Performance and Acceptability of the TfL Urban Bus Sound (AVAS) Eduardo Manzano Fontecha 1 Anderson Acoustics Grant Lewis Waters 2 Anderson Acoustics Thomasin Stuart 3 AECOM

ABSTRACT Transport for London's quiet-running bus fleet is required to include an Acoustic Vehicle Alerting System (AVAS) compliant with UNECE Regulation 138 and as part of a wider series of safety measures as set out in their Bus Safety Standard towards the London Mayor's 'Vision Zero' objectives. As a public body, TfL must ensure that the Urban Bus Sound AVAS implementation is fit-for-purpose from a safety standpoint, is considerate of driver working conditions and is consistently implemented across a number of vehicle types, manufacturers and fleet operators. Additionally, as quiet-running vehicles increase across the London bus network, it must consider the impact on the urban soundscape. Over the course of two years, Phase 2 of the project addressed practical implementation and quality assurance challenges, bringing together soundscape and acoustic testing, on-street public and key stakeholder surveys, workshops with multiple vehicle manufacturers and operators as well as engagement with bus drivers. The project concludes by providing an optimised AVAS solution with specification and vehicle checks documentation to ensure successful roll-out. The project provides an important view on considerations for how AVAS can be implemented on public transport vehicle fleets, including the need to consider and work with a wide range of public and stakeholder requirements.

1. INTRODUCTION

Anderson Acoustics, AECOM and Zelig Sound were commissioned by Transport for London (TfL) to undertake works as part of the Acoustic Vehicle Alerting System (AVAS) Phase 2 project. The project works were separated into Lots A, B and C with AECOM commissioned to undertake Lot A (Engagement), Anderson Acoustics commissioned to undertake Lot B (Sound) and Zelig Sound commissioned to undertake Lot C (Sound Design).

1 eduardo@andersonacoustics.co.uk

2 grant@andersonacoustics.co.uk

3 thomasin.stuart@aecom.com

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

2. BACKGROUND

To align with the London Mayor’s Vision Zero objectives, TfL created the Bus Safety Standard to establish a plan for implementing a range of safety measures to the bus fleet, which included a target of ‘Acoustic Conspicuity’ [1, 2]. Acoustic Conspicuity in relation to bus vehicles is the ability for road users to be aware of the presence and operation of a vehicle to sufficiently avoid risk of collision. TRL were commissioned to investigate the practical measures to achieve acoustic conspicuity for quiet-running bus vehicles, such as Electric Vehicles (EVs), Hybrid Electric Vehicles (HEVs) or Hydrogen vehicles, and their 2018 report concluded that the implementation of an AVAS was required for quiet-running vehicles [3]. Phase 1 of the project was commissioned by TfL to create a suitable Urban Bus Sound, aimed at providing a universal AVAS sound for implementation across the Greater London region, as well as to be implemented as a universal bus sound for quiet-running vehicles worldwide. The priority of this phase was to create a sound that met the UN ECE Regulation 138 performance requirements and ensured that the needs of Vulnerable Road Users (VRUs) and the bus drivers were met through extensive stakeholder engagement and involvement with the design and testing process [4]. The conclusions of Phase 1 were the selection of a final Urban Bus Sound, a successful compliance test with Regulation 138, and a trial vehicle implementation on the Route 100 fleet. A particular innovation of the Urban Bus Sound was that it comprises two sound elements, layered together, the Core sound and the Beacon sound. The Core is a broadband sound that is played continuously when the vehicle is turned on (whether it is stationary or moving) and the Beacon is a soft impulsive element or ‘ping’ that plays when the vehicle is in motion. The Beacon element was developed as a direct response to vision impairment group engagement and collaboration. 3. PHASE 2 OBJECTIVES

The overarching purpose of the Phase 2 works was to finalise an operational performance standard of the AVAS and implement measures that will ensure its consistency and acceptability as it is rolled-out across the TfL bus fleet. The steps taken to achieve this were the implementation of a Responsive AVAS that adapts the sound output level to respond to the ambient sound level of the vehicle’s location, the confirmation of acoustic conspicuity performance in various environments, the assessment of acceptability of AVAS sound ingress into the driver’s cabin and the revision of the AVAS performance specification and vehicle checks documentation to support manufacturers and operators to maintain optimal AVAS performance across the TfL bus fleet. 4. WORK PACKAGE 1: RESPONSIVE AVAS INTEGRATION

The concept of a Responsive AVAS was first discussed in the TRL report as a method to both support the optimisation of the sound output level of the AVAS to ensure acoustic conspicuity whilst reducing noise pollution impacts [3]. Due to concerns with variability and robustness of microphone-based approaches, it was proposed to implement a Responsive AVAS using city-wide noise modelling data combined with other geospatial datasets on street type and classification as part of the GPS used for the Intelligent Speed Assistance (ISA) provided by Actia. ISA is geospatial mapping system that automatically limits the speed of a vehicle based on a road’s speed limit. Anderson Acoustics worked in partnership with sub-contractor Tranquil City, an environmental data company, to deliver suitable geospatial datasets for the implementation of the Responsive AVAS via the Actia ISA system.

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

The UK Department for Environment, Food and Rural Affairs (Defra) Strategic Noise Mapping (2017) modelling output data for the London region was obtained to provide the basis for the street classification [5]. A series of steps were undertaken to confirm the suitability of this data for the London road network and correct it for any changes in traffic volumes since the creation of the data and ensure its relevance. This was originally planned to be completed via roadside acoustic surveys, but unfortunately due to the coronavirus pandemic restrictions and traffic reductions this was not possible. However, similar roadside monitoring data conducted by Anderson Acoustics on various road types between February and early March 2020 on behalf of Heathrow Airport Ltd was kindly shared for this purpose [6]. This, in addition to other sources on aircraft noise and TfL Street Type classifications (based on Healthy Streets Approach) were used to produce a cross-validated method of classifying the street network segments for the annual average daytime and night-time ‘Noise Reference Level’. Five discrete AVAS sound output levels were used, termed ‘Steps’ 0-4, ranging between 60 dB to 72 dB L Amax,F at 2 m from the vehicle in 3 dB increments. The Noise Reference Level was then used to determine the AVAS Step for each street segment based on three options for acoustic conspicuity, +5 dB (Option 1), +0 dB (Option 2) and +3 dB (Option 3). Option 3 was selected by the project team as being the most suitable for implementation as a balance between ensuring acoustic conspicuity and limiting noise pollution impact, and a geospatial dataset was produced based on the Integrated Transport Network (ITN) for system implementation. The geospatial dataset was integrated into the Actia ISA mapping system with no major issues and a partnering firmware configuration was produced by Forman Vehicle Services Ltd on their AVAS unit to respond to the AVAS Step value according to sound output level specifications. During the implementation and trial process, it was found to be important to define a specific source address channel for the correct communication between the ISA and AVAS units on the AVAS software configuration. Additionally, an unrelated issue of inconsistency of sound output level was found on some vehicles, which was determined to be caused by speaker connection issues and was resolved by the bus manufacturer, Alexander Dennis Ltd (ADL). The Responsive AVAS initial configuration was implemented on a series of ADL vehicles operating on Route 100 in early 2021 and on Optare, Wrightbus and Caetano vehicles manufacturers in summer 2021 to conduct in-situ testing and checks across a range of vehicle types. 5. WORK PACKAGE 2: ACOUSTIC CONSPICUITY & SOUNDSCAPE IMPACT

The objective of this work package was to demonstrate whether the Responsive AVAS configuration allows for lower volumes to be used whilst maintaining acoustic conspicuity and to identify the maximum sound level of operation that is considered suitable to reduce noise pollution and increase public acceptability of the AVAS. Inherently, the acoustic analysis itself can only predict acoustic conspicuity and acceptability to a certain extent, and therefore the conclusions of this section are intended to be read in conjunction with Lot A public and stakeholder engagement results. The TRL report on acoustic conspicuity recommends that an AVAS is created for the TfL bus fleet that “is identifiable by a wide range of the population”, “meets the current regulations on minimum sound”, “is not annoying or irritating to the majority of the population” and “can easily be

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

distinguished from other sounds, such as background sounds and other traffic” [3]. These objectives were used as a basis for the acoustic analysis conducted, where it was used to determine if the sound could provide information to the road users, especially concerning VRUs, on a vehicle’s Presence, Direction, Location and Operation [7]. To appraise the likelihood of audibility of the Responsive AVAS in-situ as the bus approaches and departs, the sound pressure level is assessed when the bus is in various urban environments and distances from the measurement position. The measurement position in all measurements is at a bus stop but may be at varying distances from the front of the vehicle when it stops. The typical ambient sound level, L eq,T per 1/3 octave band, is subtracted from the specific sound level of the vehicle with the Responsive AVAS passby, L eq,100ms per 1/3 octave band, for each of the environment scenarios. This typical ambient level is derived from the mean average of all ambient sound level spectra measurements without the presence of the buses to provide a consistent baseline reference noise level. The assessment is based on achieving +5 dB above the ambient or background sound level as per BS 5839-8:2013 [8]. Each road passby event was presented in the form of ‘signal to noise’ ratio spectrograms, highlighting where the +5 dB criterion was achieved. An example of this is shown in Figure 1 for a passby event of an EV Route 100 vehicle with Responsive AVAS.

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

Figure 1: Route 100 Responsive AVAS signal-to-noise ratio compared to ‘quiet’ ambient sound level environment for arrival, idle and departure. Purple areas denote the audibility criteria is met (+5 dB) and grey areas denote that the signal-to-noise ratio is between 0 to +5 dB, which indicates

marginal potential for audibility. The analysis found that the Beacon element met audibility criteria at distances of 8m or higher in quiet, moderate and busy environments on a vehicle’s approach. The Core element was found to be marginally below audibility criteria in quiet and moderate environments and not clearly identified in busy environments. It was found that the Responsive AVAS Urban Bus Sound provided a distinct sound and frequency profile compared to that of ICE (Internal Combustion Engine) vehicles that can aid the identification and association of the sound with a bus. This was particularly due to the tonal and pulsating characteristics of the Beacon and reductions in low frequency dominance typically found in ICE vehicles. The analysis indicated that to increase the Responsive AVAS’s performance for localisation, an increase in frequency content at 3000 Hz and above should be made, in both Beacon and Core elements. The AVAS was found to provide a 1% per km/h frequency shift and therefore meet requirements of UN ECE Regulation 138 on speed detection.

However, despite an idle sound only being an optional requirement of Regulation 138, the Core sound during ‘idle’ operation did not achieve minimum audibility criteria in the test environments and therefore did not meet TfL performance expectations. As a result, it is recommended that the Core sound is increased in amplitude in relation to the Beacon to increase road user detection of a vehicle awaiting departure. A comparison of the Responsive AVAS was made to existing ICE vehicles to determine the impact on noise pollution and public acceptability. It was found that the Responsive AVAS implementation could provide a significant benefit in reducing overall sound levels due to bus vehicles on quieter streets across the London network. During the daytime, the Responsive AVAS was shown to provide a 10 dB overall reduction in sound levels compared to ICE buses and a 5 dB reduction compared to if a fixed level AVAS, i.e. averaged over the different 3dB adaptive bands, was implemented. During the night-time, this benefit increased to 11 dB reduction compared to ICE vehicles and 6 dB reduction compared to a fixed AVAS solution. Benefits were also noted in moderate environments during both daytime and night-time compared to both ICE vehicles and a fixed AVAS solution. In busy environments, the results highlight a risk of sound levels marginally increasing when the AVAS operates on the highest Step level (Step 4). Therefore, it is recommended that to limit any impact in busy environments, especially during the night-time, the highest operational level of the Responsive AVAS is limited to Step 3 or 69 dB(A). Psychoacoustic metrics, including loudness and sharpness, have been analysed on both Responsive AVAS and ICE vehicles to investigate how the sound content may affect the public perception of the AVAS. The analysis shows that the Responsive AVAS has a lower perceived loudness rating than an equivalent ICE vehicle during a passby event. In real terms, this means that the Responsive AVAS is expected to be perceived quieter, despite having an equivalent or higher decibel level. This is due to the differences in sound energy across the frequency spectrum for the two vehicle types. Furthermore, the Responsive AVAS was shown to have a lower sharpness rating compared to an ICE vehicle, which indicates that the AVAS is likely to be perceived as less ‘harsh’ sounding and therefore less intrusive. When considering sound ingress of the Responsive AVAS on nearby residential properties, the Responsive AVAS was found to reduce maximum sound events (L Amax,F ) inside a property by 10 dB or above when compared to ICE vehicles. Maximum sound events, especially from road traffic noise, have been linked to increased risk of noise related health impacts [9]. The results indicate a significant benefit on sound conditions because of surface transport vehicles, when considering the wide-scale implementation across the Greater London region. Sound ingress calculations indicate a prominent tone due to the Beacon element that may cause a risk of disturbance when operating at Step 4. Therefore, it is further recommended that Step level 4 is considered for removal to limit the Responsive AVAS operational limit to Step level 3 or 69 dB L Amax,F . 6. WORK PACKAGE 3: DRIVER CABIN SOUND LEVELS

The objective of this work package was to measure and assess driver cabin sound conditions directly due to sound ingress of the Responsive AVAS implementation to discuss acoustic comfort. As with WP2, due to the subjective nature of acoustic comfort, the results of this section are meant to be read alongside the Lot A driver engagement findings (Work Package 4).

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

Due to the necessary position of the AVAS installation at the front of the vehicle, sound ingress of the AVAS sound during operation to the driver’s cabin may be an acoustic comfort constraint in the current configuration, which is subject to the physical limitations of the front cabin and, consequently, different for each bus design. UN ECE Regulation 138 does not outline any restrictions on sound ingress into the driver’s cabin or the vehicle interior. However, as driver comfort and suitable operating conditions are priorities of TfL to ensure staff, customer and road user safety, it is committed to establishing performance requirements that ensure these conditions are met in vehicles with the Responsive AVAS. A review of standards related to acoustic comfort and noise level conditions demonstrated that there are limited standards of practice or guidance in this area. Speech Interference Level or SIL outlined in BS EN ISO 9921-2003 was considered the most appropriate metric to predict acoustic comfort and to apply a robust assessment method for the AVAS specification [10]. This was considered appropriate due to the critical nature of speech communication for safe vehicle operation. ‘Good’ SIL performance is targeted with an upper limit of ‘Fair’ (contribution from the AVAS Urban Bus Sound only) to ensure comfort and to account for an allowance for non-native speakers and listeners. To achieve the above SIL performances, the total ambient sound level in the driver cabin, between 500Hz to 4kHz, should be:

• Good – 39 dB to 45 dB L SIL ; • Fair – 45 dB to 50 dB L SIL . An upper limit of 60 dBA at the driver’s cabin, produced exclusively by the AVAS sounders, is also recommended to limit low and high frequency sound content outside the mid-range frequencies accounted for in the L SIL calculation. Acoustic monitoring of driver sound conditions during operation was conducted on a range of bus routes and vehicle types, including Responsive AVAS EVs and HEVs, quiet-running EVs (without AVAS) and diesel ICE vehicles to understand the sound levels experienced in each during various times of the day. Due to the nature of monitoring, sound ingress due to environmental sound and vehicle operational sounds (such as iBus communication) were present. Therefore, the monitoring of EVs without AVAS was critical to understanding the contribution of these sounds, as well as detailed notes taken by consultants on the significance of sounds not associated with the AVAS during measurement periods. The results demonstrated that environmental noise ingress was the dominant source of sound contribution in the driver’s cabin. However, results did show that the background sound level, L A90 , increased by up to 7 dB in Responsive AVAS vehicles compared to the EVs without AVAS. This was reinforced by the finding that there was an increase in L 90 levels between 250 Hz to 1 kHz for Responsive AVAS vehicles compared to EVs without AVAS. Due to the dominance of environmental noise ingress shown, background sound level results are considered the most representative measure of Responsive AVAS sound contributions. Results indicate that driver cabin sound levels in Responsive AVAS electric vehicles are on average 3-4 dB lower than in ICE vehicles with notably reduced sound content in the 250 Hz, 500 Hz and 4

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

kHz frequency bands. The Responsive AVAS electric vehicles had slightly increased sound content at 1 kHz which is expected to be due to the AVAS operation and potential resonances due to loudspeaker location within the cavity behind the nose panel. Responsive AVAS EVs were shown to have a sound level range of between 6 dB L Aeq,1hr over the vehicles measured, whereas the ICE vehicle had a range of between 0.2-0.7 dB L Aeq,1hr . This demonstrates the effectiveness of the Responsive system at reducing sound levels in quieter environments, providing respite from noise during driver journeys. Driver cabin sound levels were measured in ADL, Optare, Wrightbus and Caetano vehicles, with overall background sound levels ranging 56-66 dB L A90 and speech interference levels (derived from the background sound level) ranging 53-56 dB L SIL . It should be noted, that these levels are including environmental sound and not the specific AVAS sound ingress only. The Go-Ahead/BYD vehicles were shown to have on average the lowest L A90 levels of the fleets tested, yet the Abellio/Caetano vehicle was shown to have the lowest L SIL level, with similar levels achieved by the Tower Transit/Optare vehicle following the AVAS hardware reconfiguration conducted in early September 2021. The Tower Transit/Optare vehicle was operating on a busier route, which may suggest that considering increased environmental noise, the vehicle may have the lowest AVAS sound ingress level across the vehicles tested. The differences between the vehicles are expected to be due to vehicle structure and design as well as AVAS system positioning that have been seen to cause resonances in some configurations. 7. WORK PACKAGE 4: PUBLIC & DRIVER ENGAGEMENT

To understand the impact of the Urban Bus Sound on stakeholders including bus users, pedestrians and those with a visual impairment, research has taken place including:

● On-street survey with pedestrians and bus users; ● Accompanied journeys with organisations representing visually impaired people and

residents’ groups; ● In-depth interviews with bus drivers. Ease of hearing the sound Respondents were more likely to hear the Urban Bus Sound in quieter areas where the AVAS was at a low setting with around half of respondents to the on-street survey saying they could hear the sound when the bus set off (52%) and when the bus was travelling (44%) compared to than those in noisier areas (31% and 34% respectively). In both these situations the AVAS is playing both parts of the Urban Bus Sound. Feedback from the accompanied journeys suggested that the ‘ping’ was the most distinctive element of the Urban Bus Sound and can be heard most above other traffic sounds. Drivers also felt the ‘ping’ was the most distinctive element with many not being aware that the base sound; some thought the hum was the sound of the “engine”. Most drivers did not think pedestrians could hear the sound enough. When the bus is stationary it only plays the core sound and slightly more respondents stated they could not hear the Urban Bus Sound when the bus was stationary (45%) than when it was moving

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

(overall 36% of respondents were not aware of the Urban Bus Sound when the bus set off and 40% when it was traveling along the road). Hearing the Urban Bus Sound on the bus Overall, 40% of bus users felt they could hear the Urban Bus Sound whilst traveling on the bus. During the accompanied journeys it was not felt that the sound was intrusive whilst making a journey. The driver’s interviewed had not received any complaints about the sound and did not feel passengers could hear it and if they could they were not disturbed by the sound. Drivers, however, could hear the sound in the cab. Some drivers reported being able to ‘zone out’ the sound and likened it to white noise and therefore were not affected. However, for some drivers the sound was intrusive and this was owning to several factors:

● Model of the bus: some models of bus appear to be louder than others but even within the

same range some buses also seem to be to louder than others. ● Other noises in the cab: some models of bus had additional features such as the mobile

eye, which drivers reported beeps a lot. This combined with the AVAS sound caused irritation and a heightened awareness of the sound. ● Personal preference: some drivers appear to be more sensitive to sound and/or other

circumstances cause them to be more aware of the sound and once conscious they are unable to zone out the sound. ● Awareness of AVAS and its function: a lack of awareness of the Urban bus sound and in

particular the elements of the sound i.e. The base hum and the ‘ping’ added to frustrations with some thinking the hum meant there was something wrong with their bus. Awareness of the Urban Bus Sound Those who were already aware of the Urban Bus Sound were significantly more likely to hear the sound than those that had not heard the Urban Bus Sound before taking part in the survey; when the bus was stationary (58% compared to 21%); when the bus was setting off (74% compared to 25%) and when the bus was traveling (67% compared to 22%). The accompanied journeys also highlighted the significance of familiarity with the sound as the visually impaired people interviewed all felt they could identify an AVAS bus more readily once they had heard the Urban Bus Sound a few times. Therefore, the findings would suggest as awareness of the Urban Bus Sound grows its ability to meet its safety function will increase. The drivers interviewed also suggested publicity would help AVAS become more effective. Volume of the Urban Bus Sound Half (50%) of respondents to the on-street survey felt the volume of the AVAS was about right but 45% felt it was too quiet. This was borne out in the accompanied journeys, where all respondents felt the sound would be more effective if it were louder. The sound was felt to be unobtrusive and therefore respondents felt the volume could be turned up without becoming a noise nuisance. Over two-thirds of the drivers interviewed did not feel the volume was loud enough to alert other road users to the presence of the bus.

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

Does the Urban Bus Sound meet its Safety Function? Almost half of respondents to the on-street survey felt the Urban Bus sound would help prevent accidents (44%) and enabled them to know there was a bus close by (47%). Respondents in the accompanied journeys also felt the Urban Bus Sound served its safety function. However, the Urban Bus Sound was thought to be difficult to hear in busy traffic but in these situations Vulnerable Road Users are alert to the presence of traffic and the dangers that presents. The drivers interviewed were all keen to improve safety, particularly pedestrian safety but most did not feel the AVAS sound was loud enough, and half felt it should be played for speeds above 12 mph. 8. WORK PACKAGE 5: URBAN BUS SOUND REVISIONS & PRE-ENTERING SERVICE DOCUMENTATION

The final work package was to implement the recommended improvements to the Urban Bus Sound AVAS and to produce suitably robust and practicable Responsive AVAS specification and pre- entering service guideline document for AVAS solution manufacturers, bus operators and others. The documents are to ensure that the Responsive AVAS is consistently implemented across the emerging quiet-running surface transport fleet. Urban Bus Sound & Responsive AVAS Revisions Based on the findings discussed in Work Packages above, the following improvement measures were implemented to optimise the system performance for both acoustic conspicuity improvement and soundscape impact mitigation.

1. Increase in ‘Core’ sound output level in relation to the ‘Beacon’ for both idle and moving

operations. This measure is intended to increase the acoustic conspicuity of the ‘Core’ element that provides a continuous sound during vehicle operation. It is particularly important to increase the audibility of the ‘Core’ during idle operation for road users in proximity to the vehicle as it may depart.

2. Increase sound energy at frequencies above 2000 Hz to further support localisation. This measure is expected to increase the ability for road users, especially VRUs, to locate the vehicle in the surrounding sound environment. This is not expected to negatively affect the urban soundscape or acceptability considerations outlined in this report, as frequencies above 2000 Hz are subject to higher propagation losses and building facades are generally more efficient at attenuating higher frequencies.

3. Reduction of Step Level 4 sound level or removal of Step Level 4 from operation. This proposal was made based on impact reduction of driver conditions and the wider urban soundscape, but also partly justified by the acoustic conspicuity results for busy environments. The maximum sound levels measured during Step Level 4 operation have been shown to be like that of an ICE vehicle passby and therefore is a risk to acceptability compared to the current vehicle fleet. When considering frequency content, the Responsive AVAS is shown to significantly reduce sound energy across the frequency spectrum compared to ICE vehicles, but tonalities are present that increase the risk of disturbance in the driver’s cabin or in nearby residential properties. It is expected that with the increase in Core sound output in relation to the Beacon, as well as the

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

increase in higher frequency content, the reduction in the overall range of Responsive AVAS sound output levels from 0-4 to 0-3, will be of little detriment in acoustic conspicuity in busy environments. Three revised sound mixes, Option A to C, were produced by Zelig Sound based on the recommendations and were implemented on a test ADL vehicle at the Go Ahead London garage at Northumberland Park on 6th December 2021. Option C became the preferred option, as it produced the most even frequency distribution with fewer tonalities and resonances when in-situ. The results demonstrate the increase in higher frequency content from 1600 Hz up to 6300 Hz. Operating at Step 3 (new maximum operating level), sound levels increased between 7-20 dB compared to the original mix on Step 4 across this frequency range. Figures 2 to 5 show the revised Urban Bus Sound Option C frequency profiles at Step 3 and 0 for 10 km/h and 20 km/h compared to the original mix.

Remix Option C at 10 km/h - Step 3 Highest Level

Max Event L max per 3rd Octave Band

90

80

70

60

Sound Pressure Level, dB

50

40

30

20

10

0

160 200 250 315 400 500 630 800 1k 1.25k 1.6k 2k 2.5k 3.15k 4k 5k 6.3k 8k 10k 12.5k 16k 20k

3rd Octave Band Frequency, Hz

Figure 2: Sound output level for Option C remix at Step 3 compared against Regulation 138

UN ECE Reg 138 Minimum - 10 km/h Option C - 10 kph Original - 10 kph (Step 4)

requirements at 10 km/h and the original mix at the highest step setting (Step 4).

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

Remix Option C at 20 km/h - Step 3 Highest Level

Max Event L max per 3rd Octave Band

90

80

70

60

Sound Pressure Level, dB

50

40

30

20

10

0

160 200 250 315 400 500 630 800 1k 1.25k 1.6k 2k 2.5k 3.15k 4k 5k 6.3k 8k 10k 12.5k 16k 20k

3rd Octave Band Frequency, Hz

Figure 3: Sound output level for Option C remix at Step 3 compared against Regulation 138

UN ECE Reg 138 Minimum - 20 km/h Option C - 20 kph Original - 20 kph (Step 4)

requirements at 20 km/h and the original mix at the highest step setting (Step 4).

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

Figure 4: Sound output level for Option C remix at Step 0 compared against Regulation 138

requirements at 10 km/h.

Remix Option C at 10 kmh - Step 0 Lowest Level Max Event Lng, Per Sid Octave Band

Figure 5: Sound output level for Option C remix at Step 0 compared against Regulation 138

requirements at 20 km/h. At Step 0, the lowest operational level, the Urban Bus Sound Option C achieves the requirements of Regulation 138 comfortably at both 10 and 20 km/h operations between 500 Hz to 5000 Hz. The results provide a clear indication of the increase in potential acoustic conspicuity of the Option C remix, in balance with the marginal reduction in overall sound output levels to reduce to Step 3 (69 dB). Regarding driver cabin sound levels, the preferred Option C mix at Step 3 provided sound conditions generally in line with the performance targets, achieving ‘Good’ and ‘Fair’ Speech Interference Level (SIL) performances for raised speech. This was most notably due to the reduction in the dominance of the Beacon and the overall reduction in sound level by limiting the maximum sound output operation to Step 3. Loudspeaker position reconfigurations are expected to further reduce sound ingress levels in the driver’s cabin. Responsive AVAS Optimisation Workshops with Manufacturers & Operators Detailed workshops were conducted with operators Go Ahead London and Tower Transit, vehicle manufacturers ADL and Optare and service providers Forman Vehicle Services and Actia to understand optimisation options to improve the Responsive AVAS performance both externally around the vehicle and internally in the driver’s cabin. AVAS hardware configurations were explored to reduce sound levels in the driver’s cabin and clear improvements were found. Following conclusions from the field trials and engagement surveys, adjustments were made to the AVAS sound and Responsive configuration to improve performance regarding acoustic conspicuity external to the vehicle, driver acoustic comfort as well as on soundscape considerations and acceptability. The revised Option C mix and Responsive AVAS configuration was implemented on an Optare vehicle on 20th December 2021 at the Tower Transit Westbourne Park garage. The two

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

Remix Option C at 20 knv/h - Step 0 Lowest Level Max Event Lng Per Srd Octave Band

loudspeakers were positioned on the front lower frame, facing towards the ground to avoid the reflection of sound into the driver’s cabin. Forward sound projection was deemed sufficient during idle operation at Step 3 and observations during on-street arrivals demonstrated that the Responsive system was working correctly. It was observed that the Core sound operation was significantly more present on arrival of the vehicle and the balance between the Core and the Beacon elements was favourable. The sound levels in the driver’s cabin under Step 3 ‘idle’ operation demonstrated that despite the Option C sound mix revision including increasing the sound level in some frequency regions, this did not result in an uplift in sound level ingress. The result suggests that sound levels in the driver’s cabin provide the minimum targeted Speech Interference Levels ratings. Internal measurements of revised AVAS performance with the optimised system configuration were conducted for inclusion in the performance specifications to ensure consistency across the fleet. During the trials, a drive around was organised with the driver’s Union Representative to observe and comment the sound revisions and configuration updates following the initial installation and trial feedback. The comments from the Representative were positive and it was noted that the sound level was notably reduced during driving operation, with the Beacon “ping” sound having a less significant presence. The revisions were considered acceptable to the Representative and the engagement process was also well received as the Union’s concerns were listened to and addressed. Urban Bus Sound Specification & Pre-Service Commissioning Documentation Draft revised specification and pre-entering service check documents were produced following the findings of the Phase 2 works and presented in a page-turner workshop with manufacturers, operators and AVAS solution providers for their comment. The drafts were intended to allow a greater degree of flexibility for manufacturers to tailor their AVAS solutions according to their vehicle design constraints, whilst still maintaining a high-performance standard. Final versions of the Specification and Pre-Service Commissioning Documentation have been produced incorporating comments and suggestions along with revised Urban Bus Sound mixes and measurement data that demonstrate the minimum performance standards of the final AVAS in-situ. The documents are the outcome of the Phase 2 project and are intended to be used as a basis for manufacturers, suppliers, operators and TfL to successfully implement the Responsive AVAS and maintain consistency of its performance across TfL’s expanding quiet-running vehicle fleet. 9. CONCLUSIONS

Phase 2 of the Urban Bus Sound implementation was completed through collaboration between Transport for London (TfL), Anderson Acoustics, AECOM and Zelig Sound. Its main focus was achieving performance consistency across the bus fleet; improving the acoustic conspicuity of the AVAS by integrating geospatial datasets into the system, thus turning the sound into an ever evolving element dependent on the various urban soundscapes; assessing the acceptability of the AVAS within the driver’s cabin by studying and controlling, where appropriate, its ingress; and supporting the different bus manufacturers and operators across the city through the implementation

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

process and specific obstacles that they may face to achieve the specification of performance required.

The recommendations have been implemented and tested in-situ and confirmed through stakeholder engagement and manufacturer confirmation. The revisions have been demonstrated to deliver improved system performance and balance optimised acoustic conspicuity with acceptability for a range of scenarios. The specification has been designed to provide wide range support for various manufacturers to supply the AVAS hardware/software and implement it to achieve the minimum suitable performance requirements. The specification encourages enhanced sound fidelity as technology improves and on-going consideration for how the Urban Bus Sound may affect the ever- changing urban soundscape.

11. ACKNOWLEDGEMENTS

We gratefully acknowledge support from Forman Services Ltd, Alexander Dennis Ltd, Go Ahead London, Optare, Wrightbus and Caetano. 12. REFERENCES

1. Greater London Authority, Mayor’s Transport Strategy, 2018. 2. Transport for London, Bus Safety Standard: Executive Summary, 2018. 3. TRL Limited, The Transport for London (TfL) Bus Safety Standard: Acoustic Conspicuity.

Evaluation of Safety Measure (2018) 4. UN ECE Regulation No. 138, Uniform provisions concerning the approval of Quiet Road

Transport Vehicles with regard to their reduced audibility (2017). 5. Department for Environment, Food and Rural Affairs (Defra), Strategic Noise Mapping (2017) 6. UK Government, DXF noise exposure contours for Heathrow, Gatwick and Stansted airports 7. Fleischer, H., Blauert, J. ‘Audibility of Some Specific Public-address Warning Signals in Typical

Environmental Noise Situations’. Applied Acoustics 27, 305-319 (1989). 8. BS 5839-8:2013 Fire detection and fire alarm systems for buildings – Part 8: Code of practice for

the design, installation, commissioning and maintenance of voice alarm systems. 9. European Environment Agency, ‘Environmental noise in Europe — 2020’. EEA Report No

22/2019 (2019). 10. BS EN ISO 9921:2003 ‘Ergonomics — Assessment of speech communication’.

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