Multi Objective Decision Making Applied to Artificial Lift Method SelectionJames Tauchnitz
Australian School of Petroleum
University of Adelaide
Honours Project 2009
This study focuses on the application of Multi Objective Decision Making (MODM) techniques to the artificial lift method selection problem. Artificial lift selection is rarely performed using a structured decision making process. Typically, production engineers select artificial lit methods from a limited knowledge of alternatives. This often results in ‘force fitting' a favored method to a particular well which inevitably results in poor performance and failures. The work outlined in this report will assist the decision maker to select the most appropriate artificial lift method for any given well.
The SMARTER (Simple Multi Attribute Rating Technique Eliciting Rankings) MODM technique was applied to the decision problem. SMARTER divides the problem into a series of smaller, more manageable tasks. The first phase of the process is framing which develops the decision relevance, context, the alternatives and the decision attributes. Modeling is the next phase of the process which involves scoring alternatives against each of the attributes and finding an overall weighted value for each alternative. The final phase, assessing, addresses any competing objectives and tests the robustness of the decision against changes to the input parameters.
In total eight different alternatives and twenty eight different attributes formed the decision pay off matrix. The decision evaluators were three independent experts from industry. Each expert had extensive experience in artificial lift selection and design. Each expert was provided identical reservoir, well, production and field data for a particular case in the THUMS E. Wilmington Field operations, offshore Long Beach California. Using this information they were asked to rank the decision objectives which were used to determine the objective weights.
The results of the analysis showed that the preferred artificial lift method for the particular case was the hydraulic jet system. Interestingly ESP's were overwhelmingly the most widespread artificial lift method in use in the THUMS E. Wilmington field. However, ESP's were ranked the 4th best alternative in this study.
It is recommended that for future work, statistical analysis be performed with the aim of quantifying any dependencies in the objectives. Additionally it is recommended to compare the results to a model that uses non linear value functions. This could be achieved using the bisection method.
In its application to artificial lift selection, SMARTER is not an optimisation tool. Rather a screening tool to select the best alternative from a defined set of objectives. The results of the SMARTER analysis should link into design calculation by performing detailed designs for all the alternatives forming the efficient frontier. Preference should be given to the higher ranked alternatives