that the same calculation approach was used in all cases.

Analysis

Figure 3 and 4 show LAeq and LCeq noise indicators for each noise condition (respectively for the structure-borne and airborne transmission arrangements). It can be seen that the C-weighted indicator is less sensitive to noise types C and D (sports event and karaoke), where there is less low-frequency noise.

Figure 3 - LAeq and LCeq indicators for tests with structure-borne transmission

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

Noise Type (A to D) and Nominal Level (1 to 5)

L

eq

(dB)

12345

Noise Type A

12345

Noise Type B

12345

Noise Type C

12345

Noise Type D

L

Aeq

L

Ceq

Figure 4 - LAeq and LCeq indicators for tests with airborne transmission

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

Noise Type (A to D) and Nominal Level (1 to 5)

L

eq

(dB)

12345

Noise Type A

12345

Noise Type B

12345

Noise Type C

12345

Noise Type D

L

Aeq

L

Ceq

The background noise levels in the test rooms were in the range 19 to 24 dB LAeq (18 to 23 dB LA90)

The equivalent C-weighted background levels were in the range 38 to 41 dB LCeq (33 to 39 dB LC90)

The raw data from noise measurements in each room (each of the 18 noise indicators for each noise condition) are included in Appendix D to this report.

Results (Questionnaire)

Subjects

Two subjects dropped out on the day due to illness, and these were replaced by subjects of the correct age/sex category from the reserve list. A full set of 60 subjects therefore took part in the testing. The average age across all subjects was just under 45 years and most subjects lived in mainly residential areas. More than half rated themselves on the sensitive half of the scale in terms of general sensitivity to noise. It is interesting to note that although all but one subject was recruited as having no hearing impairment, four subjects rated themselves as having a mild hearing impairment in the test questionnaire.

The responses given to the background information questionnaire are given in Appendix E.

Testing

The distribution of responses was examined, and extreme and outlier responses were identified. Figure 5 shows the box plots for responses on acceptability, for each noise type and level. As a brief key to the box plots, the boxes themselves indicate the bounds of the upper and lower quartiles of the responses, and the bold line in the middle is the median response. The circle points are outliers, and the star points are extreme values. Outliers and extreme values were removed from further analysis. In the case of sound D, one respondent told us that she had completed the first few iterations of the questionnaire using the scale in reverse, and that explains most of the very high extreme values for the lowest sound levels.

Figure 5 - Responses on acceptability of the overall noise level by test condition

If subjects marked that they had heard entertainment noise, they were asked about what impact the noise would have on their daily activities. The responses to these questions are shown Appendix F, with responses shown against nominal noise level. It is clear from these charts that entertainment noise at levels 1, 2 and 3 would not be considered a major disruption by most subjects, but that at level 4, the disturbance is becoming more obvious.

Analysis

Non-parametric equivalents of the one-way ANOVA and independent T-test were carried out – tests used include Kruskal-Wallis and Jonckheere-Terpstra - and these showed no significant differences in responses for source type (internal versus external) or sound type. The same tests showed a strong significance in response for noise level, as would be expected. A median test (see Table 1) showed the responses pooled into three groups - Group 1 (levels 1 and 2), Group 2 (level 3) and Group 3 (levels 4 and 5). This means that levels 1 and 2 were not significantly different to each other, but were significantly different to 3, 4 and 5. Level 3 was significantly different to 1 and 2, and significantly different to 4 and 5. Levels 4 and 5 were not significantly different to each other, but were significantly different to 1, 2 and 3.

Table 1 - Median tests on noise test levels

Frequencies

Test Level

1

2

3

4

5

Q4: Acceptability of the overall noise level

> Median

15

36

102

191

215

<= Median

225

204

133

46

21

Test Statistics (b)

Q4: Acceptability of the overall noise level

N

1188

Median

3.00

Chi-Square

551.417(a)

df

4

Asymp. Sig.

.000

a 0 cells (.0%) have expected frequencies less than 5. The minimum expected cell frequency is 110.6.

b Grouping Variable: Test Level

Analysis of potentially confounding variables

The laboratory experiments deliberately constrain some independent and confounding variables in order to test the parameters of interest to the experiment. In “real life” situations, there will be variables that could make the conclusions made here from the laboratory tests less relevant, and the conclusions and recommendations should always be viewed with the understanding that controlled experimental testing cannot, by its nature, model all combinations of variables that exist in the field.

Tests were carried out on potential confounding variables to determine if they had an effect on the level of acceptability of entertainment noise. The variables tested are listed below:

T. Gender

U. Age

V. Test houses

W. Test room

X. Types of location

Y. Locations

Z. Types of property

AA. Whether or not entertainment noise can be heard indoors at home

AB. Level of annoyance with entertainment noise heard at home indoors during the evening and night

AC. Level of hearing impairment

Significant differences were only observed in test D on Test room.

A Kruskal-Wallis test shows that the difference in the level of acceptability of entertainment noise between the rooms is only just significant, see Table 2. Examination of the ranks suggests that room 4 may be significantly different (sound considered to be less acceptable) to rooms 3 and 6. This seems likely to be a spurious result rather than a meaningful one.

Table 2 - Kruskal-Wallis test on differences between rooms

Ranks

Test Room

N

Mean Rank

Q4: Acceptability of the overall noise level

1

189

549.71

2

195

579.08

3

180

612.59

4

193

522.41

5

190

549.78

6

190

603.09

Total

1137

Test Statistics(a,b)

Q4: Acceptability of the overall noise level

Chi-Square

11.092

df

5

Asymp. Sig.

.050

a Kruskal Wallis Test

b Grouping Variable: Test Room

The data for noise metrics versus acceptability for the four different sound types are shown in Table 3. The results indicate that the best across-the-board metric was the Absolute LAeq. This metric had the strongest two correlations for all four noise types. No other metric had such a consistent predictive performance with subjective response. For the other noise metrics, different ones were strong for different noise types - e.g. the C-weighted metrics only made a showing for the dance music, IOA A-weighted (LAeq of the entertainment noise minus LA90 without entertainment noise) for the sports and karaoke.

Table 3 - Spearman’s rho coefficients for acceptability of level of sound and noise metrics

 

Spearman's Rho

Metric

All Sounds

Sound A (Rock)

Sound B (Dance)

Sound C (Sport)

Sound D (Karaoke)

IOA A-weighted

0.765

0.801

0.779

0.698

0.769

IOA Max 1/3rd octave exceedance

0.598

0.724

0.728

0.550

0.657

IOA max exceedance

0.621

0.724

0.728

0.632

0.680

Noise Act LAeq-LA99.8

0.732

0.763

0.757

0.682

0.741

BS4142 LAeq-LA90(no music)

0.756

0.800

0.762

0.697

0.750

NR Leq

0.735

0.800

0.752

0.648

0.724

NR L10

0.776

0.811

0.804

0.698

0.775

NR L90

0.529

0.769

0.397

0.468

0.469

NR Lmax

0.576

0.633

0.623

0.280

0.739

Absolute LAeq

0.781

0.828

0.784

0.720

0.781

Moorhouse max exceedance

0.499

0.717

0.761

0.107

0.540

LCeq

0.571

0.762

0.774

0.264

0.687

LCeq-LC99.8

0.399

0.660

0.766

-0.213

0.493

LCeq-LC90(no music)

0.534

0.739

0.766

0.148

0.582

LA90-LA90(no music)

0.761

0.815

0.778

0.702

0.754

LC90-LC90(no music)

0.610

0.777

0.734

0.430

0.587

LAeq-LA99.95

0.745

0.776

0.756

0.699

0.753

LCeq-LC99.95

0.397

0.655

0.771

-0.250

0.493

Note: The Noise Act assessment was based on a total 5 minute measurement from which both the LAeq and LA99.8 measurements were extracted, rather than the 15 minute period allowed in the current measurement protocol.

Key for correlation tables:

#

: largest significant correlation

#

: 2nd largest significant correlation

#

: 3rd largest significant correlation

All of the correlations are significant, and almost all are positive, indicating a strong positive linear relationship between the level of noise measured (as indicated per metric) and acceptability of the level of noise. Figure 6 shows this relationship for Absolute LAeq.

A partial correlation is the correlation between variables that remains after controlling for (e.g. partialling out) one or more other variables. Partial correlations were carried between Absolute LAeq and subjective acceptability controlling for the effect of the following variables:

AD. Thermal comfort

AE. Rating of temperature

AF. Air movement

AG. Air quality - dryness

AH. Air quality - freshness

AI. Air quality - odour

AJ. Air quality overall

AK. Noise – from outside of the building

AL. Noise – from building systems

AM. Noise – from within the building (other than from building systems)

AN. Noise overall (everything that can be heard)

AO. Comfort (with consideration of environmental factors A. to K.)

Factors J to L were found to have some effect on the relationship between level of acceptability and Absolute LAeq.

Correlation without controlled variables

0.770

Correlation controlling for noise from within the building

0.674

Correlation controlling for noise overall

0.501

Correlation controlling for comfort overall

0.598

The zero-order or uncontrolled correlation indicates that the higher the actual noise is, the less acceptable people find the level of noise. Comfort overall, noise overall and noises from within the building all appear in part to moderate this positive linear relationship.

These results are however, not unexpected. It has been shown that the less comfortable people are, the less tolerant they are. Comfort takes into account all aspects of the environment, including sound. A comfort rating would therefore by its nature reflect some of the variation in level of acceptability. This is also true of the two noise ratings that had an effect. Ratings for noise overall requires respondents to consider everything that they can hear, which includes the test sounds. The remaining rating focuses on noises from within the building. In 50% of the tests the source of the sound was from within the building. In the remaining tests the source of the sound was outside of the building. However, many subjects said that they did not realise that this was the case and rated the sounds they heard as though were being generated from within the building.

The graph below shows a plot of mean acceptability score for each Absolute LAeq value experienced. The linear regression line through the data explains around 76% of the variance in the data. This graph can be used to determine appropriate noise targets.

Figure 6 - Acceptability ratings in the laboratory for Absolute LAeq

50.00040.00030.00020.000

Absolute LAEQ

10.00

8.00

6.00

4.00

2.00

0.00

Mean Acceptability Score

R Sq Linear = 0.763

The second best predictive performance in the laboratory testing was LA90 minus LA90 (no music). For completeness, the chart below shows the scatter plot of acceptability vs. for the laboratory data. The variance explained by this regression is much lower than with Absolute LAeq, and it is clear by inspection that the regression line is a poorer predictor of the acceptability rating.

Figure 7 - Acceptability ratings in the laboratory by LA90 minus LA90 (no music)

30.00020.00010.0000.000

LA90-LA90(no music)

10.00

8.00

6.00

4.00

2.00

0.00

Mean Acceptability Score

R Sq Linear = 0.605

Inaudibility

One of the suggested noise metrics tested was inaudibility, i.e. an assessment where acceptability is presumed to be linked with inaudibility.

The chart below shows the frequencies at each acceptability rating, split by whether or not the subject reported hearing the entertainment noise. It is clear that in many cases, subjects who were able to hear the entertainment noise nonetheless considered it to be acceptable.

This indicates that an assessment method based on inaudibility would significantly underestimate the acceptability ratings of the people experiencing the noise. The data are further split by noise level, and this chart is shown in Appendix F.

Clearly

unaccept

able

987Just

unaccept

able

Just

acceptabl

e

432Clearly

acceptabl

e

Q4: Acceptability of the overall noise level

300

200

100

0

Count

No

Yes

Q2: Were you able to

hear any entertainment

noise in the last few

minutes?

Conclusions from the laboratory testing

The noise metric that appears to provide the best prediction of subjective response across the board for different entertainment noise types is the Absolute LAeq. This noise metric provided consistently high correlation coefficients when compared with the subjects’ ratings of acceptability.

According to the regression between subjective acceptability rating and noise level in absolute LAeq, the table below shows that LAeq levels associated with each value of acceptability. For example, therefore, if the objective is that the new criterion reflect the level at which householders feel the noise is “just unacceptable”, the target absolute LAeq,5 minutes should be 34.0 dB.

Table 4 - Semantic descriptor and associated value of acceptability

Semantic descriptor

Score

Absolute LAeq,5 minutes

Clearly acceptable

1

17.0

2

20.4

3

23.8

4

27.2

Just acceptable

5

30.6

Just unacceptable

6

34.0

7

37.4

8

40.8

9

44.2

Clearly unacceptable

10

47.5

6 Field trials

The primary objective of the field trials was to assess the practicability of using the assessment methods identified in the lab-testing phase as having the best correlation with subjective response to entertainment noise, in real world conditions. The field trials aimed to cover variables such as urban and rural environments, a good geographical spread within the UK, and significantly different background noise climates. Furthermore, the venues were selected such that they covered the following noise issues:

A. Having a recognised noise problem

B. Being borderline

C. Having an acceptable noise climate

The field trials were undertaken at 10 venues around the UK and a summary of the type of venue, type of music, locations and perception of the acceptability of the noise is shown in Table 5 below:

Table 5 – Summary of Field Trial Venues

Venue and LocationVenue ClassificationPerception of Noise from Pub/ClubType of music

1. London ClubUrbanBorderlineDance

2. Bristol PubUrbanProblemRecorded pop

3. Bristol ClubTownBorderlineDrum & Bass

4. Milton Keynes ClubUrbanBorderlineDrum & Bass

5. Milton Keynes PubUrbanAcceptableRecorded pop

6. North Herts Town HallRuralProblemLive Band

7. Sheffield ClubUrbanBorderlineDance

8. Nottinghamshire Miners ClubRuralAcceptableRecorded pop

9. Skipton, North Yorkshire PubTownBorderlineRecorded pop

10. Skipton, North Yorkshire PubRuralProblemLive Band

The locations for the 10 field trials provided a good range of venues (pubs, clubs and town halls), locations (urban, towns and rural) and types of music.

The equipment used for the field trials was the same as that used for the laboratory tests and comprised a Norsonic 121 noise analyser, set to record as a minimum set of measurements the following parameters and indices:

D. Broadband A and C weighted noise levels as Leq, L1, L10, L90, L99.8, L99.5, and L99 indices over 5 minute, 2 minutes and 1 minute periods

E. Low frequency noise levels in the 1/3 octave bands from 12 Hz to 250 Hz as Leq, L1, L10, L90, L99.8, L99.5, and L99, indices, over 5 minute, 2 minutes and 1 minute periods

F. Leq 125 millisecond continual time history

An assessment questionnaire was drafted by CSL with input from BRE, to be used by EHPs. This questionnaire is shown in Appendix G. The questionnaire requests that the EHPs consider the noise impact for a “one-off” or infrequent event lasting 2 hours and occurring after 23:00 hrs, e.g. once every 6 months. While it is possible that the context given to EHPs could change their responses somewhat, it was considered to be necessary to give a context to respond within. The differences between individuals responses could be exaggerated if they are also assessing on very different imagined contexts. For the analysis that was carried out, it is likely that the responses would be slightly different for different scenarios, but it is considered that the differences would not significantly alter the conclusions.

At each venue there were at least 2 and in most cases 3 EHPs that completed the questionnaires regarding their perception of the noise both inside and immediately outside each selected residential location. Where possible, EHPs from neighbouring Local Authorities were used to provide an opinion as well as the local EHPs, as existing noise issues at the venue may have affected their judgement of the noise measurements during the specific assessment periods.

Background noise measurements were generally taken immediately prior to the music being played at ‘normal’ operational levels within the pub/club. These measurements were taken both directly outside the residential property and in the worst affected habitable room.

Where possible, the noise measurements with the music noise being assessed were taken after 2300 hours. However, this was not always possible as some of the pubs closed at 2300 hours. In some cases it was practical to undertake the noise assessment soon after completion of the background noise measurements. In these circumstances the EHPs were instructed to assess the music noise as if the event was after 2300 hours.

The field trials presented many practical difficulties. The first was arranging access to residential properties that were affected by music noise. The following methods were used to determine suitable venues:

G. Using CSL’s existing Environmental Health and Club/Pub contacts (provided 8 out of the 10 venues).

H. Project Website – received visits and provided information regarding the project but provided no venues for the field trials.

I. Articles in both the weekly EH News and monthly EHP magazines, included the web site address but provided no venues for the field trials.

J. Environmental Health message board, EHNet (provided 2 venues) and Barbour Index.

Some venues were offered by Local Authorities and then withdrawn due to imminent service of notices or prosecutions. Some venues were withdrawn due to staffing difficulties within the Local Authorities that had offered the venues. Generally, the EHPs that offered the venues were able to provide colleagues from neighbouring local authorities to provide their opinions on the noise at the residential properties and to complete the questionnaires.

At the Sheffield venue, the noise assessment was undertaken within neighbouring offices rather than residential accommodation. However, this venue was considered to be ideal for the assessment and the EHPs were requested to assess the noise levels as if the offices were habitable residential rooms.

Methodology

To ensure consistency at each of the 10 field trials the following methodology was adopted.

At the majority of the noise measurement locations a 5 minute background noise measurement was undertaken both inside and outside the selected residential property prior to the start of the music event at the venue. The external levels were taken 1m from the façade of the same elevation of the residential property as the room in which the internal measurements were taken. The internal background noise measurements followed the same methodology as the laboratory tests whereby the windows were closed and the microphone was located close to the centre of the room and at least 1m away from any reflecting surfaces. These measurements were taken as close to the time of the music events as possible. In some cases, such as at Milton Keynes, the background survey was undertaken on a different evening but at the same time of day as when the music events took place.

The noise measurements with the entertainment noise present followed the same procedure and were measured at the same locations as for the background noise measurements.

Briefing documents were provided to the residents and licensees for each field trial. These documents detailed the project objectives and provided information regarding the laboratory and field trials. A further document was given to the participating EHPs, which requested the existing entertainment noise assessment criterion and enforcement methods used by the Local Authorities that participated in the trials. These documents are reproduced in Appendix J.

Prior to any music noise assessments/measurements, the Entertainment Noise Assessment questionnaires were distributed and discussed with the EHPs. This provided the EHPs with the opportunity to understand the type of questions that would be asked at the end of each individual assessment. It was requested that the questionnaires were not completed until the end of the 5 minute assessment period to take into account any variations in the noise over the 5 minute period. Furthermore, the noise from questionnaires being completed would have affected the noise levels being measured at the same time.

The wording on the questionnaires was explained to the EHPs to ensure that the first page was completed for one-off events (e.g. once every 6 months) that last for at least 2 hours and occur after 2300 hours. Where the assessment was undertaken before 2300 hours it was requested that the EHPs assess the music noise as if the event occurred after 2300 hours. The second page of the questionnaire was also discussed with the EHPs to ensure that the assessment was considered as a regular event lasting 2 hours and occurring after 2300 hours e.g. once a week.

On completion of the assessments the tests were repeated if either the source noise levels could be significantly varied of if there was another habitable room where the assessment could be repeated (this was a useful exercise when there was direct sound transmission via a party wall or floor from the venue to the residential receptor, as it was usually relatively easy to move to a room in the dwelling less affected by the noise from the licensed premises).

The completed questionnaires and the associated 5 minute noise measurements were then sent for analysis to BRE. Where possible, 15 minute samples were also taken of noise levels in the test dwelling, although this was difficult to achieve in practice as it required the full co-operation of the residents in the late evening period.

Results

The noise data was entered into the BRE spreadsheet, to ensure that the calculations were made on the same basis as the data from the laboratory testing.

A total of 14 EHPs were involved in the assessments, making a total of 75 assessments for 10 venues. Between 1 and 4 assessments were made at each venue at different locations. Dates of assessment were between 3/2/06 and 1/4/06. Times of assessments were between 21:57 and 23:45.

Analysis

A summary of descriptive analyses from the field questionnaires is given in Appendix H. There was quite a large variation in EHP response to the assessment questions.

Acceptability is rated by the EHPs on a 10 point scale, where 1 is clearly acceptable, 5 is just acceptable, 6 is just unacceptable and 10 is clearly unacceptable. Table 6 shows the mean scores for acceptability of the entertainment noise heard, when judged as a one-off or infrequent event, and when judged as a regular event. As might be expected, noise levels were more likely to be considered acceptable when judged as one-off events than when judged as regular events.

Table 6 - Acceptability ratings of entertainment noise by EHPs judged as one-off and regular events.

Venue

How would you rate the current level of entertainment noise overall….

Q5: … for a one-off event?

Q11: … for a regular event?

1. London Club

4.25

5.75

2. Bristol Pub

4.83

6.00

3. Bristol Club

2.75

4.75

4. Milton Keynes Club

2.50

3.38

5. Milton Keynes Pub

5.63

6.25

6. North Herts Town Hall

9.50

10.00

7. Sheffield Club

7.38

8.75

8. Nottinghamshire Miners Club

3.25

3.88

9. Skipton, North Yorkshire Pub (town)

4.67

6.67

10. Skipton, North Yorkshire Pub (rural)

7.33

9.33

Table 7 shows the correlation coefficients between the responses on acceptability made by EHPs to entertainment noise, when considered as a one-off event. The range of coefficients was greater than for the laboratory testing data, with some metrics giving very weak correlations. Absolute LAeq, was the strongest predictor of householder response in the laboratory testing, and it provides a fair but not strong prediction of EHP response in the field. The strongest predictor of EHP response across all assessments in the field was LA90 minus LA90 (no music).

Table 7. Spearman’s rho coefficients for acceptability of level of sound and measures of sound, for entertainment noise judged as a one-off event

 

Metric

How would you rate the current level of entertainment noise overall, for a one-off event?

All Measures

Inside measures

Outside measures

IOA A-weighted

0.511

0.375

0.651

IOA Max 1/3rd octave exceedance

0.676

0.813

0.309

IOA max exceedance

0.676

0.813

0.309

Noise Act LAeq-LA99.8

0.110

0.261

-0.195

BS4142 LAeq-LA90(no music)

0.518

0.526

0.578

NR Leq

0.403

0.304

0.322

NR L10

0.531

0.602

0.435

NR L90

0.434

0.425

0.365

NR Lmax

0.245

-0.057

0.532

Absolute LAeq

0.507

0.508

0.491

Moorhouse max exceedance

0.547

0.552

0.472

LCeq

0.545

0.628

0.520

LCeq-LC99.8

0.260

0.166

-0.112

LCeq-LC90(no music)

0.732

0.798

0.564

LA90-LA90(no music)

0.757

0.833

0.713

LC90-LC90(no music)

0.714

0.714

0.621

LAeq-LA99.95

0.128

0.302

-0.195

LCeq-LC99.95

0.303

0.290

-0.028

Note: The Noise Act assessment was based on a total 5 minute measurement from which both the LAeq and LA99.8 measurements were extracted, rather than the 15 minute period allowed in the current measurement protocol.

Key for correlation tables:

#

: largest significant correlation

#

: 2nd largest significant correlation

#

: 3rd largest significant correlation

The equivalent table for entertainment noise judged as a regular event is shown below.

Table 8 - Spearman’s rho coefficients for acceptability of level of sound and measures of sound , for entertainment noise judged as a regular event

 

Metric

How would you rate the current level of entertainment noise overall, for a regular event?

All Measures

Inside measures

Outside measures

IOA A-weighted

0.461

0.330

0.561

IOA Max 1/3rd octave exceedance

0.604

0.766

0.260

IOA max exceedance

0.604

0.766

0.260

Noise Act LAeq-LA99.8

0.079

0.225

-0.225

BS4142 LAeq-LA90(no music)

0.446

0.470

0.460

NR Leq

0.399

0.255

0.451

NR L10

0.529

0.572

0.530

NR L90

0.442

0.422

0.441

NR Lmax

0.231

-0.093

0.574

Absolute LAeq

0.498

0.461

0.562

Moorhouse max exceedance

0.523

0.483

0.543

LCeq

0.549

0.628

0.588

LCeq-LC99.8

0.274

0.207

-0.109

LCeq-LC90(no music)

0.665

0.739

0.477

LA90-LA90(no music)

0.679

0.786

0.622

LC90-LC90(no music)

0.599

0.639

0.467

LAeq-LA99.95

0.092

0.259

-0.225

LCeq-LC99.95

0.324

0.329

-0.004

Note: The Noise Act assessment was based on a total 5 minute measurement from which both the LAeq and LA99.8 measurements were extracted, rather than the 15 minute period allowed in the current measurement protocol.

In general, the correlation coefficients between EHP rating of acceptability and noise metrics were fairly weak. This is probably due to the individual variation in response between the EHPs having a greater influence on the outcome of the statistical analysis, which is probably partly due to the lower sample size of EHPs (n=14) in the field tests compared to the larger sample of members of the public (n=60) used in the laboratory tests. Additionally extraneous noise e.g. road traffic and people in the street etc. had a much greater confounding impact on the measured noise levels at some of the field test sites compared to the laboratory tests, where the influence of extraneous noise on the measurements was minimal.

It should be noted that LA90 minus LA90 (no music) requires a second measurement, as the metric requires measurement with and without the presence of the entertainment noise. This makes it a less practical tool for environmental noise assessment, particularly when dealing with enforcement for one-off events on the night in question. Missing values (where the LA90 (no music) measurement was not possible to arrange) may also affect the results for this measure.

Additional comments provided by EHPs on the questionnaires are shown in Appendix I.

Comparison of laboratory and fieldwork data

As different noise metrics had the strongest correlations in laboratory compared to field testing, some comparisons were made between the two sets of data. The strongest correlation in the laboratory testing was the Absolute LAeq. The figure below shows the fieldwork acceptability ratings (for a one-off event) plotted against the measured Absolute LAeq during the field tests. The green points are the fieldwork assessments, and the green line is the regression line through those points. It is clear that there is a great deal of variation between assessments, and the variance explained by the regression line is fairly low, at just over 40%. For comparison, the red line is the regression line for the same relationship for the laboratory data. The regression lines follow a similar trend. Note that this chart only includes those field assessments made inside a property, as it is only appropriate to compare those ratings with the laboratory ratings.

Figure 8 - Scatter plot of field assessments of acceptability as a one-off event (inside ratings)

y = 0.2907x - 5.2053

R

2

= 0.4146

0

2

4

6

8

10

12

14

0.00010.00020.00030.00040.00050.00060.000

Absolute LAeq

Acceptability

Acceptability for a one-off event Absolute LAeq Predicted Acceptability Score Linear (Acceptability for a one-off event)

The correlation between noise metrics and acceptability scores in the field was strongest for LA90 minus LA90 (no music). The figure below shows the scatter plot and regression line for acceptability ratings against that metric. The variance explained is better than for Absolute LAeq, but still not strong. The red line shows the regression line of acceptability vs. LA90 minus LA90 (no music) from the laboratory testing. It is clear that the regression lines are more similar for laboratory and field testing using this metric, but at the expense of losing some strength in the predictive power of the laboratory testing. The variance explained by this regression line is high (at over 80%) but this is at least partly an artefact of the shape of the data, with a large cluster near the origin of the graph, and a smaller cluster higher up.

Figure 9 - Scatter plot of field data of acceptability as a one-off event vs. LA90 minus LA90 (no music)

y = 0.4441x + 2.7377

R

2

= 0.8137

0

2

4

6

8

10

12

14

16

-5.0000.0005.00010.00015.00020.00025.00030.000

LA90 - LA90(no music)

Acceptability

Acceptability for a one-off event LA90-LA90(no-music) Predicted Acceptability Score Linear (Acceptability for a one-off event)

The regression lines for laboratory tests and field tests as shown in Figures 8 and 9 are not identical, but they are fairly similar, at similar levels and with similar gradients.

7 Stakeholder Involvement

Website Questionnaires

An online questionnaire was developed for interested parties to input information to the project. The questionnaire was designed to allow a variety of people to answer, from members of the general public to licensed trade professionals and Environmental Health practitioners. Depending on their answers to the initial screening questions, respondents would take a different route through the possible questions.

A flow chart for the questionnaire routing, and screenshots of the questions themselves, are shown in Appendix C.

The link to the questionnaire was included on the CSL web pages about the project and the articles in EH News and EHP magazine. To date, only a very small response (4 completions) have been received. The link was sent out very recently in a BRE e-newsletter, in hopes of attracting more interested feedback.

8 assessment laboratory and field testing

The performance of each of the proposed noise metrics is summarised in the table below. This includes the correlation coefficients with acceptability ratings of test subjects in the laboratory testing, as well as judgements on practicality, ease of comprehension and use.

Table 9 – Summary of performance of each noise metric, lab and field tests

Metric

Laboratory

Field aspects

Correlation with acceptability ratings of subjects in laboratory testing

Practicality for use in the field

Ease of understanding and use

IOA A-weighted

0.765

Medium

High

IOA Max 1/3rd octave exceedance

0.598

Low

Low

IOA max exceedance

0.621

Low

Low

Noise Act LAeq-LA99.8

0.732

Medium

Medium

BS4142 LAeq-LA90(no music)

0.756

Medium

Medium

NR Leq

0.735

Low

Medium

NR L10

0.776

Low

Medium

NR L90

0.529

Low

Medium

NR Lmax

0.576

Low

Medium

Absolute LAeq

0.781

High

High

Moorhouse max exceedance

0.499

Low

Low

LCeq

0.571

High

Medium

LCeq-LC99.8

0.399

Medium

Low

LCeq-LC90(no music)

0.534

Low

Low

LA90-LA90(no music)

0.761

Low

Medium

LC90-LC90(no music)

0.610

Low

Low

LAeq-LA99.95

0.745

Medium

Medium

LCeq-LC99.95

0.397

Medium

Low

Note: The Noise Act assessment was based on a total 5 minute measurement from which both the LAeq and LA99.8 measurements were extracted, rather than the 15 minute period allowed in the current measurement protocol.

Basis for assessment of noise metrics

The comparison of assessments between laboratory and field testing is interesting, and it is reassuring where the assessments