Recycled concrete aggregates optimized by machine learning
The availability of locally-sourced natural concrete aggregates in Switzerland is constantly diminishing as more deposits are covered by habitation and as excavation in riparian zones continues to be highly regulated. At the same time, construction and demolition waste (CDW), which contains large quantities of granular minerals from concrete, bricks, and other crushed ceramic materials, is the biggest waste flow in the country.
There is great ecological and economic potential in the production of recycled concrete (RC) aggregates out of mineral CDW. However, current approaches to the development of RC recipes are empirical and based on default methods, making the process inefficient and expensive. This is partly due to the fact that the impact of a recycled aggregate on RC performance is a multifactorial, complex phenomenon. The results of traditional empirical RC development are highly specific to the CDW stock in use, and cannot in principle be transposed to other CDW sources.
The objective of project BAROMaL is to contribute to the development of a new, adaptive RC recipe formulation tool, one that is capable of efficiently handling the variability of recycled fine aggregates. Machine learning techniques will be applied to images of recycled aggregates to enable the prediction of the physical performance of new types of RC. These machine learning techniques will also make it easier to estimate the impact that modifications to the recipe can have on the product’s performance in mechanical, economic and environmental terms.
Project BAROMaL thus completes the methodological exploration that began with project ORCADEMO.