The SPE has split the former "Management & Information" technical discipline into two new technical discplines:
- Data Science & Engineering Analytics
Vardcharragosad, Pichit (PTTEP) | Doungprasertsuk, Chonlada (PTTEP) | Srisuriyon, Jantakan (PTTEP) | Hamdan, Mohamad Kamal (PTTEP)
Abstract The well optimization technique with backward elimination aims to determine the optimum number of wells and their locations that can maximise project value and its recoverable resources, through repeated ranking of candidate wells and eliminating the poorest performer. For a greenfield development, subsurface uncertainties are typically still very large due to limited data from exploration and appraisal wells. This study outlines our approach to perform well optimization with these governing uncertainties in order to support the decision-making process. First, multiple realizations of reservoir models are constructed to represent range of possible outcomes by sampling different values from uncertainty parameters. Backward elimination for well optimization is then performed on those realizations. Wells can be ranked based on means and standard deviations of their performance, and the lowest rank candidate is eliminated from the process one at a time. At this point, project economic and resources are evaluated to find optimum set of wells for field development. Furthermore, well performance data from multiple realization models are carefully analyzed to define key subsurface uncertainties that need to be managed. Solution from this backward elimination with subsurface uncertainty workflow can maximize project valuation because it balances the risk of overspending to drill sub-optimum wells in some realizations with the risk of losing sell opportunity due to insufficient field deliverability in the other realizations. Development decision will be more robust because it is based on the optimum configuration that is applicable irrespective of the unknown uncertain quantities. Moreover, detailed analysis on well performance data allows us to better understand the risk associated with our planned wells so that appropriate de-risking plan can be developed and combined into development strategy. The backward elimination process is straightforward to implement and normally does not require very high computational expense. Thus, it is suitable to be used with uncertainty workflow with multiple realizations of reservoir models, which will increase computational requirement by multiple times. Other commonly used techniques for well optimization such as a Genetic Algorithm (GA) or an Evolutionary Algorithm (EA) are computational expensive by themselves already; and they will require even more runtime when using them with this uncertainty workflow. This paper extends backward elimination approach for well optimization to be used with uncertainty workflow. The overall uncertainty analysis workflow is discussed and provided, with key steps detailing the approach taken. Project valuation and recoverable resources can be further optimized with this new approach, and ultimately can guide the decision making in field development.
International Petroleum Technology Conference
OnePetro PDF doi: 10.2523/IPTC-22973-EA