Case Study Review to Reinforce a Decision-Based Experimental Design Methodology

Abstract

Design of experiments (DoE) is an industry methodology designed to test the impact of different variables on an objective function without running every single scenario possible. Despite its many advantages, there still exists the tendency to engage in re-work, particularly when the results of the experiments do not align with expected outcomes. This paper highlights the need for a decision-based approach for DoE studies to eliminate inefficiencies in resource utilization. A decision-based process focuses changes in variables and analyses on those that have the potential to impact the final decision for which the study is being carried out. A DoE study for a virgin reservoir development, conducted in two run cycles is used to highlight this. For the first cycle, the top variables from stakeholder engagement and initial Pareto and Tornado analyses were carried forward for final runs and analyses. The same was repeated for the second cycle, with a change in the distribution of the key variables, within the limits of analog data available. The analyses of both run cycles show close alignment of the results of both run cycles. Also, the same development decision was reached from the results of both run cycles, even though changes were made to the input variables for the second run cycle. A review of the Tornado chart for either run cycle shows that the development decision can be arrived at with either one, without the need for the other. This shows that the number of run cycles for similar studies can be minimized to ensure that only those that have the potential to change the decision to be made are carried forward. This will ensure optimum use of the limited available technical, time and computational resources, allowing the professional to do more with less.