Assessment of strength and self-healing properties of bacterial concrete using machine learning techniques and microstructural characterization
摘要
This study investigates the feasibility of utilizing bacteria to improve the mechanical properties and self-healing capabilities of concrete. Five strains of bacteria (Bacillus subtilis, Bacillus licheniformis, Bacillus flexus, Pseudomonas stutzeri, and Escherichia coli) were examined at different concentrations (10⁴, 10⁵, and 10⁶ cells/mL) and compared to a reference concrete. The workability (slump, compaction factor, Vee Bee consistometer) and compressive, flexural, and split tensile strengths were assessed before and after damage induced by the study. A total of 48 cubes, 48 cylinders, and 48 prisms were cast and tested to evaluate compressive, split tensile, and flexural strengths, respectively. Results showed that workability was increased with bacterial incorporation. Additionally, bacterial concrete showed significantly better strength properties, particularly at high bacterial content. Microstructural characterization (SEM, XRD) showed higher density, efficient crack healing, and increased durability in bacterial concrete, with Bacillus subtilis (M16) and Bacillus licheniformis (M4) demonstrating better performance. Self-healing tests also validated that bacterial concrete showed considerably higher strength recovery than normal concrete. XGBoost regression model and Forest Tree Regression model is used to predict the mechanical properties. Bacillus subtilis at 10⁶ cells/ml produced the best results. These results underpin the potential for using bacterial concrete as a sustainable, long-lasting option with improved mechanical properties and self-healing characteristics, thereby having the potential to prolong the lifespan of concrete structures.