TY - JOUR A1 - Völker, Christoph A1 - Moreno Torres, Benjami A1 - Zia, Ghezal Ahmad Jan A1 - Rug, Tehseen A1 - Firdous, Rafia A1 - Böhmer, Felix A1 - Stephan, Dietmar A1 - Kruschwitz, Sabine T1 - Presenting SLAMD – A Sequential Learning Based Software for the Inverse Design of Sustainable Cementitious Materials N2 - In recent decades, the number of components in concrete has grown, particularly in formulations aimed at reducing carbon footprints. Innovations include diverse binders, supplementary cementitious materials, activators, concrete admixtures, and recycled aggregates. These developments target not only the enhancement of material properties but also the mitigation of the ecological and economic impacts of concrete — the most extensively used material by humankind. However, these advancements also introduce a greater variability in the composition of raw materials. The material’s behavior is significantly influenced by its nanoscale properties, which can pose challenges in accurate characterization. Consequently, there’s an increasing need for experimental tuning of formulations. This is accompanied by a more inconsistent composition of raw materials, which makes an experimental tuning of formulations more and more necessary. However, the increased complexity in composition presents a challenge in finding the ideal formulation through trial and error. Inverse design (ID) techniques offer a solution to this challenge by allowing for a comprehensive search of the entire design space to create new and improved concrete formulations. In this publication, we introduce the concept of ID and demonstrate how our open-source app “SLAMD” provides all necessary steps of the workflow to adapt it in the laboratory, lowering the application barriers. The intelligent screening process, guided by a predictive model, leads to a more efficient and effective data-driven material design process resulting in reduced carbon footprint and improved material quality while considering socio-economic factors in the materials design. KW - Sustainable concrete KW - Machine learning optimization KW - Inverse design techniques KW - Scientific software KW - Data-driven material design PY - 2023 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-589993 DO - https://doi.org/10.17756/nwj.2023-s2-032 VL - 9 SP - 180 EP - 187 AN - OPUS4-58999 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Völker, Christoph A1 - Moreno Torres, Benjami A1 - Rug, Tehseen A1 - Firdous, Rafia A1 - Zia, Ghezal Ahmad Jan A1 - Lüders, Stefan A1 - Lisdero Scaffino, Horacio A1 - Höpler, Michael A1 - Böhmer, Felix A1 - Pfaff, Matthias A1 - Stephan, Dietmar A1 - Kruschwitz, Sabine T1 - Data driven design of alkali-activated concrete using sequential learning N2 - This paper presents a novel approach for developing sustainable building materials through Sequential Learning. Data sets with a total of 1367 formulations of different types of alkali-activated building materials, including fly ash and blast furnace slag-based concrete and their respective compressive strength and CO2-footprint, were compiled from the literature to develop and evaluate this approach. Utilizing this data, a comprehensive computational study was undertaken to evaluate the efficacy of the proposed material design methodologies, simulating laboratory conditions reflective of real-world scenarios. The results indicate a significant reduction in development time and lower research costs enabled through predictions with machine learning. This work challenges common practices in data-driven materials development for building materials. Our results show, training data required for data-driven design may be much less than commonly suggested. Further, it is more important to establish a practical design framework than to choose more accurate models. This approach can be immediately implemented into practical applications and can be translated into significant advances in sustainable building materials development. KW - Sustainable building materials KW - Sequential learning KW - Data-driven materials design KW - Alkali-activated building materials PY - 2023 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-584376 UR - https://www.sciencedirect.com/science/article/pii/S095965262302379X DO - https://doi.org/10.1016/j.jclepro.2023.138221 SN - 0959-6526 SN - 1879-1786 VL - 418 SP - 1 EP - 13 PB - Elsevier CY - Amsterdam AN - OPUS4-58437 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Völker, Christoph A1 - John, Elisabeth A1 - Firdous, Rafia A1 - Hirsch, Tamino A1 - Kaczmarek, Daria A1 - Ziesack, Kevin A1 - Buchwald, Anja A1 - Stephan, Dietmar A1 - Kruschwitz, Sabine ED - Ferrara, Liberato ED - Muciaccia, Giovanni ED - di Summa, Davide T1 - Beyond Theory: Pioneering AI-Driven Materials Design in the Sustainable Building Material Lab N2 - This work focuses on Artificial Intelligence (AI)-driven materials design, addressing the challenge of improving the sustainability of building materials amid complex formulations. These formulations involve various components, such as binders, additives, and recycled aggregates, necessitating a balance between environmental impact and performance. Traditional experimental methods often fall short in managing the complexity of material composition, hindering fast enough development of optimal solutions. Our research explores complex composition materials design through a comprehensive, comparative lab study between Data-Driven Design, using SLAMD - an open-source AI materials design tool, and traditional Design of Experiments (DOE). We aimed to develop a high-performance, alkali-activated material using secondary precursors, aiming for a compressive strength exceeding 100 MPa after 7-days. The findings reveal that AI-driven design outperforms DOE in development speed and material quality, successfully identif. T2 - 4 RILEM Spring Convention and Conference on advanced construction materials and processes for a carbon neutral society 2024 CY - Milano, Italy DA - 07.04.2024 KW - Secondary Raw Materials KW - Data-Driven Design KW - Sequential Learning KW - Design of Experiments KW - Alkali-Activated Binder PY - 2024 SN - 978-3-03170281-5 DO - https://doi.org/10.1007/978-3-031-70281-5_31 SN - 2211-0852 VL - 2 SP - 274 EP - 282 PB - Springer AN - OPUS4-61662 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -