400 variables are predicting asphalt maintenance

With over 5000 km of national highways in the Netherlands, the Dutch Ministry of Infrastructure and Water Management (Rijkswaterstaat; RWS) is spending approximately 200 million euro a year on maintenance of asphalt pavements. To decrease these costs, and thereby decrease the amount of tax money needed, RWS is starting to use big data techniques to predict “just in time” maintenance.

undefined
Beeldbank

RWS uses predictions of the asphalt lifetime to estimate when and where road maintenance will be needed. The current practice of programming road maintenance is based on a course prediction, in combination with many operational parameters. This has resulted in maintenance frequently being done either too early (before the materials are fully used) or too late, as an emergency repair of damaged roads, which can lead to extra costs and environmental and traffic burden.

The main failure mechanism of asphalt on Dutch highways is raveling. Raveling data is collected with a Laser Crack Measurement System. In current practice, the data of a road segment of 100 metres is typically summarised in nine parameters. The prediction of asphalt lifetime based on these parameters in consecutive years is correct one third of the time. Using the available data in a more detailed manner, e.g. considering data of shorter road segments and combining and analysing the measured data before it has been summarised, the prediction consistency in consecutive years has doubled to two third of the time. This is a major improvement!

 

undefined

The effect of using shorter road segments can be illustrated as follows: Based on the available data, a segment of asphalt currently gets a quality score between 1 and 5. However, there can be large differences in quality within a segment of 100m. Trees, for example, decrease the quality of the road underneath. Meanwhile transitions between types of asphalt, for instance coinciding with a bridge location or with the end of a batch of asphalt production, also have an influence. In a span of 100m, you can then have three different asphalt quality levels, scoring together an average of 3 (meaning no maintenance is needed), while in reality there will be parts scoring a 1 and parts scoring a 5. This can be overcome using shorter road segments.

To be able to have a clear overview of all this data, RWS will develop a dashboard which will visualise the quality of all asphalt of the national highways in the Netherlands. The creation of this dashboard is part of the European project BE-GOOD which stimulates the use of open data.

Improving the accuracy of asphalt lifetime prediction enables better maintenance planning. The planning is based on more aspects than asphalt quality alone, such as minimizing traffic hindrance, combining asphalt maintenance with other road maintenance, budget planning, contract details, and guarantee issues. At present, these other aspects sometimes dominate the maintenance planning, because the prediction of residual asphalt lifetime is of insufficient quality. With big data analysis a turnaround appears to be feasible.

As a result, premature maintenance may be significantly decreased, thus saving on costs and on environmental impact due to CO2 emissions and (raw) material usage. Furthermore, timely maintenance avoids ad-hoc and unforeseen maintenance which often lead to traffic congestion. And a very important advantage is the increased road safety for all road users due to optimised asphalt conditions.

For more information about this project, watch the video below:

 

 

Share this

Tweet Share