Optimal Investment and Operation Strategy for Oil Spill Preparedness: With Case Study on the Northern Norwegian Continental Shelf

Abstract

The development of petroleum activities in the northern area has brought an increased focus on developing offshore oil spill preparedness. Remoteness and harshclimate dominating large parts of the region pose new challenges compared to the more well established areas further south. The consequences of a major oil spill arehigh, and to prevent catastrophic environmental damage, sufficient preparednessresources are required. The establishment and operation of a sufficient preparednessregime is expensive, and the need for developing efficient solutions is present.In this thesis, a multistage stochastic programming model is developed, that offersdecision support regarding investments related to oil spill preparedness,consideringfuture oil activities. The future activity level in the northern area is highlyuncertain, and hence, the activity level is incorporated into the model as stochastic parameters. To the knowledge of the authors, no previous work has modelled the oil spill preparedness problem as a multistage stochastic programming problem that considers future scenarios of oil activity. In addition to strategic investment decisions,the model handles allocation of preparedness systems to bases and standing emergency vessels. A system is defined as a specific combination of oil spill equipment. The allocation of systems is based on predefined requirements for every installation, stating the number of systems that must be present at a spill site within a given number of hours, if an oil spill accident occurs. Offshore response in proximity to the source of the oil spill is the main focus. Several algorithms are developed to preprocess the data, enabling the final optimization model to be a specialised version of the Set Covering Problem.The implemented model is applied to different scenarios of future oil activities inthe Barents Sea South and South-East, and the waters around Jan Mayen. In addition to the stochastic model, a deterministic model is executed, to assess the value of the stochastic solution. For the given test case, the stochastic model provides almost no additional value over the deterministic model, which can be explained by the high level of flexibility incorporated into the decision process.