A PREDICTIVE MODELLING FRAMEWORK FOR DOMINANT AND INTERACTING DELAY RISKS IN METALLURGICAL PROJECTS

Abstract: Delays Large metallurgical projects routinely experience severe schedule overruns due to complex, interdependent delay risks that are often recognised only after impacts occur. Conventional risk registers describe potential issues but offer limited guidance on which risks are likely to dominate or interact under specific project contexts. This paper presents a predictive risk analysis framework that anticipates dominant and interacting delay risks before execution. Using limited historical project data, contextual similarity learning, and interaction-based amplification modelling, the framework generates probabilistic completion behaviour and decision-relevant risk prioritisation. Results demonstrate that delay exposure is highly concentrated, context-sensitive, and systemic, enabling earlier, more focused project decision-making under uncertainty.