Effective well management and a productive wellwork program are valuable and integral business objectives. Wellwork involves various well interventions and optimisation activities for enhancing and extending hydrocarbon production. These remedial processes involve substantial CAPEX and OPEX, as well as other resource allocations.
Failure to prioritize objectives and improper selection of candidate wells can have significant implications on both derived value and potential risk. A primary challenge is to ensure that wellwork is delivering production growth while maintaining cost efficiency. Well-by-well reviews with actionable decision support information will provide the best method for identifying potential production improvements. The selection and prioritisation of candidate jobs is a critical investment decision.
This paper addresses the business problem of reducing the uncertainty of well work program outcomes’ so that more informed choices can be made from all the options such that the benefits and value of an overall well work program is enhanced and optimized.
It illustrates the use of data-driven models for estimating key performance indicators for wellwork jobs and predicting the likely outcome for a new planned job using pre-determined success criteria. Nine different machine learning and advanced analytics learning schemes were applied to the training dataset of wellwork history. The competing models performance was evaluated on a separate validation data set for a balance between best fit and prediction accuracy.
The application of developed models provided intelligence augmentation for the decision-making process. This methodology embeds learning from past wellwork activities to streamline and guide complex workflows. The business value for embedding quantitative predictions into strategic and operational decision-making processes is realized in reducing less-favorable investments and maximizing the value of wellwork.