Abstract:
It is known that road geometric features significantly contributes to the variation in vehicle speed. According to conventional Highway Geometric design procedure there is no accurate method to predict actual vehicle speed with different combinations of geometric elements such as horizontal alignment and vertical alignment etc. This study explores a methodology to evaluate actual vehicle speed variation mainly considering geometry.
In this paper after classification of all the design elements, a new predictor called “curvature Index” (the degree of angle variation per unit length) is introduced to represent actual horizontal alignment variation of a road segment including number of bends and nature of bends. The curvature index measures analyzed were: bend density (number of bends per km); cumulative angle (degrees per km); mean angle (degrees); and standard deviation of angles. The research confirmed that driver’s speed choices are more strongly related to curvature Index than curve design speed, and to the approach speed environment. Similar measures were carried out for vertical alignment variation and new predictor called “Elevation Index” is introduced to represent different combination of vertical alignments. International Roughness Index (IRI) and Road width selected as other predictors to obtain the correlation between Geometry and speed.
The curvature index, elevation index and actual vehicle speed are estimated using the database of GPS (Global Positioning System) receivers. Data was collected at selected road segments in Sri Lanka under less traffic condition. A methodology developed to increase the reliability and accuracy of GPS data. Simple linear regression analysis is carried out using SPSS and MINI Tab software to develop actual vehicle speed model together with ArcGIS. ArcGIS (Geographical Information System) provides a good platform to graphical analysis of data and integrate with GPS data.
This study conforms positive correlation between actual speed and combination of Geometric elements. This model could be combined with other social environmental factors (e.g.: land use) and effectively use as speed prediction model or as a design tool.
