Getting it right for projects in early design phase: Advanced techniques to ensure accurate explosion assessments

Abstract

A large vapour cloud explosion followed by a fire is one of the most dangerous and high‐consequence events that can occur at petrochemical facilities. As the size and complexity of facilities increase, designs must consider the potential adverse effects associated with vapour cloud explosions in large congested areas. Designing topside structures to withstand a maximum credible event by avoiding unacceptable escalation of events and damage to safety critical equipment is an essential part of the path toward safer designs. One main factor inhibiting inherently safer designs is the lack of detailed geometry information in the early design phase. When congestion due to small components is not integrated into the explosion pressure analysis model, the design blast loads will be severely underestimated. This article will compare detailed computer‐aided design (CAD) models obtained for both early design phase and as‐built final detailed design, and demonstrate the gross under prediction of accidental explosion loads of the early phase models, as compared to as‐built. In fact, the explosion overpressures can differ by an order of magnitude underlining the need for guidance on the necessary levels of congestion required to obtain reliable blast loads.Gexcon has evaluated traditional congestion parameters such as volume blockage ratio, surface area to volume ratio and packing density as indicators of predicted overpressure as they are commonly used for explosion siting studies. On their own, these parameters have not proven reliable indicators of predicted overpressure and have shown no direct correlation to predicted overpressure across varying geometries. The lack of a direct relationship makes assessing the amount of required congestion to be added to early design phase models challenging. This study will focus on evaluating alternative ways of quantifying congestion, such as non‐dimensional congestion parameters, row spacing, and special obstruction density gradients, to evaluate combustion and eddy/turbulence decay downstream of obstructed areas. With these alternative parameters, we have further refined our anticipated congestion method for petrochemical facilities for accurate predictions of explosion consequences with early design phase models (concept or front‐end engineering design (FEED), phase). This methodology will define what levels of anticipated congestion are necessary to supplement CAD models in early design phase to ensure accurate assessment of explosion hazards and include recommendations for required congestion level for supplementing early phase designs. © 2018 American Institute of Chemical Engineers Process Saf Prog 38: e11995, 2019