Neural networks implementation for the environmental optimisation of the recycled concrete aggregate inclusion in warm mix asphalt

Abstract

Regarding the traditional Hot Mix Asphalt (HMA), Warm Mix Asphalt (WMA) with Recycled Concrete Aggregate (RCA) contents (WMA-RCA) requires lower production temperatures and diminishes the consumption of natural aggregates (NAs). Nonetheless, these environmental benefits may be counteracted by the higher optimal asphalt binder demanded by the WMA-RCAs. In this regard, this research develops a computational model to optimize the WMA-RCA design. In order to build a sufficiently accurate and adaptable model, it was decided to employ Artificial Neural Networks (ANNs). The ANN implementation was based on the postulates of the statistical learning theory, i.e., preferring to generate learning through low-complexity models. Also, a representative case study of the northern region of Colombia was assessed. In this scenario, the optimal coarse RCA content was 10%, and the sustainability savings were maintained up to an RCA's hauling distance of 200 km.

KEYWORDS:

• Artificial neural networks
• deep neural networks
• life cycle assessment
• recycled concrete aggregate
• sustainability
• warm mix asphalt

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study is framed in the ‘Research Project 745/2016, Contract 037-2017, No. 1215-745-59105’, funded by the Administrative Department of Science, Technology, and Innovation (COLCIENCIAS) and the Universidad del Norte.