Application of multiple criteria decision analysis to compare environmental impact factors in Statoil

摘要

In this study, Multiple Criteria Decision Analysis (MCDA) is applied to the risk assessmentframework of Environmental Impact Factors (EIFs) in Statoil. The objective for theapplication is to integrate EIFs to an indicator for overall environmental risk related toemissions and discharges from petroleum activities and operations. To reach this indicator,expert judgements of the relative importance of environmental compartments are consideredto be essential. The study is a part of the integrated HSE risk management project at Statoiland is based on the principles and experiences from the MCDA trial session in 2006.To further investigate and refine the approach, the case study of drilling technologyalternatives at the Norne field is applied. The Statoil goals of zero harm to the environmentand continuous improvement of environmental performance form the basis of the problemdesign. Five decision alternatives are identified and relevant EIF scores for these alternativesare assembled or estimated. The EIFs are tailored to act as decision criteria that reflect theneeds of scientific accuracy and practical viability, and the scores are accordingly modified.The special features of the EIF for air emissions require a different approach for this factor.For each alternative, criteria scores at the compartment level are aggregated and weightedthrough the use of an optimisation model, and a total performance indicator for eachalternative is identified. Even though the EIF scores are calculated on the basis of genericdata, the area-specific sensitivity of environmental compartments results in importanceweights that are limited to a pre-defined area. The set of weights for the relevant influencearea in the Norne case is elicited through two expert panel sessions.As a response to challenges at the first session, the problem design is additionally modified.The most important adjustments are related to weight elicitation on a unit basis and theintroduction of “risk scores”. Due to similarity of data in the Norne case and a temporarilyexclusion of air emissions, three decision alternatives are identified as equally optimal. Theresults from the second session indicate that the current problem design has increased thefeasibility of the MCDA approach, but that challenges like integration of air emissions andrelevance of sensitivity aspects remain.