Many critical factors must be considered in determining the most feasible and appropriate options for decommissioning offshore oil and gas platforms, including alternatives from complete removal to partial removal with artificial reefing. The authors of the complete paper describe how to use multivariable decision analysis to facilitate offshore platform decommissioning. They implement decision analysis as a software tool to clarify and evaluate decommissioning alternatives against a comprehensive set of objectives, including both market and nonmarket values.
Recent experience has demonstrated that an effective decommissioning process must include aspects that are inherently difficult to monetize, such as the production of greenhouse gases, gains or losses of ecological services, and regulatory compliance. Differences in stakeholder perspectives regarding the value and importance of nonmarket issues can result in distrust when choices are considered in purely financial terms.
In 2008, a comprehensive policy analysis was commissioned in California to examine the range of decommissioning options and their potential effects. A mathematical decision model called PLATFORM enabled an integrated synthesis of a full range of competing outcomes and values and provided all stakeholders with an analytic tool that supports examination of trade-offs across a range of decision options.
The multiattribute decision framework provided a comprehensive comparison of the decommissioning options against both quantitative and qualitative attributes and allowed stakeholders to effectively understand the science and assess the societal trade-offs for these different options.
Decision Options. Alternatives to the complete removal of decommissioned offshore oil and gas platforms have been proposed, including:
Only the rigs-to-reefs option eliminates the eventual need for platform removal.
A set of criteria were applied qualitatively and resulted in clearly sorting options into two categories (evaluated in detail or examined briefly and eliminated). Two “use options” (complete removal and partial removal as part of conversion to an artificial reef) and one “disposal option” (onshore dismantling) warranted detailed analysis.
PLATFORM was developed as a computer model to evaluate alternative decommissioning decision strategies and the conflicting criteria (attributes) involved. Key objectives in the design included:
Model Development. The model incorporates user interfaces, a hierarchy of influence diagrams to build and organize the model, range-sensitivity analysis to identify key sources of uncertainty or disagreement, and Monte Carlo simulation to analyze uncertainties. Model dimensions include such elements as platforms, decision options, scenarios, and attributes.
Differences among 27 platforms affected the cost and environmental effects of the two decommissioning options, as well as their suitability for artificial reefing. The preferred decommissioning method may therefore vary among platforms.
Model details are organized as a hierarchy of modules, each structured as an influence diagram. Influence diagrams identify key variables, including data types, uncertainties (oval nodes), decisions (rectangular nodes), and result variables, with the influences represented as arrows between them (for schematics, see Figs. 3 and 4 of the complete paper). For each model component, an influence diagram was initially developed that identified the top-level conceptual structure, and progressively added conceptual and technical detail as needed to complete the analysis.
Sensitivity analysis allowed users to explore which uncertainties have the greatest effect on results, and whether realistic changes in component estimates might change the choice among options.
Structuring Multiple Objectives or Attributes. Objectives were organized in a total of eight attributes. Some market and nonmarket value attributes (e.g., cost, fish biomass) can be quantified readily. Other nonmarket value attributes (e.g., impacts on marine mammals, other ecosystem services) are difficult to quantify because of inadequate data available or incomplete understanding of causal processes. These latter attributes were assessed and scored in narrative terms.
The “Compliance” attribute is unique from the other attributes because it reflects a categorical preference for one option (complete removal) rather than a gradient of costs/benefits.
Multiattribute Decision Analysis. Multiattribute utility theory provides a structured approach to evaluating decisions under uncertainty on multiple objectives or attributes. It provides ways to represent a person’s preferences over alternatives characterized by uncertain attributes as a scalar utility function. Additive independence means that a person’s preferences show no interactions among attributes—preferences over values of one attribute are not affected by the level of other attributes. Additive independence is often a reasonable approximation of people’s preference structures with limited uncertainty, and allows decomposition of the aggregate utility function into a simple weighted sum of attribute-specific utilities
This assumption allowed for the assessment of the utility function for each attribute separately, and from the weights used to combine them into a multiattribute utility function. Applying this approach involves the following steps:
Identify and organize attributes
Define a clear scale for each attribute—either cardinal, meaning quantified, as in USD for direct costs, or ordinal, meaning a list of outcomes in order of preference
Define a single-attribute utility function to score the possible levels of each attribute into a utility from 0 (worst outcome) to 100% (best outcome)
Select swing weights (or equivalent costs) to model stakeholder preferences about relative value or cost for each attribute from which one can obtain weights
Combine the swing weights and attribute scores into an overall multiattribute utility for each decision option
For the qualitative attributes, a five-point scale, ordered from the worst to best outcome plausibly possible for any platform, was developed. Intermediate points were labeled poor, medium, and good.
The three attributes based on quantitative models are Cost, Fish Production, and Ocean Access. It is reasonable to assume a linear utility function for each of these attributes over the range of interest for these decisions.
Normalized scores are displayed by attribute for Platform Harmony for the complete and partial removal options. Partial removal scores higher than complete removal on cost and all environmental impacts—except birds, for which they score the same—because both options remove the surface platform structure. Complete removal performs slightly better on changes to Ocean Access, because it removes the underwater parts of the jacket that must be avoided by many commercial-fishing-gear types. Strict Compliance is the key exception to this pattern: partial removal scores zero, and complete removal scores 100. Thus, the choice between complete and partial removal depends almost entirely on the judged importance of strict compliance relative to costs and environmental effects.
Combining Attributes and Swing Weights. PLATFORM offers two methods to assess weights for aggregating over attribute scores—SMARTS (Simple Multiattribute Rating Tool with Swing Weights) and an equivalent cost method that lets users express preferences for each attribute score in terms of cost. Swing weights recognize that the importance of each attribute should depend on the range of each attribute. Asking whether dollar cost is more important than the effect on marine mammals in the abstract is an ill-defined question; it is more meaningful to ask whether the range of Cost (a market value), from zero to USD 250 million, is more important to a stakeholder than the range of outcomes on Marine Mammals (a nonmarket value), from no effect to the death of 20 sea lions.
Sensitivity to Preference Weights. Developing the preference weights begins by examining the effect of varying each swing weight around its base value. Interestingly, Cost Uncertainty had the second-lowest sensitivity. In other words, uncertainty about the factual question (the direct cost of decommissioning) had considerably less effect on results than stakeholder disagreements about relative preferences for the top seven attributes, as reflected in their swing weights.
It was also interesting that the sensitivity bar for only one attribute, Compliance weight, reached below zero. In other words, Compliance weight was only variable to the extent that an extreme change could change the preferred decision—from partial to complete removal.
Compliance was one of only two attributes that favored complete removal. It should not be surprising that a higher weight on Compliance favors complete removal. The other attribute favoring complete removal was Ocean Access.
The use of multiattribute decision analysis effectively allowed a large set of potential decommissioning options and their combinations to be condensed to a smaller decision tree with a more limited number of options that can be subjected to additional detailed analysis. The decision analysis showed that removing the upper portion of the platform preserves the majority of the ecological value while removing potential interference with shipping. Additionally, the analysis demonstrated the benefits of leaving the remaining platform components in place as an artificial reef.
The study clarified how different stakeholder preferences affect the choice among decommissioning options. Strict Compliance with original leases remains the single most compelling reason for some stakeholders to favor complete removal. Unlike other attributes, this attribute is necessarily structured as a binary outcome—complete removal meets strict compliance, and other options do not. Discussions with stakeholders suggested that most viewed strict Compliance as less important than environmental impacts and decommissioning costs, especially if some of the savings from partial removal were applied to ocean conservation and ecosystem services.