Modeling Delay Amplification in Mega Projects Using Circular Impact Assessment and Scientific Machine Learning

Mega infrastructure projects in developing economies routinely experience schedule delays that generate cost escalation, deferred socio-economic benefit, and ongoing environmental exposure. Existing research treats these consequences as isolated, linear outcomes, leaving unexplained how delay-induced impacts interact, compound, and feed back into further delay after project completion. This working paper presents the Circular Impact Assessment (CIA) model, a four-layer framework that treats schedule delay, cross-domain impact (economic, social, environmental), and sustainability performance outcomes as a circular rather than linear outcome, connected by a theorized secondary delay intensification loop. The model is applied to two Bangladeshi mega projects which are the Padma Multipurpose Bridge Project (PMBP) and the Dhaka Metro Rail Line 6 (DMRP), using an embedded mixed-methods design combining thematic coding, VADER sentiment scoring, SPSS-based statistical validation, and a scientific-machine-learning simulation layer built on Universal Differential Equations (UDEs), which embed mechanistic cost-escalation dynamics within a neural residual component capable of approximating socio-political feedback that purely mechanistic models cannot capture. Because this work is based on an ongoing dissertation, its results section previews the patterns that the model's theoretical structure and empirically grounded parameters lead us to anticipate rather than reporting completed findings. The research describes circular, feedback-sensitive evaluation as a methodological advance over linear delay-impact models, as well as its connected contributions to SDGs 8, 9, 11, 13, and 15.