Exploring the road safety impacts of Platform and Easy Access Stops in Melbourne Final Report

Transcription

1 Exploring the road safety impacts of Platform and Easy Access Stops in Melbourne Final Report Julian Reynolds CIV5308: Case Studies in Transport Engineering Institute of Transport Studies Department of Civil Engineering, Monash University June 5, 2009

2 Executive Summary Since 1999, two new types of tram stops have been developed and installed in 288 locations throughout Melbourne. Platform stops have replaced safety zone stops, providing station platform style facilities in the centre of the road. Easy access stops have replaced kerb side stops, providing level access for passengers stepping from the traffic lane to the tram floor. While improvements in accessibility, reliability and journey times are important outcomes of the recent upgrades to tram stops, the road safety implications of these changes have not yet been fully investigated. This project investigates the road safety impacts of the two new types of tram stops by examining crash statistics at all new stops. Unfortunately, there is insufficient crash data available to state with any certainty whether crash rates have increased or decreased. A discontinuity in the VicRoads CrashStats data set has severely hampered crash data analysis based on police reports of casualty crashes, resulting in insufficient data to draw any conclusions. Revisiting this data set in about two years when more data is available may be appropriate. Internal Yarra Trams data has indicated that incident rates involving pedestrians have increased significantly following the installation of platform stops. However, the reliability of this data is not sufficient to be confident in this result. This data is collected for operational logging purposes rather than crash reporting. Yarra Trams may wish to consider implementing a formal crash reporting and recording program similar to and interlinked with the VicRoads Crash Stats program to improve the opportunities for crash data analysis across the tram network. To make up for the lack of data a road safety audit approach has been used in this project to assess safety at typical safety zone, platform, kerb side and easy access stops. Perceived risks were rated on the likely frequency of crashes and the resulting crash severity. This approach showed that platform stops reduce the risk to passengers being struck by a tram due to the wider waiting areas and clear separation provided by the platform edge. However, the platform stops have not improved safety for passengers moving between the stop and footpath, because not all crossing points are controlled by zebra crossings or signals. The location of the tram stops on the downstream side of signalised intersections also exposes passengers waiting to cross at pedestrian signals to traffic leaving the intersection. Pedestrians crossing at the other end of the platform may also be at risk of being struck by trams departing the stops. There is an improvement in safety for passengers at the new easy access stops. The traffic calming effects of the ramp up to the raised platform reduces the speeds of approaching vehicles and makes the stop more visually obvious to drivers. While there is an increased risk for motorists due to the 300mm drop off between the platform and the tram tracks, this is small risk compared to the hazards facing passengers at untreated kerb side stops. A range of changes to the standard design have been recommended to improve the safety of passengers and other road users at platform and easy access stops. These include simple aspects such as pedestrian fencing and zebra crossings as well as more substantial items such as refuge islands, crash barriers and offset platforms. Some or all of these concepts could be considered for inclusion at future tram stop upgrades, within the constraints of the conditions at each individual site. As more platform and easy access stops are constructed and a larger volume of crash data becomes available, it will be possible to quantitatively assess the safety impacts of these new stop types. While stop improvement programs are primarily aimed at improving access, there remains significant opportunity to research and improve the safety of passengers and other road users in the vicinity of tram stops. ii

3 Contents 1 Introduction 1 2 Tram Operating Environments Tram stops in Melbourne Older style stops New stops Literature Review Stops in exclusive and semi-exclusive environments Stops in mixed traffic Safety for passing public vehicles Summary of key findings from the literature review Crash Data Analysis VicRoads Crashstats data Yarra trams data Road safety audits Road safety audit findings Risk assessment Proposed changes to stop design Platform stops Easy access stops Conclusions 58 8 References 60 9 Acknowledgments 61 A Concept designs for platform and easy access stops 62 B Site photographs 65 iii

4 1 Introduction Melbourne has one of the largest tram networks in the world, with 250 kilometres of tram tracks, 1770 tram stops and 500 trams operating across 25 different routes (Purdy 2007). Seventy five percent of the tram network is in mixed traffic operations, with trams sharing space on the road with cars, trucks and other public vehicles. With this sharing of space comes the possibility of conflict between trams and vehicles, or between vehicles and passengers who are boarding, alighting or waiting for a tram. Tram tracks in Melbourne predominately run down the centre of the road in a median reservation, an exclusive tram lane or a lane shared by trams and other traffic. Many tram stops in Melbourne have a passenger waiting area in the centre of the road, between the traffic lanes and the tram tracks. The stops pose risks for passing motorists running into the fencing and other infrastructure at the stop, for passengers waiting in the stop being struck by a passing vehicle or tram, or for passengers crossing the traffic lanes to get to and from the stops. At kerb side stops, where passengers wait on the footpath and cross the traffic lanes to board the tram directly when it arrives, there are risks of passengers being struck by a passing vehicle. In recent years, there have been many innovative upgrades to tram services and stops in Melbourne to improve reliability and speed of services and to accessibility to public transport for disabled and mobility impaired passengers. Since 1999, two new types of tram stops have been developed and installed in 288 locations throughout Melbourne (Yarra Trams 2009a). Most of these new stops are platform stops, which provide station platform style facilities in the centre of the road. A small number of new stops are easy access stops, which retrofit existing kerb side stops to provide level access from the traffic lane to the tram floor. Kerb access stops have also been trialed along some tram routes, where the traffic carriageway is narrowed to provide direct access from the kerb to the tram for passengers. While improvements in accessibility, reliability and journey times are important outcomes of the recent upgrades to tram stops, the road safety implications of these changes have not yet been fully investigated. This project investigates the road safety impacts of the two new types of tram stops by examining crash statistics at all new stops and assessing the risk and exposure of road users to hazards at a number of representative stops. Recommendations for changes to the existing stops and future stop design will be a key outcome of this proposed project. In the following section the different types of tram operating environments are identified and discussed. Section 3 examines developments in tram, train and bus stop design that could be adapted to improve safety at the new types of stops in Melbourne. Sections 4 and 5 directly compare the safety of the new Melbourne stops with the previous safety zone and kerb side stops. Finally, in Section 6 recommendations are made on possible changes to the design of platform and easy access stops to improve the safety of passengers, motorists and other road users. 1

5 2 Tram Operating Environments There has been a long history of tram operations in Australia and around the world, but many tram systems were shutdown in the middle of the 20th century to make way for cars and buses on city streets. It is only in recent years that trams are returning to cities as a flexible alternative to bus and train public transport systems. Trams can provide higher capacity, better ride comfort and a stronger image than buses, at a lower cost and with greater flexibility than urban train systems. Topp (1998) describes five German cities where trams are being returned to city streets and old tramways are being upgraded. These five case studies how trams can operate in a range of different environments and roles such as in Hannover, where trams are run through a metro style underground in the inner city and along streets in outer areas, and in Karlsruhe, where railroad tracks are shared between regional trains and trams. Trains can operate only in a separate right of way, with either controlled or grade separated crossings. Buses are typically used in mixed traffic environments, sharing the streets with other vehicles although modern Bus Rapid Transport systems and busways show how buses can be adapted to different types of operations. In contrast, trams are more flexible and are regularly operated in a wide range of environments and alignments. Table 1 shows the nine categories of tram operating environment reported by Korve, Farran & Mansel (1996). These categories range from full grade separated alignment (type a) where there is no interaction between trams and other vehicles or pedestrians, to mixed traffic operations (type c.1) where the tram uses the same traffic lane as other vehicles on the road. Table 1: Light Rail Transit Alignment Classification - reproduced from Korve et al (1996) Class Category Description of Access Control Exclusive Type a Fully grade-separated Type b.1 Separate right of way Type b.2 Shared right-of-way protected by 6-inch high curbs and fences Semi-Exclusive Type b.3 Shared right-of-way-protected by 6-inch high curbs Type b.4 Shared right-of-way protected by mountable curbs, striping, and/or lane designation Type b.5 LRT/Pedestrian mall adjacent to a parallel roadway Type c.1 Mixed traffic operation Non-Exclusive Type c.2 Transit mall Type c.3 LRT/Pedestrian mall The entire range of alignment types may be encountered in a single system, or even along a single route. An example is the 96 tram route to St Kilda Beach in Melbourne, which runs through almost the complete range of alignment conditions from type a to type c.3 (Metlink Melbourne 2009), as shown in Table 2 and Figure 1. Along route 96 there is a large range of different stop types to match the different operating environments. Platform stops are found in the CBD and along the exclusive and semi-exclusive alignment from Port Junction to St Kilda station. In the mixed traffic operations close to the St Kilda terminus there are centre of the road safety zones and kerb side stops, where passengers need to step up from the roadway to the tram floor. 2

6 Figure 1: Route 96 map, from Metlink Melbourne (2009) 3

7 Table 2: Melbourne Route 96 CBD to St Kilda - Alignment and Stop Types Section Type Description Stops 1 Parliament to b.4 tram tracks separated from centre of the road platform Swanston Street traffic lane by lane designation stops 2 Swanston Street to c.3 tram route through Bourke platform stops Elizabeth Street 3 Elizabeth Street to Port Junction 4 Port Junction to Wright Street 5 Wright Street to St Kilda Station 6 St Kilda Station to Luna Park 7 Luna Park to St Kilda Beach Street Pedestrian Mall b.4 tram tracks separated from traffic lane by lane designation centre of the road platform stops a fully grade separated platform stops b.1 separate right of way, but with some at-grade road and pedestrian crossings b.4 tram tracks separated from traffic lane by lane designation c.1 Mixed traffic operation, tram lane shared with traffic lane platform stops centre of the road safety zones and kerb side stops kerb side stops In comparison to trams, other modes have few different types of stops. Trains usually stop only at stations with platforms that are well away from passing traffic. In Melbourne, many train stations have pedestrian overpasses or tunnels, or automatic gate controlled at-grade crossings that allow passengers to safely cross the train tracks. Buses stop mostly at kerb side stops in mixed traffic operations and passengers make use of normal pedestrian facilities to access the bus stop. Bus interchange stations and busways are relatively rare examples where a higher standard of stop is provided and passengers are partially or fully separated from vehicles. When considering safety at tram stops, can we directly transfer the experiences of bus and train operators? Huston, Cardimen & Halperin (1995) seek to do just this in their paper on a proposed tram system operating in a type b.4 semi-exclusive alignment. The paper suggests that there would be a significant reduction in fatalities on a major road if a tram line was installed in place of a traffic lane Huston et al. (1995) compare the average fatality rate of intra-urban rail systems across the USA to the fatality rate on the existing road to determine the reduction in fatalities if travellers transfered from cars onto a tram. The analysis shows a significant reduction in fatalities and injuries and that the cost of the new tram system would be repaid in 5 to 20 years if the economic benefit of crash reduction were included in the cost assessment. However, Huston et al. s (1995) analysis may be over simplifying the safety of tram systems. The national intra-urban fatality rate use in the analysis includes systems operating in all types of alignments, from exclusive (type a) to non-exclusive (type c). Unfortunately, semi-exclusive and non-exclusive alignment types account for the majority of total accidents on tram systems, even while these alignment types make up only a small percentage of the total track miles. A survey of 10 USA light rail systems found that between 70 and 100 percent of total crashes occurred in alignment types b.2 to b.5 and c.1 to c.3, even when these alignment types averaged only 30 percent of the total track miles on each system (Korve et al. 1996, page 3). It is also 4

8 important to note that these crash rates do not include collisions between motor vehicles or between motor vehicles and pedestrians that were caused by the nearby location of a tram track or stop (Korve et al. 1996, page 2). These unreported accidents are only present in semiexclusive and non-exclusive alignment types, and are a further example of how tram operations in these alignments differ from metro style train services. Clearly, the semi-exclusive and nonexclusive alignment types have a lower level of safety than a system operating in an exclusive right of way. By adopting the national rate for all alignment types Huston et al. (1995) are dramatically overstating the safety benefits of trams in a semi-exclusive operating environment. Korve et al. s (1996) findings meet the logical expectation that transit services with a higher degree of separation from other traffic and pedestrians would have a higher level of safety. Increasing the separation between pedestrians, vehicles and transit services reduces the level of conflict and should therefore reduce the rate of crashes. When considering tram stop safety across the entire range of alignment types, there is little that can be learnt from train operations in type a alignments alone. Rather, when seeking to transfer experiences of bus and train operators into the design and operation of tram services, it is important to consider the type of operating environment in which each mode operates. The design of train platforms and stations clearly has relevance for tram systems in type a and type b.1 alignments. Experiences from busways and bus rapid transit systems may be transferrrable to trams in semi-exclusive operating environments, while the approaches used to operate regular street-running buses may be adaptable to trams in non-exclusive mixed traffic alignments. 2.1 Tram stops in Melbourne Currie & Smith (2006, p.43) identify five approaches to providing stops in centre of the road tram operations in mixed traffic environments in Melbourne. These are: Safety zones, Platform stops (previously known as super stops), Kerb side stops, Kerb access stops, and Easy access stops In addition, Melbourne s tram network also has some major tram interchanges with station style facilities with multiple platforms. Other sections of the tram network are located in the central median of wide roads or in a separate right-of-way where stops consist of a raised platform or ground level boarding area Older style stops Safety zone stops are common in the central areas of Melbourne and along arterial roads without a medium. The stops consist of a distinctive yellow safety zone traffic island as shown in Figure 2. Passengers wait in a narrow area next to the tram tracks, with a fence separating them from passing road vehicles. Tram passengers need to step up from the road surface level to the tram floor to board a tram. At signalised intersections safety zones are placed on the approach side to the intersection. Outside central areas of Melbourne, where there are lower passenger volumes and tram frequencies, many tram stops are located on the kerb rather than in the centre of the road, as shown in Figure 3. At these kerb side stops passengers wait behind the kerb or on the footpath for the tram to arrive. To board the tram passengers cross the traffic lane and step up from 5

9 Figure 2: Safety Zone tram stop at Elizabeth and Bourke Street, Melbourne CBD the road surface level to the tram floor. There is no physical separation between boarding passengers and passing motorists. It is illegal for a driver to pass a stopped tram at a tram stop and kerb side stops rely on drivers obeying this rule to maintain safety for passengers. Older style major interchanges exist in a few locations on the Melbourne tram network. Notable examples are the Elizabeth Street Terminus in the CBD and the Domain interchange on St Kilda Road. These interchanges have many of the same characteristics as safety zones, although generally with greater separation from adjacent traffic lanes New stops Platform stops are the most common and visually striking type of new tram stop design in Melbourne. These stops have replaced safety zones in many locations, particularly in the central business district. As shown in Figure 4, a platform stop occupies the same position in the middle of the carriageway as a safety zone, with raised platforms providing level access to the tram floor level. To create space for the wide platforms one of the adjacent traffic lanes is removed, with traffic restricted to only one lane. Platform stops have been constructed with stops for each direction on only one side of the intersection, unlike safety zone stops which had stops for each direction on the approach to the intersection only. Some new stops have been constructed at mid block locations instead of at intersections. Similar platform stops have been installed at stops where trams operate in a separate right of way. Figure 14 shows the new raised platform at the Fraser Street stop on route 96, where the tram operates in a separate right of way. While these stops improve accessibility, there is little effect on safety other than to improve separation between the passenger waiting area and the tram dynamic envelope. A number of major interchanges have been upgraded or created using similar platform designs. The new interchange at St Vincent s Plaza has three different platforms, while the 6

10 Figure 3: Schematic of kerb side stop, from Currie & Smith (2006) interchange at Melbourne University has a single island platform that serves trams traveling in both directions. Kerb access stops were trialed in Clarendon Street in South Melbourne as part of the Think Tram Program (Cliche 2006, p.455). These stops consisted of a kerb extension on the downstream side of a signalised traffic intersection and allowed passengers to step directly from the kerb into the tram. Only a single shared traffic lane was provided at the stop so other vehicles had to wait behind the tram until it departed the stop. The kerbside stops were removed following the trial, primarily due to local trader s concerns about the loss of kerb side parking (Cliche 2006, p.456). Currie & Smith (2006, p.43) describe the easy access stop, a new and innovative modification to kerb side stops to provide level access to the tram floor. On the approach to the tram stop the level of the traffic lane is raised above the level of the tram tracks, allowing passengers to step from the traffic lane directly across to the tram floor at the same level, as shown in Figure 5. With 288 new tram stops completed to date (Yarra Trams 2009a), Melbourne is continuing to replace the older style stops with platform and easy access stops. In the following section, developments in tram, train and bus stop design to improve safety will be compared with the new Melbourne stops. Sections 4 and 5 will directly compare the safety of the new Melbourne stops with the previous safety zone and kerb side stops. Finally, in Section 6 recommendations will be made on possible changes to the design of platform and easy access stops to improve the safety of passengers, motorists and other road users. 7

11 Figure 4: Platform Stop at Bourke Street and Queen Street, Melbourne CBD Figure 5: Schematic of easy access stop, from Currie & Smith (2006) 8

12 3 Literature Review This section details the current developments in overseas tram, bus and train stops that could be applied to facilities in Melbourne to improve safety. Tram stops in mixed traffic environments are of particular interest in this study, as the majority of new Melbourne stops are located in this environment. Section 3.2 describes the available research into safety at tram and bus stops in mixed traffic environment. Overseas developments in this area should be directly transferrable to the Melbourne tram network. However, the review begins with an examination of research into tram stop safety in exclusive and semi exclusive environments. While operations in these environments are quite different to those in mixed traffic, some of the developments in stop design may be transferrable to Melbourne s mixed traffic tram stops. 3.1 Stops in exclusive and semi-exclusive environments The vast majority of research into tram safety considers tram operations in semi-exclusive environments. American research, in particular, is mostly focused on operations in alignments running along separate rights of way (class b.1) and in medians or adjacent to highways (classes b.2 and b.3). While much of this research may be more applicable to Melbourne s train network than the trams, approximately 25% of the Melbourne tram network operates in these types of exclusive or semi-exclusive environments (Purdy 2007). Developments in stop design from class b.1, b.2 and b.3 alignments may also be transferrable to the design of platform stops in Melbourne s mixed traffic, class b.4 or b.5 operating environments. Korve, Farran & Mansel (1995, pp ) recommend a range of pedestrian crossing treatments for trams operating in semi-exclusive alignments, including along the medians of arterial roads. Many of these treatments are in use across the USA and include: Passive warning signs, Active warning signs, including signs that warn of a second tram and which direction pedestrians should look, Pedestrian bedstead barriers, as shown in Figure 6, Z-crossing channelization, as shown in Figure 7, Manually operated self closing swing gates which alert pedestrians to the LRT tracks and forces them to pause before crossing them as shown in Figure 8, and automatic pedestrian gates. The key aim of these treatments is to prevent pedestrians running out onto the tracks without first looking for approaching trams. Similar types of treatments can be seen across the Melbourne urban rail network, particularly in the vicinity of at-grade rail-road crossings and close to stations. The type b.1 sections of the number 96 tram route also have pedestrian barriers that are similar to the Z-crossing channelization and bedstead barriers. It may be possible to adapt these types of barriers to Melbourne s centre of the road platform stops, in order to improve pedestrian safety when crossing tram tracks and traffic lanes. Siques (2001) provides a similar list of devices to Korve et al. (1995), but importantly identifies the following four key approaches to promote pedestrian safety: improving pedestrian awareness of the crossing, controlling the pedestrian path across the crossing, 9

13 Figure 6: Pedestrian bedstead barricade and crossing, from Korve et al (1995) Figure 7: Pedestrian z-crossing, from Korve et al (1995) 10

14 Figure 8: Pedestrian swing gates, from Korve et al (1995) 11

15 improving awareness and sight distance to approaching trams, even if the crossing is actively controlled with automatic gates or pedestrian lanterns or active signs, and using public outreach and education programs to improve knowledge of the hazards at crossings. Siques (2001) also highlights the use of tactile pavement markers, train-coming icons, audio warnings and pedestrian refuge areas to improve safety at tram crossings in a semi-exclusive environment. Examples of train-coming icons and pedestrian refuges are shown in Figures 9 and 10. Pedestrian refuges and tactile pavement markers are already used in and around platform tram stops in Melbourne to improve pedestrian safety, whereas audio warnings and train-coming icons would seem more suited to isolated pedestrian crossings in high speed environments. Indeed, within Siques s (2001) paper, reference is made to the automatic swing gates which are used to protect at-grade train crossings in Melbourne. Many of the treatments outlined in these two papers are more relevant to the higher alignment classes such as b.1 and b.2. The restricted space and lack of separation in the lower alignment types such as b.4, b.5 and mixed traffic may limit the practicality of these types of installations. Figure 9: Train-coming icon, San Francisco LRT system, from Siques (2001) A range of pavement marking and delineation devices used in Barcelona are described by Farran (2006). Unlike treatments such as z-crossings and swing gates, these pavement markings can be used in areas with limited space to separate pedestrians and trams, and may even be appropriate for stops in mixed traffic environments. Barcelona has sections of two-way track and uses a tram directional striping to warn pedestrians from which way trams will approach, as shown in Figure 11. This type of treatment may be unnecessary in Melbourne as all sections of track are doubled and trams always travel on the left, so pedestrians familiar with the environment are less likely to look the wrong way for approaching trams. Figure 12 shows pavement markings of the dynamic envelope of trams, to encourage Barcelona pedestrians to wait well clear of the tracks when crossing at-grade. Similar line markings are used throughout the Melbourne tram network at centre of the road safety zone stops, although without the same width, symbols or level of delineation. 12

16 Figure 10: Pedestrian refuge area, from Siques (2001) Figure 11: Tram direction striping, from Farran (2006) 13

17 Figure 12: Dynamic envelope delineation with pavement texture, paint and tram symbol from Farran(2006) Separating pedestrians and transit vehicles was given particular importance in the design of at-grade crossings at stops along the New Britain to Hartford Busway (Watkins, Sorenson & Garrick 2002). A range of different platform layouts were evaluated through internal agency review and public consultation, with the goal of providing a safe busway station within the constraints of a restricted right of way and the need to allow express buses to skip the station. The adopted platform design is shown in Figure 13, and allows pedestrians to cross between platforms behind the bus stop. Watkins et al. (2002) identify that this modified opposite platform design prevents pedestrians crossing the bus way at any location, as is the case when platforms are directly opposite. By offsetting the platforms, pedestrians are encouraged to walk to the end of the platform before crossing. Figure 13: Modified opposite busway platform design, from Watkins et al. (2002) Encouraging passengers to cross behind the bus is inconsistent with the recommendations of Unger, Eder, Mayr & Wernig (2002), who identify that pedestrians crossing behind a bus is highly dangerous. However, Unger et al. s (2002) study was predominately of passenger injuries in mixed traffic conditions and does not appear to be relevant to the busway s separate right of way. At the busway, pedestrians only cross one lane behind the bus. There is no need to prevent passengers being struck by vehicles from the opposite direction, as is the case in mixed traffic. A similar approach can be found at the Fraser Street stop on the number 96 tram route in Melbourne, as shown in Figure 14. At this stop the platforms are not directly opposite 14

18 each other. Passengers are therefore encouraged to walk to the pedestrian crossing connecting the end of both platforms rather than directly across the tram tracks. This contrasts with the nearby Middle Park tram stop which has directly opposite platforms, as shown in Figure 15. At this stop, fencing has been installed to physically prevent passengers crossing the tram tracks other than at the pedestrian crossing. Figure 14: Modified opposite tram stop platform, Fraser Street stop on route 96, Melbourne The major difference between the offset platforms described by Watkins et al. (2002) and the Fraser Street stop in Melbourne is the direction of the offset. The busway pedestrian crossing is located at the rear of the stop, so that the bus crosses the pedestrian crossing prior to disembarking passengers. With Melbourne trams traveling on the left, at the Fraser Street stop the tram stops prior to the crossing. Unlike the busway crossing, at the Fraser Street stop there is potential for passengers to disembark and try to cross in front of the tram as it departs the stop. Pedestrians using the crossing may also be uncertain whether an approaching tram is stopping or continuing through without stopping. However, swapping the offset of the Fraser Street tram stop so that the crossing is behind the tram may not be practical or safe. Sight distance requirements have dictated the layout of the tram stop and pedestrian crossing. Pedestrians need to cross two tracks in one bound. If passengers crossed behind a stopped tram the tram would obscure vision of trams approaching from the opposite direction. In contrast, the busway crossing shown in Figure 13 only requires pedestrians to cross one bus stop lane before reaching a pedestrian refuge. Pedestrians do not need to watch for a vehicle approaching from the opposite direction as there is only one lane to cross in the first bound. Clearly the only way to adapt the design proposed by Watkins et al. (2002) to the Melbourne tram network is to include a pedestrian refuge in between the two tram tracks, which would allow pedestrians to cross one track at a time. 15

19 Figure 15: Opposite tram stop platform, Middle Park stop on route 96, Melbourne - note fencing preventing direct pedestrian crossing between platforms 16

20 A pedestrian refuge or offset platforms may not be necessary to prevent pedestrians being struck by a tram from the second direction. Korve, Ogden, Siques, Mansel, Richards, Gilbert, Boni, Butchko, Stutts & Hughes (2001, p.75) show an operational approach to preventing collisions between recently disembarked passengers and a second tram approaching in the opposite direction. As shown in Figure 16, tram drivers are trained to use their vehicles to block the pedestrian crossing when a second tram is approaching a tram stop from the opposite direction. This prevents recently disembarked passengers from stepping onto the crossing and potentially being struck by the second tram. There may be opportunities to include similar operational instructions to tram drivers in Melbourne. 3.2 Stops in mixed traffic In comparison to the volume of research into exclusive and semi-exclusive operations there is little literature examining the safety of tram stops in mixed traffic operating environments. This may be due to the rarity of this type of operation around the world, other than in Melbourne. The scale of the mixed-traffic operation in Melbourne is... approximately six times larger than... in all the cities of the United Kingdom combined; it is equivalent to 76% of all the operations in all cities in the United States and 46% of all the operations in all cities in Germany. (Currie & Smith 2006, p.39). With such limited experience with operations in modern mixed traffic environments, it is no surprise that there is a smaller body of literature for this type of operation. Danger from passing traffic has been identified as a major passenger safety concern at kerb side stops, with an pedestrian to road vehicle accident rate of 25 per year across Melbourne s 1,200 kerbside tram stops (Currie & Smith 2006, p.39). In examining four alternative stop designs, Currie & Smith (2006, p.43) give a qualitative assessments of the level of passenger conflicts with road traffic at each type of stop, but there is no quantitative data available on the safety of passengers at each stop type. Currie & Smith s (2006) paper describes the new easy access stop design in detail and identifies that a primary benefit of the easy access stop is a reduction in vehicle speeds on the approach to and through the treatment, which enhances patron safety while boarding and alighting a tram (Currie & Smith 2006, p.44). This stop design appears to be unique in using traffic calming techniques to improve passenger safety. Safety zones, kerb side stops and other types of tram stops in Melbourne do not have this same traffic calming approach. Rather, the adjacent traffic lanes are mostly uninterrupted and public vehicles pass the stop at the similar speeds to other parts of the road network. With only limited literature available on tram stops in mixed traffic, perhaps examining bus stops can provide some insights into safety in this environment. Buses typically operate in type c.1 alignments, sharing road space with other vehicles. These are the same alignment types common in many tram operations in Melbourne. A number of research projects have examined the safety of school-aged children at bus stops in mixed traffic conditions. Baltes (1998) examined the safety of school-aged passengers by assessing driver s comprehension of the Florida school bus stop law. This law requires drivers to stop when a school bus is stopped and displaying a Stop sign and flashing lights. The results showed significant confusion on the part of motorists regarding their driving responsibilities... (and) the intent of various school bus signalizations (Baltes 1998, page 15). A similar law exists on Melbourne s streets, with drivers required to stop behind a tram and give way to alighting and boarding passengers at kerb side stops. While there is no available research on the comprehension of the Melbourne law by Australian drivers, if there is a similar confusion as in Florida this is cause for concern. The safety hazard for tram passengers may be greater as passengers enter and exit the tram from the traffic lane rather than the kerb side as is the 17

21 Figure 16: Tram blocking a pedestrian crossing to prevent collisions between a second tram and recently disembarked passengers, from Korve et al (2001) 18

22 case for Florida school bus passengers. Unger et al. (2002) looked at the safety outcomes for school-aged children by examining the medical records of children injured or killed at tram and bus stops in Austria over a 6 year period. The study examined 30 incidents and found that crash severity was high, with two deaths and nine other children with long term-physical health impacts. This study was not exclusive to urban areas and the relatively high speeds of passing vehicles in the Austrian countryside are likely to have contributed to the high level of fatalities and serious injuries. While trams tends to operate in an urban environment where speeds are much lower, high crash severity may still be a factor at tram stops. Korve et al. (1996) and Farran (2006) report that while only 10 percent of tram related accidents are collisions between pedestrians and light rail, these account for at least 50 percent of all tram-related fatalities. Unger et al. s (2002) study also used a questionnaire to obtain information about each crash. Nineteen of the 30 crashes involved a child being struck by a passing vehicle, while the remaining 11 crashes involved injuries caused by a bus or tram. Twenty four of the 30 crashes involved a lack of visibility between a driver and the child. Children stepping out onto the road from behind the bus or tram and being struck by a vehicle from the opposing direction was the cause of 10 of the 19 crashes involving cars. While the sample size for this study is quite small and the results may not be statistically significant for assessing risk, there is an indication that lack of visibility of passengers for passing motorists and people stepping out from behind public transport vehicles are particularly dangerous at stops in mixed traffic environments. New technology may help to reduce the risk to passengers in mixed traffic environments, and at transit stops in general. A range of sensing technologies are identified by Bu & Chan (2005) as having potential for reducing the likelihood of pedestrians being struck by buses. In future installations sensors will identify pedestrians, while a computer system analyses the movement of the bus and pedestrians and warns the driver if there is the potential for a collision. When these systems are fully developed they may be a useful tool for bus drivers to avoid collisions caused by lack of visibility. However, the high costs are identified as a limiting factor for full-scale roll out of these types of systems. It may be that a more modest approach could deliver similar results without the high costs of complicated sensing and computing technology. A video camera rear-view system has been installed on some of the modern trams in Melbourne, as shown in Figure 17. This system allows drivers to view boarding passengers from a number of angles that would not be possible with line-of-sight and mirrors. Little other research on bus stop safety in mixed traffic conditions has been found during this review. Kerb side bus stop design is not particularly well covered in the literature, which is probably due to the inherent safety of passengers waiting on the footpath and boarding the bus directly from the kerb. This safety is not present for tram passengers due to the location of tram tracks in the centre of the road in Melbourne. Passengers need to cross traffic lanes to board trams or reach the stop and this creates a risk of passenger to public vehicle collisions. There is a particular lack of research in passenger to public vehicle collisions in and around tram stops. Some research only considers tram to passenger and tram to vehicle collisions, for example Korve et al. (1996, pages 3-4) Passenger to vehicle collisions at or near tram stops are not considered, primarily due to a lack of data collection by transit operators. This appears to be a significant oversight by operators as passengers are exposed to the risk of collisions with vehicles when accessing tram stops, even though they are still technically pedestrians. 3.3 Safety for passing public vehicles The literature on safety at transit stops that has been reviewed so far focuses on preventing pedestrians or passengers from being struck by transit vehicles or passing public vehicles. Col- 19

23 Figure 17: Rear view video system on modern Melbourne tram, route note video is displayed on the two smaller screens on either side of the main screen lisions between public vehicles and transit vehicles are also examined in the available literature, for example by Farran (2006) and Korve et al. (1996). However, the safety implications of physical infrastructure at tram stops for passing vehicles is not well considered in the reviewed literature. This is likely due to limited experience with public vehicles crashing into transit stops for most public transport operators. In exclusive and semi-exclusive operating environments there is little opportunity for public vehicles to impact stops, due to the separation between traffic lanes and the public transport alignment. In mixed traffic operations, these types of incidents are not relevant to the majority of public transport operators running buses with kerb side stops. If a motorist crashes into a kerbside bus stop it is similar to any other crash involving roadside furniture such as a power poles, fences and trees. Treatments to reduce the frequency and severity of these sorts of crashes have been well established by road authorities, and typically consist of removing the hazards, relocating the hazard further from the edge of the traffic lane, or protecting the hazard with a crash barrier (Austroads 2004, p.75). Centre of the road tram stops may pose a greater hazard to passing public vehicles than kerb side stops. Purdy (2007) presents a number of photographs, reproduced in Figures 18 and 19, of public vehicle collisions with centre of the road tram stops. These types of incidents are clearly hazardous to occupants of the vehicle and passengers waiting at stops. Along with other obstructions in the roadway, such as traffic islands, speed humps and roundabouts, centre of the road tram stops may require treatment to reduce the likelihood and severity of crashes. 20

24 Figure 18: Public vehicle to tram stop crash in Melbourne, from Purdy (2007) Figure 19: Public vehicle to tram stop crash in Melbourne, from Purdy (2007) 21

25 3.4 Summary of key findings from the literature review This literature review has considered safety at a range of different types of transit stops in exclusive, semi-exclusive and mixed traffic operating environmented. The key findings that apply to safety at tram stops are: Trams operate in a wide variety of environments, from fully-grade separated rights of way to mixed traffic. The wide range of environments results in a wide range of stop types and the need for different approaches to safety at stops, even along a single route. Experiences from urban train networks may be transferable to trams operating in exclusive or separate rights of way. Experiences from busways and bus rapid transit systems may be transferrrable to trams in semi-exclusive operating environments. Experience from street-running bus operations is less adaptable to trams in mixed traffic operations. Buses typically stop at kerb side stops, while trams typically run down the centre of the road. A range of treatments are available to improve pedestrian safety when crossing semiexclusive tram alignments including: passive warning signs, active warning signs, pedestrian bedstead barriers, pedestrian refuge areas Z-crossing channelization, manually operated self closing swing gates automatic pedestrian gates. tactile pavement markers, train-coming icons and directional indications, audio warnings, and pavement markings of the dynamic envelope of trams. Four key approaches to pedestrian safety at tram track crossings are: ensuring that pedestrians are aware of the crossing, controlling the pedestrian path across the crossing, providing sight distance to approaching trams, even if the crossing is actively controlled, and public outreach and education programs. Platform layout can be used to encourage pedestrians to cross safely. Locating the crossing point behind the stop can prevent passengers stepping in front of the tram after disembarking. However, passengers stepping out from behind a vehicle are particularly at risk of being struck by other passing vehicles in mixed traffic environments, so these treatment is only appropriate in semi-exclusive alignments where a pedestrian refuge can be provided so pedestrians cross only one direction at a time. 22

26 Operational instructions to drivers to block a pedestrian crossing may help to preventing pedestrians being struck when a second tram is approaching a stop from the opposite direction. Laws demanding public vehicle drivers stop behind a transit vehicle that is boarding or disembarking passengers are confusing and little understood by motorists. Treatments that rely exclusively on these laws may be inherently unsafe. Pedestrian detection using sophisticated sensing technology may become useful, particularly in reducing the incidence of passenger to tram collisions. However, this technology is expensive and still under development. More modest current technology, such as vehicle mounted video systems, can allow drivers to avoid passenger to tram collisions without use of high cost new technology. Centre of the road tram stops may require treatment to reduce the likelihood and severity of crashes, similar to other obstructions in the roadway, such as traffic islands, speed humps and roundabouts. 23

27 4 Crash Data Analysis Data is collected by a number of agencies to record crashes and incidents on roads and trams in Melbourne. This data is used for overall analysis of crash trends, to identify and treat crash locations, research purposes and to meet statutory reporting requirements. Crash data can be used to compare crash rates before and after the installation of a new facility, or between a group of treated sites and a control group of untreated sites. This can allow an assessment of the effectiveness of a particular treatment in reducing the rate of crashes. Austroads (2004, pp.77-79) provides an indication of the effectiveness of a wide range of treatments in decreasing certain types of crashes, which is based on before and after or control group studies across a large range of sites and projects. Using a similar approach, this section will analyse crash and incident data at tram stops that have recently being upgraded. This analysis aims to assess whether the new platform and easy access types of stops increase or decrease the rate of crashes and incidents when compared to the safety zone and kerb side types of stops. A before and after comparison approach has been adopted, as this reduces the externality effects of site conditions on the crash rates. There is certainly scope for a direct comparison between a group of treated sites and a control group of sites, but this has been beyond the scope of this project in the available time. Two sources of crash data were available for this analysis: the VicRoads CrashStats program is an online database of all casualty crashes in Victoria (VicRoads 2009). Yarra Trams, the Melbourne tram operator, collects data for internal use which includes the details of crashes and incidents involving trams and passengers (Yarra Trams 2009b). 4.1 VicRoads Crashstats data Victoria Police is responsible for the reporting of all crashes on Victorian roads that result in injury or death. This data is transferred to VicRoads and loaded into the CrashStats accident summary and mapping tool, which is accessible via the internet. Data on individual crashes can be downloaded including information on date, time, location, vehicles involved and seriousness of injuries. Crashes are recorded using the Definition for Coding Accidents (DCA) codes (Austroads 2004, pp ), which provides a system to categories and identify the movements and types of crashes. The CrashStats database currently contains data from 1987 to the end of Not all crashes for this period are available, with approximately 1% of incidents for 2006 and 2% for 2007 yet to be approved by Victoria Police for processing by VicRoads (VicRoads 2009). Unfortunately for this project there is a discontinuity in the data series at the end of A new data collection system was introduced by Victorian Police in December 2005 which changed the number of incidents which are reported. VicRoads advises is that it is not valid to compare crash rates from before 2006 with crash rates after 2005 due to this data discontinuity (VicRoads 2009). Therefore, in this project tram stops constructed between 1999 and 2005 have been analysed separately to tram stops constructed in 2006 and 2007, to account for the discontinuity in the data set. For tram stops constructed between 1999 and 2005 five years of before data has been compared with the data from between the construction date and the end of There is much less data available for trams stops constructed in 2006 and 2007, and whatever is available has been presented. For all stop 4 months of data from around the construction date has been discarded to account for the interruption to traffic during the construction works and to allow for interim effects as road users become accustomed to the changed road conditions. 24

28 Three types of crashes have been examined in this analysis: Crashes involving both a pedestrian and a public vehicle in the vicinity of a tram stop, Crashes involving both a tram and a pedestrian in the vicinity of a tram stop, and Crashes involving a vehicle hitting a tram stop. The second and third types of crashes are relatively simple to identify in the CrashStats database as there are few crashes involving trams and there is a specific data field for hit safety zone. However, crashes involving a passenger and a public vehicle are more difficult to identify, as it is not possible to identify from the crash data whether a pedestrian involved in a crash was accessing or departing a tram stop or just happened to be crossing the road nearby. For this reason no attempt has been made to differentiate between pedestrians and passengers other than only considering crashes during typical tram service operating hours (5am to 1am) on the same leg of an intersection as the tram stop. New tram stops in exclusive rights of way and operating in the central median of arterial roads have not been considered in this data analysis. Replacement of an existing ground level boarding area with a raised platform at these stops is not considered to result in significant changes to the road environment, and the focus of this study in on stops in mixed traffic environments and in close proximity to traffic lanes. Upgrades of tram terminuses and multiplatform interchanges such as St. Vincent s Plaza and Melbourne University have not been considered due to the large number of variables and external factors at these sites. There are very few safety zone to platform tram stop conversions with sufficient data to allow a meaningful comparison of before and after crash rates. Of the 31 platform tram stops installed between 1999 and June 2005, 10 were new stops on tram line extensions rather than upgrades to existing safety zone stops. A further 11 stop upgrades were in exclusive rights of way or wide medians on arterial roads, well separated from passing traffic (Yarra Trams 2009a). Of the remaining ten stops it has only been possible to analyse data from seven stops, as one stop was removed as part of the Citylink project and the other two platform stops are at major interchanges. A summary of the crash data from these seven stops is presented in Table 3. The average crash rate was approximately one crash every eight months before the new stop construction, when the stops were safety zones. After conversion to platform stops the average crash rate was one crash every 15 months. However, with a sample of only six platform stops there is insufficient data to be confident that there is a true reduction in crash rates due to the new type of stop. In particular, the stop with the largest reduction in crash rates is the St Kilda Road stop at Federation square. It is likely that this large reduction in crash numbers is due more to the relocation and consolidation of the stop to a single location on the south side of the Flinders Street intersection, rather than to the new design. Before the upgrade, there was a southbound tram stop on the north side of Flinders Street, as well as a northbound stop on the south side of the intersection. With the consolidation of the stop to a single side of the intersection, there are fewer places where passengers are exposed to the hazards of passing vehicles: this may be the reason for the significant decrease in pedestrian related crashes in the vicinity of this tram stop. Removing the St Kilda Road Federation Square tram stop from consideration, the average crash rate before the stop upgrades was one crash every 15 months. The average crash rate after the stop upgrades was one crash every 26 months, which is still a reasonable reduction. However, with only 157 months of data across 5 platform stops we cannot be confident that this result would apply to all platform stops or is a true indication of average crash rates. 25

29 There was only one easy access stop constructed between 1999 and June There were no crashes in the before or after periods at this site and hence no conclusions can be made about this type of stop from this data. The second period of available CrashStats data was from 2006 and Unfortunately, 27 of the 29 stops constructed between July 2006 and June 2007 were in exclusive rights of way, or in the median of wide arterial roads, where there are few hazards to passengers that have been changed as part of the stop upgrade works. Crash data from the two stops upgraded between July 2006 and June 2007 in mixed traffic environments are summarised in Table 4. With only one crash recorded at these two stops, there are no significant conclusions that can be drawn from this data. At least 36 of the 67 stops that have been upgraded since June 2007 are within mixed traffic environments or close to traffic lanes (Yarra Trams 2009a). When the 2008 crash data is available on the VicRoads CrashStats database it may be possible to analyse crash rates at these stops and have a statistically significant sample size. 26

30 Table 3: Before and after crash frequency at new tram stops, 1999 to 2005 Before Stop Construction After Stop Construction Stop Data Period Public Vehicle Tram Public Vehicle Data Period Public Vehicle Tram Public Vehicle Location (months) vs Pedestrian vs Pedestrian Hit Tram Stop (months) vs Pedestrian vs Pedestrian Hit Tram Stop Platform Stops St Kilda Rd at Federation Square Collins St at Swaston St Collins St at Spring St Victoria St at River Bld Victoria St at Burnley St Collins St at Spencer St Total Total excluding St Kilda Road Easy Access Stops Danks St at Harold St 27

31 Table 4: Before and after crash frequency at new tram stops, 2006 to 2007 Before Stop Construction After Stop Construction Stop Data Period Public Vehicle Tram Public Vehicle Data Period Public Vehicle Tram Public Vehicle Location (months) vs Pedestrian vs Pedestrian Hit Tram Stop (months) vs Pedestrian vs Pedestrian Hit Tram Stop Platform Stops Commercial Rd btw Punt and St Kilda Easy Access Stops Victoria Av at Birdport St 28

32 4.2 Yarra trams data Yarra trams has provided data from the internal Yarra Trams operations database from May 1994 to May The database contains descriptive information of incidents that occur across the network and details how tram services were effected and what operational changes were made to reroute or adjust services. The database is not a formal record of crashes like the CrashStats database, and many entries lack information such as exact location, extent of injuries or vehicle directions. Rather, the database is more of an operational log and contains narratives of the incidents and actions taken by Yarra Trams staff. The Yarra Trams database does not include all incidents at or near tram stops. Only incidents that directly involve or delay trams, or that are considered important by tram drivers are reported to the Yarra Trams operations centre (Reynolds 2009). Hence, the data set will not include all incidents and it is likely that public vehicle vs pedestrian crashes will be under reported. The lack of a formal incident coding process has made analysis of the incidents somewhat difficult. While incidents are categorised based on incident type, there is no additional coding to allow similar incidents to be identified. In a number of cases the incident type does not closely match the description of incident, and there appears to be inconsistency in the way that the type of incident is recorded. For example, the wayside pedestrian knockdown and wayside passenger knockdown categories both contain incidents involving trams striking passengers and non-passengers, as well as incidents involving public vehicles. Yarra Trams should consider implementing a formal incident reporting and data collection program using the techniques outlined in chapter 5 of Austroads (2004), which details the approaches to coding, collecting and storing crash data. Implementing a formal crash reporting system could provide a more useful data set from the analysis of crash locations and incidents across the tram network, and allow direct comparison with the VicRoads CrashStats database. Each individual incident has been briefly examined and categorised into the same three groups as used in the CrashStats analysis; public vehicles vs pedestrians, trams vs pedestrians and public vehicles vs tram stop infrastructure. However, due to the frequency of incorrect categorisation of incidents in the database we cannot be certain that all of these types of incidents have been identified. No attempt has been made to classify incidents on severity, so unlike the CrashStats data this analysis will include some incidents where no one was injured. Table 5 shows a summary of incidents before and after the construction of platform stops. Only stops with at least one crash are shown in the table. Incident rates are low, with only 103 incidents recorded across 67 stops. The remaining 86 stops with platforms have had no incidents either prior to or after the new stop construction, between May 2004 and May At the bottom of Table 5 incident rates and frequencies are shown across all platform stops, including stops without any incidents. Following the installation of the new platform stops the rate of public vehicle vs pedestrian incidents has doubled, and the rate of tram vs pedestrian incidents has increased by approximately 15%. The rate of vehicles striking infrastructure at stops has dropped by approximately 20%. The data shown in Table 5 shows incident rates at all platform stops, including stops that were constructed before the start of the available data. These stops without before data may be influencing the results. To allow a direct comparison of before and after crash rates Table 6 shows the incident rates only at stops with at least 6 months of before and after data. This limited data set shows a similar result, with the rate of pedestrian related incidents increasing after the construction of the new platform stops, and the rate of vehicles striking infrastructure dropping. The increase in pedestrian related incidents is quite unexpected. However, as outlined above there is considerable uncertainty in the reporting of pedestrian related incidents. Also, there is a limited amount of data, with only 103 incidents occurring at platform stops in the 5 years of available data. This number of incidents may be too small a sample to allow statistically 29

33 significant conclusions to be reached. The infrequency of collisions can make it difficult to assess the safety performance of tram related facilities. In a study of pedestrian at-grade crossing treatments Siques (2002) instead used risky pedestrian behaviour as a measure of safety and treatment effectiveness. Because risky behaviour movements are more frequent than collisions, they can be used as a surrogate safety indicator in lieu of the number of collisions (Siques 2002, p.66). This approach also has the advantage of not relying on data collection over a long period of time by outside parties, such as tram operators and the police. Rather, the researcher or engineer can directly collect data through observational surveys over a short time frame to realistically assess the safety performance of a facility. Adapting this method to tram platform and easy access stops is an avenue for future research in this area that is likely to provide a better assessment of safety performance than analysis of infrequent crashes and incidents. There have been insufficient time and resources available in this project to undertake the detailed field observational surveys of risky pedestrian behaviour. Instead an alternative method of assessing road safety risk has been used; the road safety audit, which is described in the next section. 30

34 Table 5: Incidents at platform tram stops, Yarra Trams data Before Stop Construction After Stop Construction Stop Data Period Public Vehicle Tram Public Vehicle Data Period Public Vehicle Tram Public Vehicle Location (months) vs Pedestrian vs Pedestrian Hit Tram Stop (months) vs Pedestrian vs Pedestrian Hit Tram Stop Bourke St at Spencer St Bourke St at Spring St Bourke St at Swanston St Bourke St at Spring St Burwood Hway Stops 59,61-63 Burwood Hway Stop 64,65,67-69 Burwood Hway at Blackburn Rd Burwood Hway at Stanley St Casino West Collins St at Elizabeth St Collins St at Swanston St Collins St btw Spencer/King Collins St Collins St Temp Terminus 31

35 Table 5: continued: Incidents at platform tram stops, Yarra Trams data Before Stop Construction After Stop Construction Stop Data Period Public Vehicle Tram Public Vehicle Data Period Public Vehicle Tram Public Vehicle Location (months) vs Pedestrian vs Pedestrian Hit Tram Stop (months) vs Pedestrian vs Pedestrian Hit Tram Stop Federation Square Fitzroy St at Park St Flemmington Rd at Grattan Flinders St at Elizabeth St Flinders St at King St Flinders St at Spencer St Flinders St at Spring St Flinders St at Swanston St Melbourne University North Port Peel St Nth Melbourne Plenty Rd Stops Plenty Rd Stops

36 Table 5: continued: Incidents at platform tram stops, Yarra Trams data Before Stop Construction After Stop Construction Stop Data Period Public Vehicle Tram Public Vehicle Data Period Public Vehicle Tram Public Vehicle Location (months) vs Pedestrian vs Pedestrian Hit Tram Stop (months) vs Pedestrian vs Pedestrian Hit Tram Stop Plenty Rd Stops Plenty Rd Terminus Queens Pde Spencer St at Collins St South Melbourne Light Rail St Kilda Rd Arts Centre St Kilda Rd Stops 31,32,34,35 Victoria Pde at Wellington Victoria Pde at Brunswick St Whitehorse Rd Nelson Rd Total - all platform stops Frequency per year Average period years between incidents at average stop 33

37 Table 6: Incident rates at platform stops with at least 6 months of before and after data Before Stop Construction After Stop Construction Data Period Public Vehicle Tram Public Vehicle Data Period Public Vehicle Tram Public Vehicle vs Pedestrian vs Pedestrian Hit Tram Stop vs Pedestrian vs Pedestrian Hit Tram Stop Incidents months months Frequency per year Average period years between incidents at average stop 34

38 5 Road safety audits Crashes are relatively rare events on any section of road. While this low crash rate is positive from an overall road safety point of view, it can make it difficult to quantitatively compare crash sites. The results described in Section 4 is an example of how the low number of crashes can prevent statistical analysis of crash rates. In this case, there have been too few crashes to allow a meaningful comparison of road safety risk before and after the installation of platform stops using crash data alone. As a result this project has included road safety audits to identify and analyse road safety risk at each of the four main types of tram stop in Melbourne. A road safety audit is a formal examination of a future road or traffic project or an existing road, in which an independent, qualified team reports on the project s crash potential and safety performance (Austroads 2009, p.7). The objective of an audit is to highlight potential road safety problems that the project manager can then consider rectifying within the overall constraints of the project. During an audit the site is examined under a range of conditions, including at night. The auditors also examine the design plans and any other background information to assist in making a number of findings about road safety at the site. An audit is not a check of the design against standards, but rather an examination of the road s fitness for purpose and whether the project will work safely for its expected users (Austroads 2009, p.4). Audits can be undertaken at all stages of design and construction including: concept, preliminary or detailed design - to assess the safety of the proposed design. construction - to assess the safety of traffic management during the roadworks. pre-opening - to assess the constructed road and pick up any remaining or introduced safety issues. existing conditions - to assess an operational section of road. Road safety audits have been undertaken as part of many of the tram stop upgrade projects in Melbourne at concept design (Price 2007b), functional design (Desmond & Natalizio 2007), detailed design (Harris 2006) and post-opening (Price 2007a) stages. These audits consider the road safety implications of the new tram stops and highlight safety problems with the designs and finished works at each site. However, the focus of these previous audits is on improving the new stops, rather than making any comparisons of the level of road safety with the previous safety zone and kerb side stops. The audits are also focused on the specifics of the individual stop conditions and design, including such details as linemarking, signs and individual lane widths. While these audits are a valuable resource, this project is focused on the safety of platform and easy access stop concept as a whole, rather than the features of specific stops. A road safety audit is undertaken by a small team of people with road safety engineering or other relevant experience. While the make up and size will differ for different types of audits, a road safety audit team will typically include traffic engineers, road designers, police officers and others with relevant experience who can independently assess the road environment and any proposed designs. Audits undertaken as part of this project were undertaken by just a single auditor, Julian Reynolds on Sunday 3 and Monday 4 May. Julian is a VicRoads accredited road safety auditor and has participated as a team member on over 90 road safety audits during the past 4 years, including preliminary design, detailed design, post-opening, roadworks traffic management and existing conditions audits. He also has experience in road design, road fatality investigation and blackspot projects. While it would have been ideal to have a larger audit team including an expert on tram operations, this was not possible due to the time and resource constraints on 35

39 this project. It would also have been ideal to have an independent road safety auditor, separate from the rest of this project, undertake the audits, but again this was not possible due to time and resource constraints. A number of example platform stops, easy access stops, safety zone and kerb side stops have been examined. No attempt was made to make the sites a representative sample, rather the sites were chosen to include a range of different road environments, operating conditions and passenger volumes. The sites include: Safety zone stops 1. Elizabeth Street and Latrobe Street 2. Elizabeth Street and Bourke Street 3. Swanston Street and Grattan Street 4. Lygon Street and Park Street Platform stops 5. Collins Street at Swanston Street 6. Collins Street between Russell and Exhibition Streets 7. Bourke Street at Queen Street 8. Fraser Street - light rail stop on route Middle Park - light rail stop on route 96 Kerb side stops 10. Birdport Street and Ferrars Street 11. Swanston Street and Latrobe Street 12. Royal Parade and Gatehouse Street Easy Access stops 13. Victoria Avenue and Birdport Street and Merton Street 5.1 Road safety audit findings The key outcome of a road safety audit is a formal report that describes the issues identified by the audit team. The audit team might also provide recommendations on how the project manager might rectify each of the identified road safety issues. The tables on the following pages describe the audit findings for each of the four types of stops. Where relevant, stops have been identified using the numbers shown in the above list. 36

40 Table 7: Safety Zone - table of audit findings and recommendations Findings Recommendation Priority 1 There are gaps in the railings of many safety zones. Some safety zones have ap- Important proximately 5 metre long U shaped railings, with 1 metre gaps between each railing. Many passengers were observed crossing to and from the safety zone through these gaps, rather than walking to the adjacent formal pedestrian crossing points at traffic signals. Other safety zones (such as the southern leg at site 2) have fully connected pedestrian fencing and shelters without gaps. However, many pedestrians were observed walking out the approach side of the tram stop, near the safety zone prowl, rather than using the pedestrian crossing at the traffic signals. Pedestrians who cross the traffic lanes part way along the safety zone or close to the safety zone approach prowl may be at greater risk of injury as public vehicle speeds will be higher on the approach and pedestrians have to cross in gaps in the traffic stream. Many pedestrians were observed crossing in between stationary vehicles which may increase the chances of a pedestrian being struck by another vehicle or a cyclist, or when the traffic signals change to green. Pedestrians crossing from part way along a safety zone are unlikely to be expected by approaching drivers. See Figures 21 and Many disembarking passengers were observed crossing to the opposite side of the road around the rear of the tram. These pedestrians are not visible to drivers approaching in the opposite direction. The tram blocks sight distance and these pedestrians will have to step out onto the tram tracks to see around the tram they just got off. These pedestrians are at risk of being struck by trams or vehicles traveling in the opposite direction. See Figure Some of the safety zones do not have a shelter in the safety zone. During inclement weather many passengers will wait in shelters, under shop awnings or in front of buildings on the footpath. When a tram arrives these passengers may cross the traffic lanes in an unsafe manner to avoid being left behind by a tram. Prevent or discourage tram passengers from crossing traffic lanes other than at formal pedestrian crossings. Where possible remove gaps in the safety zone fencing. However, this will not prevent pedestrians crossing at the wrong end of the safety zone near the approach prowl. Additional fencing, fencing on the kerb, instruction signs and public education campaigns could be considered as methods to reduce the number of pedestrians crossing to and from safety zones in inappropriate locations. Where possible prevent or discourage passengers from crossing behind a tram. Pedestrian fencing on the opposite kerb line or between the tram tracks could be used to physically prevent passengers crossing behind a tram. Where ever possible provide shelters at the passenger boarding location rather than on the footpath. While operational instructions to drivers to wait for any passengers on the footpath during inclement weather could prevent passengers being left behind, it is unlikely that passengers will stay on the footpath if a tram is approaching or wait for traffic signals to change to the pedestrian phase. Important Noted 37

41 Table 7: continued: Safety Zone - table of audit findings and recommendations Findings Recommendation Priority Urgent 4 The passenger waiting area in safety zones ranges from approximately 1 to 3 metres wide. There is very little separation between waiting passengers and approaching trams. At busy tram stops there is limited space for disembarked passengers to safely walk along the safety zone while a tram departs. The edge of the trams dynamic envelope is marked with a white line, but this may be insufficient to separate pedestrians and trams. There is no guidance for visually impaired passengers to stay clear of trams at safety zones. See Figure 23 The narrow width of the passenger waiting areas and lack of physical separation may increase the risk of a passenger being struck by a tram. 5 The pedestrian fencing separating the safety zone from the adjacent traffic lane does not appear to be crash worthy. At some safety zones there are shelters, but the glass panels separating passengers from the adjacent traffic lane do not appear crash worthy. There are short sections of barrier type kerb between the fencing and the traffic lane, but this is unlikely to prevent errant vehicles from crashing into or through the fencing. Many passengers lean against the fencing or walls of the shelter and would be hit if there was a vehicle impact. The fencing, kerb and glass panels may be insufficient to prevent a passenger from being struck and injured by an errant public vehicle. See Figure There is insufficient separation between the traffic lanes and the safety zone fencing. Passengers leaning against the safety zone fencing may be at risk of being clipped by passing vehicles, in particularly the wing mirrors of trucks. Increase the width of safety zones. Improve and formalise the separation between the passenger waiting areas and the area used by the tram, perhaps by installing tactile surface markers along the separation line. Consider approaches to prevent passengers from leaning against the existing fencing at safety stops or increase the separation between passengers and the adjacent traffic lanes. Consider installing crash worthy nonflexible safety barrier to separate waiting passengers and the adjacent traffic lane. Increase the separation between the traffic lane and the safety zone fencing. Where possible discourage passengers from leaning against the fencing. Urgent Noted 38

42 Table 7: continued: Safety Zone - table of audit findings and recommendations Findings Recommendation Priority Urgent 7 Most safety zones have a large yellow island prowl on the approach side. The shape of the prowl may increase the chance of a vehicle flipping. The prowl is not a crash worthy end treatment and may also pose a spearing hazard to errant vehicles. The prowl is typically located directly in the path of the right wheel for approaching vehicles and is likely to be hit by any driver who fails to move left on the approach to a safety zone. Delineation of the approach to the safety zone islands is generally of a reasonable standard. See Figure The safety zone prowl reduces the available carriageway width for public vehicles. Kerb side parking the parking generally stops a short distance prior to the prowl. The narrowing of the carriageway at the prowl may act as a pinch point and increase the risk of side swipe crashes or of cyclists being struck by a vehicle. 9 The safety zone layout at traffic signals prevents the provision of a separate right turn lane. Hook turns are used. with right turning vehicles queuing on the left side of the road within the intersection, clear of through vehicles. These queued vehicles then turn right as the lights change to red. The hook turn maneuver is an uncommon and unconventional treatment which may be confusing for drivers, particularly those unfamiliar with driving in the Melbourne city environment. 10 Pedestrians regularly jay walk at the signalised crossings associated with tram stops. Passengers rushing to catch a tram often cross against the lights, from both sides of the road. Passengers disembarking from a tram occasionally dash across in front of the tram before it departs the stop, or cross traffic lanes against the lights. This type of pedestrian behaviour may increase the risk of pedestrian vs tram or pedestrian vs public vehicle crashes. Remove the safety zone island prowls and replace with a crash worthy end treatment such as a crash cushion or energy absorbing end treatment. Review delineation on the approach to safety zone islands and ensure that it is of a high standard and regularly maintained. Where possible consider increasing the distance between the end of kerb side parking and the island prowl. Ensure that any lane narrowing or redirection is well delineated and that there is sufficient space and separation for cyclists. While this audit will not investigate the safety of hook turns in detail, relocating safety zones away from signalised intersections may allow space for a right turn lane to be provided. Engineering treatments to prevent passengers jay walking at traffic signals are unlikely to be practical due to the limited space in safety zones. Education and enforcement campaigns may be considered to reduce the incidence of jay walking. Noted Noted Important 39

43 Table 8: Platform stops - table of audit findings and recommendations Findings Recommendation Priority 1 The platform stops are fully fenced, with access points at either end of Noted the platforms. One end of the platform has a formal pedestrian crossing, either at signals or a zebra crossing. The other end of the platform stops have an informal crossing point where pedestrians give way to vehicles. In comparison to the safety zone, pedestrian movements are more controlled, with no pedestrians crossing traffic lanes part way down the stop. However, at stop 6 many pedestrians were crossing east of the zebra crossing to short cut around the pedestrian fencing. At stop 5 a number of pedestrians were observed walking in the traffic lane to short cut diagonally to the platform access points. Unpredictable pedestrian movements in the vicinity of crossing points may be unexpected by approaching drivers and increase the risk of a pedestrian vs public vehicle collision. See Figures 26 and The uncontrolled crossings where pedestrians give way to vehicles may be hazardous due to the variable traffic conditions, particularly where the stop is located on the approach to a signalised intersection. Traffic speeds in the vicinity of the uncontrolled crossings at stop 5 varies between 40km/hr, as vehicles approach the intersection on a green, and stationary, as vehicles queue when the lights are red. During the site visit drivers and pedestrians were observed visually negotiating who should go first when traffic speeds were low. However, traffic speeds are not consistently low and the area is not a formal shared zone. The lack of clear priority at the uncontrolled pedestrian crossing points and rapidly changing flow conditions may increase the risk of a pedestrian vs public vehicle crash. See Figure During the site visits no disembarking passengers were observed walking around the back of trams to cross the road. However, there is nothing preventing pedestrians stepping off the platform and crossing to the opposite platform directly. Howver, in this case there is risk of being struck by trams only and not public vehicles as well. The platform stops represent a significant improvement in pedestrian control over the safety zones. However, additional pedestrian fencing may be needed on the kerb side to channel pedestrians towards crossing points and enforce 90 degree crossings of traffic lanes. Consider installing formal crossing treatments such as zebra crossing or signals at all access points to platform stops. In locations with high volumes of crossing pedestrians consider installing pedestrian fencing between the two tram tracks to prevent pedestrians crossing the tram tracks other than at designated crossing locations. Offset platform design could be considered to relocate crossing points to behind trams, as used is the New Britain to Hartford Busway (Watkins et al. 2002). Important Noted 40

44 Table 8: continued: Platform stops - table of audit findings and recommendations Findings Recommendation Priority 4 The passenger waiting area is approximately 3 metres wide and clearly separated Comment from the tram operating area by the platform edge. 5 Shelters are provided at the passenger waiting area. During inclement weather passengers will not need to cross from kerb side shelters when trams arrive. 6 The fencing separating the platform from the adjacent traffic lane does not appear to be crash worthy, although it is more substantial than fencing at safety zones. In some locations the handrail or fencing is exposed to passing traffic and act as a snagging or spearing hazard for errant vehicles and motorcyclists. There is a barrier kerb separating the platform from the traffic lane in most locations. As in the safety zone, passengers leaning against the fencing or shelters may be at risk if an errant vehicle crashes into the platform stop. However, separation between passengers and passing traffic is larger than at safety zones. Average traffic speeds at platform stops may be lower due to the reduction in the number of traffic lanes, therefore lowering the risk of errant vehicles. However, it is unclear if the speed of the quickest vehicles has also decreased. There may still be a similar risk of errant vehicles crashing through the fencing or shelters at platform stops as at safety zones. 7 Separation between the traffic lane and the pedestrian waiting area is larger at platform stops than at safety zones. Handrails are located on the inside of fences. The fencing is chest high at stop 5, but waist height at stops 6 and 7. With seats and shelters provided at the stop there appears to be less incidence of waiting passengers leaning against the fencing. However, a person leaning against or over the fencing may still be at risk of being struck by part of a passing vehicle, such as a wing mirror. 8 Bollards are located on the approaches to the platform stops to prevent vehicles crashing into the fencing and stops. The bollards are likely to be safer than the prowls at safety zones as they will not flip errant vehicles. However, head on impact with a solid bollard at speed is still a significant hazard. Delineation on the approach to the bollards is generally of a high standard. See Figure 28. The risk of tram vs passenger collisions should be significantly reduced when compared to narrow safety zones Consider installing crash worthy safety barrier to separate the platform stop and the adjacent traffic lane. Consider adopting the higher style of fencing used at stop 5 across all platform stops. Consider installation of crash worthy frangible bollards, crash cushions or other devices that would reduce the severity of a head on collision with the end of the platform stop. Comment Urgent Noted Important 41

45 Table 8: continued: Platform stops - table of audit findings and recommendations Findings Recommendation Priority 9 The width of the carriageway reduces to a single traffic lane passing the platform Important stops. The narrow lane width may be a pinch point for cyclists. At stop 5 an approximately 500mm wide cycle lane has been installed between the trafic lane and the kerb along the length of the platform. This treatment may encourage drivers to overtake cyclists, but the cycle lane is too narrow to provide enough separation between cyclists and vehicles. There may be a greater likelihood of public vehicle vs cyclist collisions in the vicinity of platform stops. 10 The positioning of some platform stops away from intersections provides additional space for traffic lanes. However, hook turns are still in place at intersections in the Melbourne CBD. 11 At zebra crossing access points public vehicles give way to pedestrians accessing the platform stop. Pedestrians continuing to cross the road then need to give way to any trams. This situation requires two different actions by pedestrians at two similar situations in very close proximity. Pedestrians may continue to cross the tram tracks as if trams are going to give way to them as well. At stop 6 a Z crossing treatment has been installed, which slows pedestrians down on the approach to the tram tracks. At other stops there is no impediment to pedestrians continuing to cross the tram tracks in a straight line after crossing the traffic lane. Pedestrians may cross the tram tracks immediately after crossing a zebra crossing and not realise that they need to give way to trams, thereby increasing the risk of a tram vs pedestrian collision. See Figure The path through the intersection for westbound traffic approaching stop 5 may be unclear. The stop is on the downstream side of the intersection and there is little delineation on the immediate approach to the bollards and platform. There is a lateral shift for through traffic of approximately 3 metres to the left, due to the left turn lane on the east side of Swanston Street. The path may be unclear for drivers. Many passengers are waiting for the pedestrian phase in this area while westbound traffic crosses the intersection. There may be an increased risk of westbound vehicles striking the bollards, platform or passengers on the downstream side of the intersection. Remove the cycle lane so that cyclists have to take the lane, or increase the cycle lane width to provide greater separation. Consider provision of cycle lanes at other platform stops. Otherwise, ensure that cyclists can safely merge into the traffic stream on the approach to platform stops and take the lane. Hook turns still in place and there does not appear to have been a significant change to road safety risk. Give way to trams signs have been installed on fencing at pedestrian crossings, but these are unlikely to have a significant impact on pedestrian safety. Consider installation of z-crossings, bedstead barriers, manually operated self closing swing gates or similar treatments recommended by Korve et al. (1995, pp ) to slow pedestrians down and stop them crossing the tracks directly. Where possible provide a straight alignment for drivers travelling through an intersection with a platform stop on the down stream side. Consider installing additional delineation for through drivers, although this may be limited by a need to keep the intersection clear of confusing linemarking for drivers traveling in other directions. Review the safety of pedestrians waiting in the centre of the road at the end of the platform stop and consider installing a protective island treatment. Noted Important Important 42

46 Table 8: continues: Platform stops - table of audit findings and recommendations Findings Recommendation Priority Important 13 Recently disembarked passengers at stop 5 tend to congregate at the end of the platform stop near the intersection waiting for the signals to change and allow them to cross to the kerb. The area used to wait is exposed to passing traffic. During periods of high passenger flows the area can become quite congested and pedestrians spill over close to or into the traffic lane or tram dynamic envelope. Although not examined in detail, this circumstance has been observed at tram stops at the intersection of Flinders Street and Swanston Street where passenger volumes are also high. Pedestrians standing in informal holding areas off the end of tram platforms may be a greater risk of being hit by a passing car or tram. See Figure Pedestrians regularly jay walk at signalised crossing associated with tram stops. Passengers rushing to catch a tram often cross against the lights, from both sides of the road. Passengers disembarking from a tram occasionally dash across in front of the tram before it departs the stop, or cross traffic lanes against the lights. These pedestrian behaviours may increase the risk of pedestrian vs tram or vs public vehicle crashes. 15 Pedestrians were observed stepping in front of stationary trams as passengers boarded and alighted. Some pedestrians also crossed the tracks while the tram was approaching the stop, assuming that the tram would be stopping and it was therefore safe to cross. There may be a risk of tram vs pedestrian crashes as trams leave platform stops, or if a tram passes through a stop without stopping. While this behaviour was also present at some safety zone stops it was more prevalent and potentially hazardous at platform stops on the downstream side of traffic signals. See Figure 32. Consider methods to protect pedestrians waiting in these areas, or to encourage pedestrians to wait within the fenced platform area. Automatic pedestrian gates linked to the pedestrian walk phase could be considered, or a separate traffic island in advance of the platform stop could be used to protect passengers standing in this area. Engineering treatments to prevent passengers jay walking at traffic signals are unlikely to be practical. Education and enforcement campaigns may be effective approaches to reduce the number of jay walking pedestrians. Consider methods of preventing pedestrians crossing in front of stationary trams. Engineering treatments and education campaigns could be considered. Offset platform design could be considered to relocate crossing points to behind trams, as used is the New Britain to Hartford Busway (Watkins et al. 2002). However, current operational approaches to preventing these type of crashes, such as drivers being aware and sounding a warning gong, appear to be reducing the risk. Important Important 43

47 Table 9: Kerb side stops - table of audit findings and recommendations Findings Recommendation Priority 1 The kerb side stop treatment relies on approaching Urgent drivers obeying the law and giving way to tram passengers when a tram is at the stop. The law is unlikely to be known by drivers from regional areas of Victoria, interstate or overseas and is not immediately apparent. While many trams have flashing yellow lights and some have small stop signs that fold out when doors open, these warning devices are not clearly visible or likely to attract the attention of drivers. Stopping behind a stationary tram is an uncommon event for most drivers and there are few visual reminders of the need to stop until after the tram has stopped and passengers are on the road. Vehicles approach trams stops at speeds of up to 60km/hr. Passengers alighting or boarding a tram at a kerb side stop are vulnerable to low, medium and high speed crashes with vehicles that fail to obey the give way to passengers law. Few passengers were observed looking carefully for approaching traffic before alighting from a tram. 2 Cyclists in Swanston Street where observed regularly passing stopped trams at stop 11. Some cyclists weaved in and out of passengers as they crossed the traffic lane. Other cyclists edged along the edge of the traffic lane to the traffic signal stop line while the tram was stopped. Cyclists are generally traveling very slowly past stop trams, but there is still potential for a crash and injury to both cyclist and passenger. Converting kerb side tram stops to an alternative style of stop where passengers can step directly from a protected platform or kerb is clearly an ideal method of reducing the risk of being struck by a passing vehicle, however it is unlikely to be practical across the entire tram network due to costs and the impacts on passing vehicles. Consider programs and projects to decrease both the incidence of drivers failing to stop behind a tram and the severity of any crashes. Education and enforcement programs are clearly a key aspect of any strategy to reduce the number of violations of the give way to tram passengers law. Engineering treatments such as improved warning signs, tram mounted devices or pavement markings could also be considered to remind and warn drivers. ITS based approaches, such as the illuminated Give Way to Passengers signs shown in Figure 20 that light up when a tram is present could be introduced at other kerb side stops to improve warning and remind drivers. Treatments such as speed humps, kerb extensions and similar engineering measures could also be considered to reduce vehicle speeds on the approaches to kerb side stop, thereby reducing the severity of any crashes. Consider additional education and enforcement programs targeted specifically at cyclists in Swanston Street. Noted 44

48 Table 9: continued: Kerb side stops - table of audit findings and recommendations Findings Recommendation Priority 3 Many kerb side stops are located on the approach to traffic signals, such Important as at stops 11 and 12. When approaching a stopped or slowing tram, public drivers may get conflicting messages if the traffic signal is displaying the green phase. The lack of a clear message to give way to passengers may increase the likelihood of drivers continuing past a stationary tram when passengers are alighting or boarding. 4 Many boarding passengers at stops 11 and 12 were observed stepping onto the road before the tram had arrived at the stop and traffic had come to a complete stand still. Some drivers were observed racing the tram to the stop to avoid being stuck behind the tram as passengers board and alight. There is a lack of clarity of whether the give way to passengers law applies only when a tram is stopped, or also when the tram is approaching the stop. There is significant potential for a collision between a boarding passenger who steps onto the road too soon and a passing vehicle. 5 Waiting passengers were sometimes observed stepping onto the traffic lane to see around parked vehicles and look for the next approaching tram. Passengers stepping onto the traffic lane are at risk of being struck by a passing vehicle. See Figure 34. Consider methods to prevent a green signal being shown when a tram is stopped on the approach to an intersection. Dynamic traffic signal control systems triggered by the presence of a tram may allow a red phase to be displayed while passengers are boarding and alighting, although this may require significant investments in sensing and signal technology or even active tram driver control to distinguish between trams passing straight through an intersection and trams stopping prior to the intersection. The use of presignals before a tram stop could also be considered as a method of halting public vehicles prior to the tram stop. Comments above in item number 1 apply. Public education campaigns could possibly be used to encourage passengers to wait on the kerb until the tram has stopped. Where possible ensure that kerb side parking does not block passenger s view of approaching trams. Continued implementation of accurate real time tram arrival information may assist in informing passengers of the proximity of the next tram. Important Noted 45

49 Table 9: continued: Kerb side stops - table of audit findings and recommendations Findings Recommendation Priority 6 Kerb side stops are not visually distinctive for approaching drivers. Some Noted stops are located at traffic lights, while other stops are midblock. There may or may not be a shelter on the kerb side. Roadside furniture at some stops only consists of a single pole mounted sign. Approaching drivers may not realise that there is a stop ahead and fail to slow down on the approach to a tram stop when behind a tram. 7 Not all passengers alighting from trams cross the traffic lane to the nearest kerb. Some passengers cross directly in front or around the back of the tram to reach the opposite side of the road. Passengers crossing in front of the tram they just disembarked are at risk of being hit as the tram departs. Passengers crossing behind the tram are at risk of being hit by trams or vehicles traveling in the opposite direction as the tram they disembarked blocks sight distance. See Figure There is no protection for passengers waiting at a kerb side tram stop from errant vehicles. Kerb side shelters and the waiting area are typically immediately behind the kerb, well within the clear zone. Any errant vehicle departing the carriageway close to a tram stop is likely to strike waiting passengers and the shelter structure Increase the visual impact of kerb side tram stops so that approaching drivers can clearly identify tram stops and the need to slow down and stop behind a tram. Public education and enforcement programs may be somewhat effective in reducing the number of passengers crossing incorrectly after alighting from a tram. Fencing between tram tracks at stops could be considered to prevent passengers crossing behind a tram, but is likely to pose a significant hazard to other vehicles. Where possible relocate passenger waiting areas well clear of the traffic lanes. Consider the need for safety barrier to redirect errant vehicles away from tram stop waiting areas, particularly at high volume stops or in high speed areas where the severity of such an event would be high. Important Important 46

50 Table 10: Easy access stop - table of audit findings and recommendations Findings Recommendation Priority See Figure 35. Noted 1 Like the kerb side stop, the easy access stop treatment relies on approaching drivers obeying the law and giving way to passengers boarding and alighting at a stop. While interstate or overseas drivers are no more likely to know the law at this type of stop than any other, the visual impact of the easy access stop is more likely to alert approaching drivers to the presence of the stop and passengers in the traffic lane ahead. The raised platform at the easy access stop forces drivers to reduce speed, acting as a speed hump and reducing speeds to 20-30km/hr. A risk of passengers being struck by a passing vehicle remains, but the likely severity of a crash is significantly lower at easy access stops than at kerb side stops. 2 Easy access stops have not yet been constructed in areas of high cyclist volumes. It is likely that the issues identified in item 2 of Table 9 would continue to apply if existing kerb side stops were converted to easy access stops 3 Easy access stops have not yet been constructed on the approaches to signalised intersections. It is likely that the issues identified in item 3 of Table 9 would continue to apply if existing kerb side stops were converted to easy access stops 4 Boarding passengers stepping out onto the road prior to the tram arriving at the stop will also occur at easy access stops, and may actually be more frequent as the edge of the traffic lane is less distinctive to pedestrians as there is no kerb or level difference between the footpath and traffic lane. However, with the raised platform on the approach to the stop public vehicle drivers will be less likely to race past the tram to avoid stopping for passengers. Vehicle approach speeds will be lower and there is a lower potential for a high speed collision between a boarding passenger who steps onto the road too soon and a passing vehicle. Noted Noted Noted 47

51 Table 10: continued: Easy access stop - table of audit findings and recommendations Findings Recommendation Priority 5 Parking bans and the alignment of the kerb on the approach Noted to stop 13 make it possible for a waiting passenger to see an approaching tram. There appears to be no need for passengers to step onto the traffic lane to see whether their tram is approaching at this stop, but at future implementations of easy access stops the site conditions may not be as favourable. Passengers stepping onto the traffic lane may be at risk of being struck by a passing vehicle. However, the lower vehicle speed at easy access stops would reduce the likelihood and severity of any crash between a passing vehicle and a passenger in the traffic lane. 6 The easy access stop is visually distinctive and stands out from the road environment. Approaching drivers are likely to notice the stop ahead and be alert to passengers. However, there are a large number of warning and traffic instruction signs, which are not likely to be read by drivers. 7 The raised traffic lane design may reduce the incidence of alighting passengers crossing in front of or behind the tram after disembarking. The level difference between the traffic lane and the tram tracks provides a clear boundary to passengers and may discourage passengers from stepping directly from the traffic lane onto the tram tracks. Where possible ensure that kerb side parking does not block passenger s view of approaching trams, as has been achieved at stop 13. Continued implementation of accurate real time tram arrival information may assist in informing passengers of the proximity of the next tram. Consider reducing the number of warning and traffic instruction signs on the approach to the easy access stops. A single sign warning drivers of a tram stop ahead may be more effective and reduce visual clutter. Noted Comment 48

52 Table 10: continued: Easy access stop - table of audit findings and recommendations Findings Recommendation Priority 8 There is no protection for passengers waiting at an easy access Noted tram stop from errant vehicles. Kerb side shelters and the waiting area are close to the traffic lane and within the clear zone. Any errant vehicle departing the carriageway close to a tram stop could strike waiting passengers and the shelter structure. However, the ramp on the approach to the tram stop is likely to reduce approach speeds and therefore the potential for errant vehicles and the severity of any crash. 9 The edge of the raised traffic lane is approximately 300mm higher than the tram tracks. An errant vehicle falling off the edge or a vehicle that drives up the ramp with one set of wheels on the tracks and other in the traffic lane could be at risk of rolling or crashing. There are many Keep Left bollards along the edge of the raised platform and a 60 metre long tram track separation kerb, which delineate the edge and encourage drivers to stay well clear. However, the tram track separation kerb is driveable and the separation between the edge of the traffic lane and the drop off is only approximately 300mm. There may be insufficient separation and protection to prevent errant vehicles falling off the edge of the raised traffic lane. However, the straight road alignment reduces the likelihood of errant vehicles. See Figure 36. Where possible relocate passenger waiting areas well clear of the traffic lane. Consider the need for safety barrier protection to redirect errant vehicles away from tram stop waiting areas, particularly at high volume stops or in high speed areas where the severity of such an event would be high. Consider increasing the separation between the edge of the traffic lane and the edge of the raised platform. Installation of semimountable or barrier kerb on the approach to the ramp may be more likely to prevent vehicles straying close to the edge of the platform, or entering with one set of wheels on the ramp and other on the tram tracks. Kerbing could also be used along the top of the platform to increase the separation between traffic and the edge of the platform, although gaps would be required to allow passengers to access the tram doors. Guardrail or similar treatments could also be considered to redirect errant vehicles, but the risk and severity of impacting the guardrail would have to be weighed against the likely lower risk and similar severity of a vehicle falling off the edge of platform. Existing delineation appears to be satisfactory, and drivers approaching at regular speeds are unlikely to fall off the raised platform. Important 49

53 5.2 Risk assessment Having received the audit report, the next major step in the road safety audit process is determining what should be done about the identified safety risks. This is done through a formal response to the audit report prepared by the project manager, in which each audit finding is accepted or rejected and changes rectify the safety issues are identified. Austroads (2009, p.31) shows a method of assessing the safety risks identified in a road safety audit, as a tool to decide which hazards will be treated within the limited project funds and other constraints. The risk is determined by assessing the likely frequency and severity of a crash due to the each identified safety issue, using Tables 11, 12 and 13 as a guide. Austroads (2009) identifies that this is not a scientific system and professional judgement should be used. However, this approach will allow a quantitative comparison of level road safety risk identified at each of the four different types of tram stops as part of the audit process. Table 11: How often is the problem likely to lead to a crash? from Austroads(2009) Frequency Frequent Probable Occasional Improbable Description Once or more per week Once or more per year Once every five or ten years Less often than once every ten years Table 12: What is the likely severity of the resulting crash type? from Austroads(2009) Severity Description Examples Catastrophic Likely multiple High-speed multi-vehicle crash on a freeway. deaths Car runs into crowded bus stop. Bus and petrol tanker collide. Collapse of a bridge or tunnel. Serious Likely death High or medium-speed vehicle/vehicle collision. or serious injury High or medium-speed collision with a fixed roadside object. Pedestrian or cyclist struck by a car. Minor Likely minor injury Some low-speed vehicle collisions. Cyclist falls from bicycle at low speed. Left turn rear-end crash in a slip lane. Limited Likely trivial injury Some low-speed vehicle collisions. or property damage only Pedestrian walks into object (no head injury). Car reverses into post. 50

54 Table 13: The resulting level of risk, from Austroads(2009) Frequent Probable Occasional Improbable Catastrophic Intolerable Intolerable Intolerable High Serious Intolerable Intolerable High Medium Minor Intolerable High Medium Low Limited High Medium Low Low In the following Tables 14, 15 and 16 this method of assessing road safety risk has been applied to each of the audit findings identified in Tables 7, 8, 9 and 10. Where relevant a range of crash severity has been identified to account for the range of circumstances encountered at different stops and under different traffic conditions. The crash frequency has been assessed based on likely crash frequency across the entire network of tram stops, rather than at a single stop as crash frequencies at a single location would all be occasional or improbable rather than producing a range of results. 51

55 Table 14: Safety zone stops - assessment of risk Item Description Frequency Severity Risk Comments 1 Gaps in railings Occasional Serious High Pedestrian struck by passing vehicle 2 Passenger crossing behind tram Occasional Serious High Pedestrian stuck by passing tram or vehicle 3 Passenger crossing from shelter Improbable Serious Medium Pedestrian stuck by passing vehicle 4 Narrow safety zone waiting area Probable Minor to Serious High to Intolerable Passenger struck by arriving or departing tram 5 Public vehicle crashes through safety Probable Minor to Serious High to intolerable Ranges from damage to safety zone stop and zone fencing minor injuries to car occupants, to larger scale impacts that could seriously injure waiting passengers 6 Passengers leaning against fence Occasional Minor Medium Passenger struck by mirror of passing vehicle 7 Safety zone prowl Probable Serious Intolerable Vehicle impacts prowl and possibly flips 8 Narrowing due to safety zone Probable Minor to Serious High to Intolerable Cyclist or vehicle side swiped due to lateral shift and narrowing 9 Hook turns Probable Minor to Serious High to Intolerable Vehicle undertaking hook turn T-boned by through vehicle from the same direction 10 Jay walking at signals at safety zones Probable Serious Intolerable Pedestrian struck by tram or vehicle 52

56 Table 15: Platform stops - assessment of risk Item Description Frequency Severity Risk Comments 1 Pedestrian movements in the vicinity of Occasional Minor to Serious Medium to High Pedestrian struck by vehicle crossing points 2 Uncontrolled crossing Probable Minor to Serious High to Intolerable Pedestrian struck by vehicle 3 Crossing behind trams Improbable Serious Medium 6 Non crash worthy fencing Occasional Minor to Serious Medium to High Lower vehicle speeds in the vicinity of platform stops due to reduction of carriageway width. Barrier kerb may also deflect vehicles somewhat away from collision with fence. 7 Passengers leaning against fence Improbable Minor Low Passenger leaning over fencing struck by passing vehicle mirror 8 Bollards on the approach side of stops Probable Minor to Serious High to Intolerable 9 Squeeze point for cyclists Probable Serious Intolerable Cyclist struck by vehicle in narrow traffic lane beside platform stop 10 Hook turns Probable Minor to Serious High to Intolerable Vehicle undertaking hook turn T-boned by through vehicle from the same direction 11 Pedestrians have priority over vehicles at zebra crossings, but then give way to trams 12 Unclear path for vehicles approaching stop on downstream side of intersection 13 Crowds of disembarked passengers in exposed position at end of stop 14 Jay walking at signals at platforms stops 15 Passengers crossing in front of stationary tram at uncontrolled departure side of stop Occasional Serious High Probable Minor High Improbable Catastrophic High Probable Minor to Serious High to Intolerable Pedestrian struck by tram or vehicle Probable Minor to Serious High to Intolerable Tram speed likely to be low 53

57 Table 16: Kerb side stops and Easy access stops - assessment of risk Item Description Frequency Severity Risk Comments Kerb side stops 1 Reliance on drivers obeying law Probable Serious Intolerable 2 Cyclists regularly flouting law Frequent Limited to minor High to Intolerable 3 Conflicting messages at signals Occasional Serious High 4 Passengers stepping onto road as tram Probable Serious Intolerable arrives 5 Passengers stepping onto road to look for tram Improbable Serious Medium 6 Stops not visually distinctive Occasional Serious High 7 Alighting passengers crossing directly Probable Serious Intolerable to opposite side of road 8 No protection for passengers waiting at stop Occasional Catastrophic Intolerable Easy access stops 1 Reliance on drivers obeying law Occasional Minor to Serious Medium to High 2 Cyclists regularly flouting law Frequent Limited to minor High to Intolerable Assume new stop in same location as existing kerb side stops where this is an issue 3 Conflicting messages at signals Occasional Minor to Serious Medium to High 4 Passengers stepping onto road as tram arrives 5 Passengers stepping onto road to look for tram Improbable Minor Low Improbable Minor to Serious Low to Medium 6 Visual clutter Improbable Minor Low 8 No protection for passengers waiting at Improbable Catastrophic High stop 9 Vehicle falls off raised platform edge Improbable Minor to Serious Low to Medium 54

58 Table 17: Identified number of risks for each crash type at each stop type Stop type Crash type Low Medium High Intolerable Safety zone Pedestrian vs public vehicle Pedestrian vs tram Public vehicle only Platform stop Pedestrian vs public vehicle Pedestrian vs tram Public vehicle only Kerb side Pedestrian vs public vehicle Pedestrian vs tram Public vehicle only Easy access stop Pedestrian vs public vehicle Pedestrian vs tram Public vehicle only Table 17 summarises risks identified in the road safety audits at each type of stop, categorised by the type of crash and the level of risk. Where a range of crash risk was specified in Tables 14, 15 and 16 the higher risk has been adopted. Table 17 allows some comparison of the overall levels of risk at each type of stop. There is a slightly lower level of risk from pedestrian vs public vehicle crashes at platform stops compared to safety zone stops. While pedestrian access points are more controlled at the platform stops, conflict between pedestrians and vehicles still remains an issue at crossing points without zebra crossings and due to jay walking at signalised crossings. The greater protection afforded to passengers waiting at a platform stop is clearly an improvement, but the potential hazard of being knocked down by a vehicle while getting to and from a stop does not appear to have been significantly reduced. There has been a slight worsening of risk for pedestrian vs tram and public vehicle only crashes, with the two new items (numbers 12 and 15) due to the location of the tram stop on the downstream side of an intersection. Crash risk for passengers at easy access stops is much lower than at kerb side stops. Due to the lower approach speeds of vehicles there is a much lower chance of a crash occurring and any crash islikely to be less severe. However, this reduction of risk for passengers has come at the cost of an increase in crash risk for motorists, primarily due to the hazard of the drop off between the traffic lane and tram tracks. On balance, it would seem that this is a reasonable trade off as passengers in the traffic lane are in a more exposed position and the risk to motorists is only low to medium. While upgrading an existing kerb side stop to an easy access stop using the current design will significantly improve road safety, the same can not be said for the current design of platform stops. In the next section a range of proposals will be discussed to improve safety at platform and easy access stops, by reducing the risks identified in the road safety audits and incorporating aspects of overseas stop designs previously discussed in Section 3. 55

59 6 Proposed changes to stop design This section outlines a range of proposed changes to the typical design of tram stops in Melbourne. The results of the road safety audits have been combined with options outlined in Section 3 and in (Austroads 2004) to produce a model concept design for each type of stop that would have improved road safety characteristics. 6.1 Platform stops A sketch of possible alterations to the typical platform stop design is shown in Appendix A The sketch has been based on the existing conditions at stop 5, but many elements should be applicable to other platform stops. The key alterations to the typical platform stop design are: 1. Offset the platforms so that the platform is further away from the downstream side of the intersection. This also has the effect of preventing passengers from stepping in front of a tram about to leave the stop. A pedestrian refuge or holding point between the two tram tracks would be required to prevent pedestrians being struck by a second tram approaching from the opposite direction. 2. Install zebra crossings or pedestrian signals at all access points to the platform stops. Isolated pedestrian signals could possibly be linked to signals at an adjacent intersection, and act as pre-signals. 3. Install a crash barrier to prevent errant vehicles crashing through into the stop. A crash worthy end treatment or frangible bollards should also be installed on the approach to the tram stop. Installing a crash barrier at a public transport stop is not a typical treatment, and may seem unnecessary were average approach speeds are low. However, it should be noted that a safety barrier is a typical treatment at long term roadworks sites where clearance between traffic lanes and workers is less than 3 metres and the speed limit is 40km/hr (State Government of Victoria 2004, p.46 figure 12). While there is a significant difference between a roadworks site and a tram stop, similar issues exist such as the proximity of traffic and the safety of people standing close to the road. The use of crash barriers under these conditions at roadworks sites could be considered an example of good practice that could equally be applied at trams stops to improve safety. 4. Install a traffic island or crash cushion to protect passengers waiting to cross at the signalised intersection end of the platform stop from errant vehicles. 5. Install Z crossings, bedsteads or manual swing gates at all locations where pedestrians have to give way to trams. 6. Install fencing between tram tracks and along the kerb to prevent pedestrians crossing tracks and traffic lanes in unsafe locations. 6.2 Easy access stops No improvement options that would improve safety at easy access stops were found in the literature review. This style of tram stop is unique to Melbourne and there is nothing in overseas experience to povide guidance for improving the safety of the design. There have been no crashes at either of the existing stops to highlight the actual types of crashes likely at this style of stop. However, the road safety audit results provide guidance for alterations to the standard easy access stop design to improve safety. 56

60 The key alterations to the typical easy access stop design are: 1. If the stop is located on the approach to a signalised intersection install pre-signals linked to the intersection controller in advance of the stop. A tram sensing system should be established that triggers the signals to show red when a tram is at the stop, to give a consistent message to approaching drivers. A similar approach could used at stops away from signals, with active flashing signs such as those shown in Figure 20 installed on Swanston Street in advance of a stop to warn drivers to give way to passengers when a tram is at the stop. 2. Indent parking or provided kerb extensions so that passengers waiting at the stop can clearly see trams approaching without the need to step onto the traffic lane as is currently the case at the stop at Victoria Avenue and Birdport Street. 3. Install barrier kerb or safety barrier along the edge of the ramp on the approach to raised platform to redirect errant vehicles away from the platform edge. These options are also shown in a sketch in Appendix A Figure 20: Swanston Street kerb side stop at Latrobe Street, note active flashing sign on left 57

61 7 Conclusions Maintaining and improving safety at tram stops in mixed traffic environments will be a key aspect of future operations across the Melbourne tram network. With more platform and easy access stops likely to be installed in the next few years to continue improving accessibility for passengers it is important to consider the safety implications of these new types of stops and how safety for passengers and road users can be enhanced. There is little research literature available to provide guidance on safety at tram stops in mixed traffic environments. However, while overseas experience is mostly in semi-exclusive and exclusive tram operating environments it may be possible to adapt some of the developments in safety systems and facilities to the mixed traffic environment typical of most of Melbourne tram network. Comparing the before and after crash rates at recently constructed platform and easy access tram stops was a key component of this project. Unfortunately, there is insufficient crash data available to state with any certainty whether crash rates have increased or decreased. A discontinuity in the VicRoads CrashStats data set has severely hampered crash data analysis based on police reports of casualty crashes, resulting in insufficient data to draw any conclusions. Revisiting this data set in a couple of years when more data is available may be appropriate. Internal Yarra Trams data has indicated that incident rates involving pedestrians have increased significantly following the installation of platform stops, but the reliability of this data is not considered sufficient to be confident in this result. There is little certainty that this data includes all crashes at the stops considered, or even a representative sample. The quality of the data set is not sufficient to allow detailed analysis, as the data is collected for operational logging purposes rather than crash reporting. Yarra Trams may wish to consider implementing a formal crash reporting and recording program similar to and interlinked with the VicRoads Crash Stats program to improve the opportunities for crash data analysis across the tram network. This project focused on comparing crash rates at each new stop from before and after the stop construction. There are certainly avenues for research that compares crash rates at the new stops with a control group of untreated safety zone and kerb side stops at other locations, but there has been insufficient time available during this project to undertake such a large amount of data analysis. However, any crash data analysis is likely to have issues due to the infrequent occurrence of crashes resulting in small sample sizes. Overseas research into risky behaviour at crossings highlights one method of collecting enough data to reliably assess the risk of crashes where the actual crash rates are low. Adapting this style of data collection to pedestrian behaviour at Melbourne s tram stops is a promising approach for future research. In this project road safety audits have been used to make up for the lack of data and provide an assessment of safety at typical safety zone, platform, kerb side and easy access stops. Perceived risks to be rated as either low, medium, high or intolerable based on the likely frequency of crashes and the resulting crash severity. During this project the frequency and severity were simply assessed by the auditors opinion, but there is an opportunity to use crash statistics to more accurately assess these elements in future research. The road safety audits have shown that platform stops reduce the risk to passengers waiting in the stop of being struck by a tram due to the wider waiting areas and clear separation provided by the platform edge. However, the platform stops have done little to reduce the hazards for passengers moving between the stop and footpath. Not all of the crossing points at the platform stops are controlled, and where pedestrians have to cross between stationary vehicles or in between gaps in traffic the safety situation is no better than at safety zones. At traffic signals the relocation of the tram stop to the downstream side of the intersection also creates some safety issues, with passengers waiting to cross at the pedestrian signals exposed to traffic leaving the intersection. Pedestrians crossing at the other end of platform may also 58

62 be at risk as trams depart the stop. In contrast, there is a greater improvement in safety for passengers where a kerb side tram stop has been converted into an easy access stop. The major improvement is due to the traffic calming effects of the ramp up to the raised platform, which significantly reduces the speeds of approaching vehicles and makes the stop more visually obvious to drivers. However, the gains in passenger safety have come at the expense of motorists due to the creation of a 300mm high drop off between the platform and the tram tracks. On balance, it would seem that this is a reasonable trade off as the risks to motorists of falling of the edge of the platform are much lower than the risks to passengers of being hit by a passing vehicle. Safety zone and kerb side stops will be gradually phased out from Melbourne s tram network as the operators continue to roll out trams and facilities that are compliant with the relevant disabled access requirements. The current platform and easy access stop designs are a key component of providing access to all members of the community and improving the usability and level of service to passengers. As well as providing improved access, there is great potential for the new types of stops to improve safety. A range of changes to the standard design have been recommended to improve the safety of passengers and other road users at platform and easy access stops. These include simple aspects such as pedestrian fencing and zebra crossings as well as more substantial items such as refuge islands, crash barriers and offset platforms. Some or all of these concepts could be considered for inclusion at future tram stop upgrades, within the constraints of the conditions at each individual site. With one of the largest tram networks in the world and a large proportion of mixed traffic alignment Melbourne is at the forefront of tram operations and stop design. The new platform and easy access style of stops represent innovative approaches to providing access to trams operating in mixed traffic conditions. The effects of these new types of stops on passengers, pedestrian movements, safety and operation of the tram network are yet to be fully investigated and this project provides only a small insight into the road safety of these new stops. As more platform and easy access stops are constructed and a larger volume of crash data becomes available it will be possible to quantitatively assess the safety impacts of these new stop types. While stop improvement programs are primarily aimed at improving access, there remains significant opportunity to research and improve the safety of passengers and other road users in the vicinity of tram stops. 59

63 8 References References Austroads (2004). Guide to Traffic Engineering Practice Part 4 - Treatment of Crash Locations, Austroads. PIRON Pty Ltd, Canberra. Austroads (2009). Guide to Road Safety Part 6: Road Safety Audit, Austroads. Third edition, Austroads Incoporated, Sydney. Baltes, M. R. (1998). Measuring motorist comprehension of Florida s school bus stop law and school bus signalization devices, Transportation Research Record 1640: pp Bu, F. & Chan, C.-Y. (2005). Pedestrian detection in transit bus application: Sensing technologies and safety solutions, 2005 IEEE Intelligent Vehicles Symposium Proceedings, pp Cliche, D. (2006). Trams - prioirty to moving people, not vehicles, Delivering Sustainable Transport - it s got legs, Proceedings of the 2006 AITPM National Conference, Australian Institute of Traffic Planning and Management Inc., pp Currie, G. & Smith, P. (2006). Innovative design for safe and accessible light rail or tram stops suitable for streetcar-style conditions, Transportation Research Record 1955: pp Desmond, K. & Natalizio, E. (2007). Bourke street tram stops at Russell Street, Queen Street and William Street: Functional design road safety audit, Technical report, GTA Consultants for Yarra Trams, Melbourne, Australia. Farran, J. I. (2006). Pedestrian and motor vehicle traffic control practices for light rail transit; innovations in new light rail transit system, Barcelona, Spain, Transportation Research Record 1955: pp Harris, P. (2006). Birdport Street/Merton Street, Albert Park - easy access tram stop stage 1 works: Intersection works and pedestrian crossings detailed design stage road safety audit report, Technical report, Road Safety Audits Pty Ltd for the City of Port Phillip, Melbourne, Australia. Huston, R., Cardimen, P. & Halperin, K. (1995). Design for the alleviation of transportation fatality risk by the implementation of new light rail transit, Journal of Rail and Rapid Transit, Institution of Mechanical Engineers 209: pp Korve, H. W., Farran, J. I. & Mansel, D. M. (1995). Integration of light rail transit into city streets, Seventh National Conference on Light Rail Transit, Vol. 1, Transportation Research Board, National Research Council, pp Korve, H. W., Farran, J. I. & Mansel, D. M. (1996). Integration of Light Rail Transit into City Streets Report 17, Transit Cooperative Research Program. Korve, H. W., Ogden, B. D., Siques, J. T., Mansel, D. M., Richards, H. A., Gilbert, S., Boni, E., Butchko, M., Stutts, J. C. & Hughes, R. G. (2001). TCRP Report 69 Light Rail Service: Pedestrian and Vehicular Safety, Transit Cooperative Research Program. Metlink Melbourne (2009). Tram 96 route map [online]. Available [Accessed May 2009]. 60

64 Price, I. R. (2007a). Birdport Street at Merton Street, Albert Park - new easy access tram stop post opening stage road safety audit report, Technical report, Road Safety Audits Pty Ltd for the City of Port Phillip, Melbourne, Australia. Price, I. R. (2007b). Flinders Street at Swanston Street, Melbourne - proposed central island platform concept design stage road safety audit report, Technical report, Road Safety Audits Pty Ltd for Yarra Trams, Melbourne, Australia. Purdy, P. (2007). Improving trams for better travel, Yarra Trams presentation to Engineers Australia. Reynolds, J. (2009). Personal communication with Ian Paul, Yarra Trams Traffic Control Manager, 7 april. Siques, J. T. (2001). Pedestrian warning and control devices, guidelines, and case studies, Transportation Research Record 1762: pp Siques, J. T. (2002). Effects of pedestrian treatments on risky pedestrian behaviour, Transportation Research Record 1793: pp State Government of Victoria (2004). Road management act 2004 worksite safety - traffic management code of practice, Victoria Government Gazette S 276. Wednesday 22 December, Victorian Government Printer, Victoria, Australia. Topp, H. H. (1998). Renaissance of trams in Germany - five case studies, Proceedings of the Institution of Mechanical Engineers 212(3): pp Unger, R., Eder, C., Mayr, J. & Wernig, J. (2002). Child pedestrian injuries at tram and bus stops, Injury International Journal of the Care of the Injured 33: pp VicRoads (2009). VicRoads CrashStats [online]. Last updated 20 February 2009 Available Watkins, K. E., Sorenson, T. & Garrick, N. (2002). Platform layout for the New Britain to Harford busway, Transportation Research Record 1791: pp Yarra Trams (2009a). Raised platform stops - list of completed and proposed stops, provided by Colin Tyrus, Communication Manager Yarra Trams via April Yarra Trams (2009b). Yarra trams incident database - 1 May 2004 to 30 April 2009, Internal data. Pedestrian Knockdown, Passenger Knockdown or Vehicle Collision with Infrastructure incidents only. Provided via 6 May Acknowledgments The author wishes to acknowledge the assistance of the following people during the course of this project: Colin Tyrus, Paul Matthews, Brian Vicars, John Clarke and Ian Paul from Yarra Trams, Professor Graham Currie, Chair of Public Transport at Monash University, and Dr Michael Reynolds. 61

65 A Concept designs for platform and easy access stops 62

66 63

67 64

68 B Site photographs 65

69 Figure 21: Elizabeth Street north side of Bourke Street, passengers crossing between safety zone railings Figure 22: Elizabeth Street south side of Bourke Street, passengers crossing behind a stopped tram 66

70 Figure 23: Elizabeth Street north side of Bourke Street, narrow safety zone provides limited space for waiting passengers 67

71 Figure 24: Elizabeth Street south side of Bourke Street, fencing, kerbing and shelters separating traffic lane and passenger waiting area 68

72 Figure 25: Elizabeth Street south of Bourke Street, passengers crossing at the approach side of the safety zone Figure 26: Collins Street west of Swanston Street, pedestrian short cutting through traffic lane 69

73 Figure 27: Collins Street west of Swanston Street, passengers crossing at uncontrolled crossing point Figure 28: Collins Street west of Swanston Street, looking west at approach to platform stop 70

74 Figure 29: Collins Street west of Exhibition Street, passengers crossing a random locations near zebra crossing Figure 30: Collins Street west of Exhibition Street, pedestrians crossing tram tracks near zebra crossing 71

75 Figure 31: Collins Street at Swanston Street, passengers standing at end of stop exposed to passing vehicles Figure 32: Collins Street west of Swanston Street, passengers crossing in front of a departing tram 72

76 Figure 33: Swanston Street north of Latrobe Street, passengers crossing between stopped trams Figure 34: Clarendon Street South Melbourne, waiting passenger stepping onto into the traffic lane to look for approaching trams 73

77 Figure 35: Easy access stop Figure 36: Platform drop off at easy access stop 74