Probabilistic Assessment of Coarse Aggregate Reliability for Infrastructure in Northeast India using Bayesian Networks and Variable Elimination Method

Coarse materials are critical in construction, playing a pivotal role in the durability and structural integrity of concrete and other infrastructure components, particularly in disaster-prone regions. This research investigates the resilience and reliability of coarse materials sourced from various quarries in Assam, Northeast India, which is essential for disaster prevention and infrastructure resilience. By employing Bayesian Networks and the Variable Elimination technique, a comprehensive probabilistic framework was developed based on nine essential laboratory tests including water absorption capacity, specific density, flatness index, elongation index, stability, material crushing strength, material impact strength, alkali-silica reaction and Los Angeles abrasion strength. Specimens were collected from 18 sites and the assessment results were analyzed to determine the probabilistic conditions of each node in the Bayesian network. The findings revealed significant variations in material quality across different locations, with some areas exhibiting reduced reliability and increased vulnerability to structural failures. These insights are crucial for implementing targeted interventions such as enhancing quality control, sourcing higher-grade materials and optimizing construction techniques to improve disaster resilience. Additionally, regular monitoring and maintenance can mitigate potential infrastructure failures, thereby strengthening disaster preparedness. This study provides valuable insights to support the Assam Road Research and Training Institute (ARRTI) and similar organizations in improving construction practices and ensuring the long-term resilience of infrastructure projects in the region.