Bow-tie risk assessment combining causes and effects applied to gas oil storage in an abandoned salt cavern

Abstract A semi-quantitative risk assessment is presented for the storage of gas oil in depleted salt caverns in the Twente region, the Netherlands. It is based on a bow-tie model, in which an incident, leakage of gas oil from the storage system (cavern and wells), is evaluated by assessing its possible causes and effects. The causes are all the events that may lead to leakage from the storage system. The effects are the consequences of the leakage. It is considered that the most serious of the subsurface risks is contamination of the groundwater due to upward migration of the gas oil to the surface. A unique aspect of our risk assessment is the combination of causes and effects. The effects of containment/failure are quantified at multiple time scales using a numerical flow model for multiphase flow through porous medium, based on the geohydrological properties of the subsurface of the Twente area. The probability of occurrence of loss of containment/failure (causes) is quantified semi-quantitatively, using the causal relationships between the causes and effects. Modelling of the leakage shows that, as expected, leakage from the well above the hydrogeological base in the phreatic aquifer produces an immediate risk of contamination of the upper groundwater. However, leakage at a deeper level does not pose a risk of contamination of groundwater, because the low porosity and permeability of the geological layers prevent the upward migration of leaking gas oil. The semi-quantitative approach to the probability of failure finds that for multiple scenarios (e.g. well failure, unstable cavern, high pressure) and in the absence of human intervention, the probabilities of failure are medium to high. If human intervention is assumed, these probabilities of failure diminish considerably, especially those associated with the well. These findings are consistent with those from other hazard studies on storage in salt caverns. The causes (probabilities of failure) and effects (modelling of leakage) together indicate that for most scenarios the risk is low when human intervention (e.g. monitoring of the well) is assumed. Notwithstanding our conclusion that the risk of leakage associated with gas oil injection and storage in salt caverns is low, an extensive monitoring plan should be formulated to monitor the containment of the gas oil in the storage system and its long-term stability, to ensure timely human intervention that reduces the risk considerably.