A statistical forecast model for road surface friction. For the elaborate report on the Pulp Friction Pilot click here.
Background:
- Wintertime poor driving conditions occur when ice, snow or frost is present on the surface;
- Adverse conditions are caused by snowfall or freezing rain or when wet or damp surface cools below freezing temperatures;
- Accident rates increase dramatically under slippery conditions;
- Maintenance activities (e.g. salting) are performed to improve the driving conditions;
- EU FP7 ROADIDEA project for the development of new innovative services and techniques for the traffic and transport sectors.
For a live example of the Pulp Friction Map of Finland, click the picture below:
Friction:
- The grip between car tires and road surface;
- Varies between values 0.1 and 0.82;
- Large spatial and temporal variations;
- Snow and especially ice reduce friction dramatically;
- Condition of car tires have a significant impact – Issue not covered in this study.
Measuring Friction:
- Mechanical measurements, e.g. by making braking tests while driving and measuring the braking distance and time;
- Estimation by analyzing the reflection of the surface;
- Vaisala optical sensor DSC111 measuring, with additional measurements of the thickness of ice/snow/water on the surface;
- C. 100 DSC111 sensors along the road network in Finland (as of September 2009).
Friction modeling:
- Statistical regression model based on Vaisala DSC111 observations;
- Friction is represented as a function of temperature and ice / snow or water layer on the surface;
- Estimation of surface friction based on temperature and water / ice / snow storage;
- Minimum friction <-> 0.1 / Maximum friction <-> 0.82;
- Model is a component of FMI numerical road weather model;
- Model users are (i) road maintenance authorities monitoring road conditions, (ii) operational meteorologists making road weather forecasts;
- Production for drivers.
Model validation:
- Friction model was run at road weather stations with DSC111 sensors installed;
- Model was developed using observed friction, road temperature, and water/snow/ice, utilizing Utti, Anjala, Orivesi and Kuopio station data from winters 2007-2008 and 2008- 2009;
- Correlation between observed and modeled friction under icy and/or snowy conditions was 0.86 and under wet/damp conditions 0.98, when applied on independent Utti station data of 2009-2010;
- Results look quite promising, but the model produces too low friction values in many cases because of too much snow and ice in the model.
Future:
- Improve the wearing of snow and ice to reduce too big amounts of snow and ice on the surface in the model;
- Idea for improvements: Define station specific correlations at all computation points ( = road weather stations with DSC111 sensors );
- To investigate: Probabilistic friction/slipperiness forecast.
Acknowledgement: Finnish Transport Agency for providing observation data