Pavement Surface Life Predictor

Ensuring our roads are safe for New Zealanders is vital to taking care of our communities. Waka Kotahi NZ Transport Agency asked our team to analyze how, where, and why we maintain our road network to understand and create any efficiencies in the process and ultimately make our roads safer.

Initially, we focused on understanding the life expectancy of surfaces along the New Zealand state highway network.  Our data scientists, leveraging large historic and live datasets, were able to use a Kaplan-Meier survival model to generate insights into which chip seals would provide the best outcomes for the network.

The analysis and models were subsequently refined to identify the impact of skid-driven resurfacing on expected chipseal surface lives. This allowed us to create a predictive longevity value along the state highway network, based on surface information, helping us predict and identify areas along the network which may degrade faster than others and need early preventative maintenance. We also built a regression model to assess the independent effects of over 60 candidate risk-factors, after adjustment for the other modelled control variables.

The result? A model that could reasonably distinguish between surfaces at higher risk of failure than lower risk surfaces, correctly identifying which one would fail first 70% of the time.