Delays in oil and gas construction projects present major challenges due to the sector's complex operations, high capital investment and strict timelines. Such disruptions affect project execution and can have serious socioeconomic impacts. The purpose of this paper is to identify key factors contributing to schedule delays and develop predictive models using artificial neural networks (ANN), decision trees (DTs) and multiple linear regression (MLR) to estimate delay percentages.
This study adopts a mixed-methods approach, combining qualitative and quantitative techniques. Expert interviews and a literature review identified and prioritized key delay factors. Data from completed oil and gas construction projects selected through purposive sampling to reflect varied sizes, complexities and locations were analyzed using correlation and outlier tests to ensure reliability. Predictive models: ANNs, DTs and MLR, were then developed, trained and validated with real project data.
Design consulting experience, project location and scope changes are assessed as the leading factors influencing delay percentages in oil and gas construction projects. The MLR demonstrated the highest accuracy, exceeding 96%, outperforming both ANNs and DTs.
This paper focused on the quantitative models rather than the qualitative aspects of predicting schedule delays in construction projects in the oil and gas industry. The proposed models support planning stages by enabling more realistic project schedules and decision-making processes. The proposed models also empower project stakeholders to optimize project outcomes.
