After the interview with Data Scientist Researcher Christoph Völker on the Sequential Learning App for Materials Discovery, we explore the further development of this application in Reincarnate. As a European project, Reincarnate seeks alternative approaches to designing and manufacturing building materials to achieve a green circular economy. The project’s ultimate goal is to develop the technical and social means to give new opportunities to buildings, construction products and materials — thus maximizing their life cycle and determining if they are suitable for reuse. 

To achieve desired material properties, such as a minimum compressive strength, conventional material development workflows typically commence with a prescriptive-based material composition formulation, which is then validated through laboratory testing. However, predicting material compositions becomes challenging for various reasons, rendering a closed-form solution unattainable.

Dr. Völker and his colleagues introduce an innovative approach in this application, revolutionizing the material development process by employing AI to screen vast quantities of compositions and predict ideal materials. Consequently, this method has the potential to uncover novel materials that satisfy the desired specifications, which might otherwise remain undetected using traditional design workflows.

The innovation behind SLAMD: scaling logic and optimal solutions 

Traditionally, engineers only developed a couple of well-working formulations because they involved a lot of manual labour. SLAMD scales this up through built-in domain logic and helps to uncover solutions that may not be obvious but are optimal concerning various properties. 

SLAMD digital lab twin helps to create thousands of possible formulations and augment information such as CO2 footprint using detailed data from the constituents. This data can then later be used for AI optimization.

An ecologically informed innovation: calculating CO2 footprint

The ecologically informed function of the App comes from the fact that the technology does not only seek the best mechanical properties (such as classical lab-based research) but considers the ecological footprint at the same time. 

For instance, the CO2 footprint can be calculated ahead. The App does this for hundreds and thousands of possible formulations. In this way, the Eco-Info acts like a filter that guides decisions toward the most sustainable solutions from the start. The system works surprisingly well, and because of that, we can find high-quality materials that maintain an efficient footprint.

The future of data-driven approaches in building materials

Based on the study by Dr Völker and his team, future research should focus on developing data-driven approaches to building materials optimization that aim to automate the process and implement the approach at larger scales. There have been limited automation experiments in the building material sector, and we need to optimize or discover new materials. 

This month Dr Völker and his colleague Sabine Kruschwitz have attended the RILEM Association Spring Convention 23 in Rabat, Morocco, to present their work on AI-driven materials design using the SLAMD App and their paper on how machine learning could help to find ideal sustainable building materials. All in all, spreading research findings focus on sustainable building materials through AI-driven design frameworks.

Construction Wastes: Types, Causes, and Recycling Strategies - The Constructor

Do you want to learn more about Reincarnate or even contribute to our mission? 

Subscribe to our newsletter and follow us on LinkedIn and Twitter, so you don’t miss a thing!