MYTH-BUSTING IN MELBOURNE Left-turn Vehicle vs. Pedestrian Crashes At Signalised Intersections Presentation by Mark O Brien AITPM Thursday 21 September 2017
Assumed Common Wisdom Our study commenced over an argument about whether to include left-turn slip lanes at major signalised intersections in new residential subdivisions Two conflicting points of common wisdom were put forward: Planners, urbanists and others: VicRoads: everyone knows that slip lanes are unsafe for pedestrians and so they should be removed wherever possible and no new ones should be installed slip lanes significantly improve operations and safety and should be maintained and included in new or upgraded intersections 2
Seeking Evidence Both sides asked the other to present evidence to support their contention Evidence against slip lane safety : Finding: Both sides genuinely believed their views to be soundly based in fact and only noted the lack of evidence when challenged was limited to anecdotal reports, particularly relating to some pedestrians stated perception that slip lanes feel unsafe Evidence for slip lane safety : the most relevant safety data available for slip lanes related to vehicle vs. vehicle crashes but data on slip lane pedestrian safety was very limited Resolving the policy dispute was considered unlikely without data as the opinion-based positions were too entrenched Consequently, we were engaged to examine the left-turning vehicle vs. pedestrian crash history at signalised intersections in Melbourne 3
Study Outline The study included: Literature review Road rule and policy review Crash and left-turn treatment data assessment Stakeholder liaison VicRoads staff, pedestrian, bicycle and visionimpaired persons peak bodies Left-turn treatment design and human factors review Note: The original study is 288 pages to fully present its findings and methods is outside of the scope of this presentation. It can be made available to those who are particularly interested in the topic and/or methods used 4
Literature Review The literature available was relatively limited but some key highlights included: German study : highlighted the view limitations for truck drivers performing left-turns at signals without slip lanes (i.e. pedestrians and cyclists are located in a blind spot at significant times during the movement) truck specific crash data showed conflict types very similar to Melbourne data involving all vehicle classes USA studies difficulty for vision-impaired pedestrians in correctly selecting safe gaps at slip lanes applied to higher-speed free flow slip lanes without pedestrian priority for crossings (i.e. different to the Melbourne situation) 5
Road Rule & Policy Review A key finding from the road rule and policy review was : That VicRoads had a long-standing policy of only allowing zebra crossings on slip lanes where sufficient pedestrian demand could be demonstrated to maintain the specialness of zebra crossings to avoid drivers becoming complacent about zebra crossings at mid-block locations HOWEVER Many drivers we spoke with assumed that the different crossing layouts had different meanings and that as per mid-block crossings : drivers yield to pedestrians on a zebra crossing pedestrians yield to vehicles on an unmarked crossing UNFORTUNATELY Finding: The assumption underlying the zebra crossing policy may be creating a significant safety risk rather than solving one According to the Victorian Road Rules drivers must yield on a slip lane in both cases 6
Crash and Treatment Study Area Melbourne Metropolitan Area (pop. 4 million +) The Metropolitan Melbourne boundary was adopted: It exactly matches the only common geographic boundary in the signals & Crashstats databases maintained by VicRoads 7
Crash Rate Based on Treatment Proportion The treatment proportion method of comparison involved the following steps: 1. Categorise all left-turn treatments at traffic signals in study area 2. Identify crashes involving left-turn vehicles and pedestrians at signalised intersections 3. Compare frequency of the various left-turn treatments in the network against their share of the recorded crashes 4. Examine crash characteristics for each type of treatment Why not use an exposure-based comparison of treatment crash characteristics? Insufficient volume data and low crash frequency per site Finding: Not following the standard crash rate methodology offers an opportunity for opinionbased policy to withstand challenge because the study results can be dismissed as nonstandard but these results are the best evidence available. They should not be considered definitive but are strongly suggestive of particular courses of action 8
Study Area Traffic Signals Signals database 2010 snapshot of network based on aerial photography and selected site inspections 2,141 signalised intersections contained 6,922 approaches with left-turns The main categories of left-turn treatment were shared lanes, exclusive lanes and slip lanes although each had variations 9
Study Area Traffic Signals Slip lanes were further sub-categorised for various features but this introduced/exacerbated sample size issues so this presentation focusses on the aggregated group results 10
Stakeholder Input Aside from VicRoads staff, stakeholder input was sought from: Vision Australia, the Pedestrian Council of Australia and Bicycle Victoria All 3 groups identified: the same preference ranking for left-turn pedestrian crossing treatments with their current design features concerns with the safety of unsignalised slip lane crossings a preference for normal stand-up lane treatments possible design improvements for various left-turn treatments 11
Stakeholder Input Particular points made: Vision Australia indicated that many vision-impaired users are specifically trained to avoid unsignalised slip lanes at certain locations because : drivers tend to look to the right for a gap in traffic rather than forwards towards the pedestrian crossing at (particularly unsignalised) slip lanes of potential difficulties for vision impaired pedestrians hearing whether a nearby vehicle is (a) on the slip lane or (b) slowing to yield drivers only have to yield to pedestrians after they have stepped onto the roadway at zebra crossings a more risky act for pedestrians who cannot see or clearly hear if a vehicle has detected them and is about to yield Finding: this policy recommendation by a stakeholder could be artificially improving the safety performance of unsignalised slip lanes by relocating vulnerable users to other crossing locations 12
Stakeholder Input Particular points made: Pedestrian Council of Australia generally deferred to vulnerable user peak bodies (e.g. Vision Australia) on safety matters noted that drivers seemed not to be aware that they need to yield to pedestrians at both zebra and unmarked crossings on slip lanes Bicycle Victoria were generally opposed to slip lanes as cyclists using the road (a different discussion for another forum but with similar rules of thumb to follow) expressed dislike for zebra crossings that legally requires cyclists using them to dismount 13
Available Crash Data VicRoads Crashstats Database 5-year period data (1 Jan. 2004 31 Dec. 2008) Used the more detailed restricted access Crashstats data which includes police officer descriptions BUT Crash categorisation & reporting issues required manual record inspection of each potential qualifying crash & an expanded search was required e.g. identifiable pedestrian crashes at slip lanes returned: 6 crashes - from a direct database search of the slip lane category 42 crashes from a detailed wider database search Finding: Too many studies would have stopped at the direct sub-category search carefully check the data and assume it may be incorrect or incomplete 14
Available Crash Data VicRoads Crashstats Database The expanded search included: traffic signals pedestrian signals/crossings STOP or GIVE WAY intersections (including slip lanes) The broad search found 1748 crashes involving collisions between vehicles and pedestrians identified with the following crash types: 281 left-turn (16.1 %); 862 through (49.3 %); 520 right-turn (29.7 %); 6 reversing (0.3 %); 1 U-turn (0.1 %); 34 trams (1.9 %); and Finding: Through and right-turn vehicle movement crashes are far more common (and probably have more serious consequences) than left-turn vs. pedestrian crashes 44 other (including 2 possible left-turns and 29 unspecified vehicle movements) (2.5 %). 15
Available Crash Data Of the 281 left-turn vs. pedestrian crashes identified in the study area : 195 left-turn vs pedestrian crashes were recorded at signalised intersections over 5 years These crashes resulted in 3 fatalities, 70 serious injury & 122 other injury Finding: the severity of the crashes involved in left-turn vs. pedestrian crashes is relatively low particularly when compared to higher speed red light running incidents 16
Results Crashes & Injuries All Treatments Crashes by treatment: Shared lanes much higher share of crashes than treatments Exclusive lanes slightly lower share of crashes than treatments Slip lanes much lower share of crashes than treatments Injuries & fatalities by treatment: Were very similar to crash proportions for each treatment 17
Clear Patterns of Major Crash Conflict Types The crash assessment classified the pedestrian/vehicle conflict types for each incident Each treatment type had dominant conflict types The clear patterns strongly suggest potential design and operations improvements Some patterns matched expectations from literature review and stakeholder consultations but others were unexpected Zebra crossing policy impacts appear to be obvious for slip lane crashes when viewed by sub-category 18
Results Crash Types for Non-Slip Lanes Para. B Shared & exclusive lanes crashes dominated by Para. B conflicts with most of the remainder being Para. F conflicts Exclusive lanes much lower share of the Para. F conflicts than the shared lanes probably due to clearer view when approaching leftturning vehicle from the front Para. F 19
Results Crash Types for All Slip Lanes All slip lanes crashes are dominated by Perp. L conflicts Perp. L Para. B Perp. R Crash narratives, user group feedback & literature review suggest that Perp. L crash dominance at slip lanes is likely due to: the vehicle driver looking right to find a gap in traffic (i.e. fails to observe a pedestrian approaching from the left) Finding: the amount of jaywalking Para B. crashes was unexpected and may lead to pedestrian fencing to block this pedestrian movement without blocking vehicle sightlines 20
Policy Impacts on Crash & Slip Lane Type Results Unmarked crossings much lower share of Perp. L crashes compared to their share of slip lanes in the network Zebra crossings much higher crash proportion than treatment frequency due to high pedestrian demands 21
Policy Impacts on Crash & Slip Lane Type Results These results: Are counter-intuitive as zebra crossings tend to be more prominent and their meaning well understood by road users Are skewed by VicRoads zebra crossing policy Can be adjusted to overcome skewed data by combining the zebra & unmarked crossing results Combined result 87% slip lanes have 88% of Perp. L crashes Each slip lane type then has Perp. L crashes essentially in proportion to treatments in the network Finding: potential policy bias on results must be understood to avoid incorrect conclusions in this case that zebra crossings should be removed and replaced with unmarked crossings 22
Recommendations Provision of left-turn facilities: Left-turn vs. pedestrian crashes appear not to be the most pressing pedestrian crash problems in the network with all treatment options appearing to work reasonably well from a safety viewpoint Slip lanes appear to be the most preferable treatment from a safety data perspective Slip lane design improvements are needed to address stakeholder concerns, literature review findings and distinctive crash type patterns in the data which all tend to point in similar directions Signalised crossings are likely to be preferable for vulnerable user groups but the crash data sample size was insufficient to recommend this crossing type for more widespread application given the costs involved Exclusive stand-up lanes are likely to be preferable to shared stand-up lanes where slip lanes are not achievable 23
Useful Rules of Thumb The left turn study highlighted the following useful rules of thumb : Make decisions based on data over opinion wherever possible and acknowledge the limitations of the data State your assumptions you cannot confirm or challenge an idea without knowing its basis Check the data it may reveal methodology flaws or important extra information Ask key stakeholders for input and review they may know the intricacies of your subject matter better than you do or offer a very different perspective on what you think you know Check if policy choices are impacting your results these could be your policies or someone else s 24