A new study highlights the power of machine learning (ML) in tackling a major challenge in road maintenance: moisture damage in asphalt pavements.
Study: Prediction of moisture susceptibility of asphalt mixtures containing RAP materials using machine learning algorithms. Image Credit: ABCDstock/Shutterstock.com
Published in the International Journal of Pavement Engineering, the research explores how ML algorithms can predict the moisture susceptibility of asphalt mixtures containing reclaimed asphalt pavement (RAP) materials. The study evaluated four ML techniques—decision tree (M5P), random forest (RF), gradient boosting (GB), and extreme gradient boosting (XGBoost)—to determine their effectiveness in optimizing pavement durability.
Background
Artificial intelligence (AI) is transforming infrastructure development, from predicting potholes to designing more durable concrete. Civil engineering researchers are increasingly leveraging AI to create resilient and sustainable infrastructure by optimizing the use of recycled materials, industrial by-products, and alternative components. This not only improves material efficiency but also minimizes costs related to labor, energy, and long-term maintenance.
In line with this trend, the study focused on the ability of ML algorithms— a subset of AI— to predict how well asphalt mixtures with RAP materials withstand moisture, a key factor in pavement durability.
Methods
Moisture infiltration weakens asphalt by breaking the bonds between its components, increasing the likelihood of cracks, potholes, and other forms of pavement distress. This issue is particularly significant in cold and wet climates where moisture-induced damage is more prevalent.
As such, the study assessed four ML algorithms to predict moisture damage in asphalt mixtures containing RAP. A comprehensive database, incorporating factors such as mixture properties, binder characteristics, environmental conditions, and recycled material composition, was used to determine which variables most influence moisture resistance.
Results and Discussion
The findings demonstrated that ML models can accurately predict moisture damage in asphalt mixtures, helping engineers optimize material selection and assess the probability of pavement failure over its lifespan.
Among the tested models, XGBoost showed the highest predictive accuracy, with a correlation coefficient of 0.96 for training data and 0.75 for testing data. These results highlight the potential for ML-based predictive models to improve asphalt pavement durability. Additionally, comparisons of actual and predicted tensile strength ratio (TSR) values, along with residual analysis, confirmed the robustness and reliability of these models.
The study’s implications are significant, particularly in light of the United States' road maintenance challenges. In 2021, state and local governments spent over $206 billion on road maintenance, while in 2023, the Department of Transportation identified a $1 trillion backlog in bridge and road repairs. By optimizing asphalt mixtures, ML could reduce these costs and extend road lifespan.
Traditionally, identifying the ideal combination of RAP and other materials for specific weather conditions requires extensive and costly laboratory testing. AI-driven models offer a faster, more cost-effective alternative, making infrastructure development both more economical and environmentally friendly.
Conclusion
This research underscores the potential of ML models to enhance the reliability and efficiency of moisture susceptibility predictions, ultimately contributing to the development of more durable asphalt pavements. The findings can aid engineers, transportation departments, and private-sector stakeholders in adopting more sustainable and cost-effective design approaches.
Beyond pavement durability, AI and ML have broader applications in infrastructure management, including optimizing road and bridge designs, improving waste management, and monitoring railroad integrity. AI also plays a crucial role in risk management and disaster resilience. For instance, during emergencies, AI can help plan optimized evacuation routes tailored to specific conditions, ensuring greater safety and efficiency.
By integrating AI into infrastructure planning, the industry can move toward smarter, more sustainable solutions that balance cost, durability, and environmental impact.
Journal Reference
Afshin, A. & Behnood, A. (2024). Prediction of moisture susceptibility of asphalt mixtures containing RAP materials using machine learning algorithms. International Journal of Pavement Engineering, 25(1). DOI: 10.1080/10298436.2024.2431610, https://www.tandfonline.com/doi/full/10.1080/10298436.2024.2431610
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.