Effective Road Safety Management Using Network-wide Historical Speed Data – Commercial Speed Data As Big Data Source to Improve Road Safety Intelligence

Effective Road Safety Management Using Network-wide Historical Speed Data – Commercial Speed Data As Big Data Source to Improve Road Safety Intelligence


Timo Hoffmann, PTV Group, Karin Hitscherich, PTV Group


Commercial network-wide speed data is a valuable big data source for road safety management. It can improve road safety intelligence in several ways. Five use cases are described in this paper.


Commercial map data providers are offering custom speed data in selected countries. As opposed to speed data coming from measurement devices, this data was gathered by analyzing a large number of vehicle GPS trajectories. A major advantage of this big data approach is the general availability of the data on the whole road network (some limitations apply) instead of just punctual information or for predefined routes like when travel time data is collected by Bluetooth device tracking. Another advantage is the effortless and fast availability of the data.
Some negative aspects of this kind of data exist and are addressed in the full presentation.
The data available allows for a custom definition of date period (data is generally available from 2008 depending on region/country in the case of TomTom data), weekly hour groups within this period (e.g. weekday morning peak, Friday & Saturday night,...) and for freely definable routes within the road network. For this kind of data, also the per segment speed distribution in 5-percentile steps is provided, allowing the analysis of V85 values and more.
We have tested this data (“Custom Travel Time” data by TomTom) and identified the following use cases and benefits for the road safety researcher and practitioner:

1. Comparison of design speeds with actual driving speeds
Roads are generally designed to safely handle a predefined posted speed and in turn maximum expected speed. In most cases a corresponding speed limit is explicitly posted or implicitly given, but this is not always the case. Also, over time, changes in maximum speed limits can occur or the speeding likelihood may change. In turn, a mismatch between initial design speed and the speeds generally driven might arise. Speed data can now be used to evaluate network elements where such discrepancies occur. The introduction of a new speed limit regime or the upgrading of those parts of the network where the speeding behavior does not match the design speed can be focus of a “Systemic Safety Approach” to improve the inherent safety performance of this part of the network.

2. Assess the effect of road infrastructure changes or speed enforcement tactics on the driving speeds at the site and in its vicinity
For many crash black spot treatments, speed reduction is a main goal. However it is hard to evaluate the long term effect of any measure at the treatment site and its surroundings. Speed data can now be used to get an indication of the historical speed distribution as well as the changes in speeds after implementation of the treatment. This is – in addition to what is possible with traditional single location speed camera measurements – available also in adjacent parts of the road network.

3. Analyze the general speeding behavior at crash sites where speed is a possible contributing factor
If speeding was a contributing factor of a crash is not always easy to assess. In turn, the quality of this attribute in crash data – if available – is mediocre at best. Commercial speed data can help to identify if speeding is generally a problem at a high risk section and at what times.

4. Check necessity for posting a general speed limit as opposed to time dependent speed limit changes
Building on top of the previous use case, if knowledge is available on when critical speeding occurs, the road administration has the option to consider temporal speed limits (e.g. only at night times), which might be in line with local or regional policy to not impose “unnecessary” speed limitations while at the same time consider road safety aspects.

5. Calibration of microsimulations to better model the real driving behavior
The use of microsimulation software to assess safety levels is a growing trend among researchers and consultancies specialized in road safety evaluations. The most common toolset for this (as observed at Road Safety and Simulation conference 2013 in Rome or Transportation Research Board Annual Meeting 2014 in Washington) is PTV Vissim as microsimulation software and SSAM as analytical engine for surrogate measures to assess safety. While PTV Vissim offers a default driving behavior and speed distribution functions, the quality of the model (i.e. how well the model simulates "real" driving behavior) can be increased by using local speed distributions from the commercially available speed data as input. Also, total travel times can be compared between the model output and the measured data to calibrate iteratively.


Association for European Transport