Data and method risk and especially the severity Exponential Model - PDF document

32 nd ICTCT Conference in Warsaw, Poland Agenda Background Data Large-scale study on speed- Method calming effect from speed hump Results Niels Agerholm, Aalborg University, Denmark Summary Research Question Why is it of interest? Speed humps are used very many places in Which effect has the distance between urban areas for speed calming speed humps on the speed? Distance between speed humps is based on traditions rather than knowledge In case of: •Too short distance: No additional effect and expensive •Too long distance: Too high speeds Effect from hump No effect from hump Speed and risk Inappropriate speed increases accident Data and method risk and especially the severity Exponential Model Rune Elvik 2012 Slightly injured Seriously injured Fatalities

Data collection Selection of road segments Floating Car Data (FCD) from ITS • Comparable conditions of the road segments • Speed limit 50 km/h Platform • No other speed calming measures/effects as 431 persons got an OBU installed • Curves • Speed calming measures Region North Denmark • ”give way roads” • Further design May2012 – December 2014 1 Hz data 1.3 Billion positions Driven distance: 15 Million km Segment data Segment definition • Other speed-affecting variables • The distance between two speed humps is defined as a segment • Road type 1. Local suburban road • Segments from urban zone starts to speed humps 2. Distributor road between neibourhoods are removed 3. Through roads • Segments where other speed affecting are recorded • Road side characteristics are removed • Bicycle paths • Pavement • Bicycle lane Segment • Road width • Distance to physical buildings From single observations to road segments I From single observations to road segments II Unit for estimations: Map matching Segment 1 Segment 2 •1 observation per •Each position is connected to one road segment trip •85% percentile for each segment Expected area Actual area Road centre line

Overall results 570 speed humps 378 segments Transverse Modified sinus bump Overall results II Connection between distances and speed • Outlier removal (8) • Low data volume (12) • Few trips • Few individual vehicles Connection between road type, distances and speeding Modelling Modelling: Spee violation: � Univariate � Local road: 40 % � Distributor roads: analysis 87 % � Explaining � Through roads: variables: 99 % � Distance � Total: 91 % � Road type � Distance to buildings Distributor Distributor Local Through Road Local Road Through Road Road Road Road

Summary Acknowledement • 150 m between speed humps seems to We would like to thank: be too long •The Danish Innovation Fund via the • Above 190 m there seems to be no DiCyPS Center connection between speed and distance •The ITS Platform • Road type and distance to building • European Regional Development Fund seems to be most important • The North Denmark Region Thank you Lasse Høyrup Sørensen lhsn@cowi.com Pelle Rosenbeck Gøeg Aalborg University prg@civil.aau.dk Niels Agerholm Division of Transportation Engineering Dep. of Civil Engineering Aalborg University +45 61 78 04 55 na@civil.aau.dk