A Method for Integrating Response Surfaces into Optimization Models with Real Options: A Case Study in Gas Flooding

Abstract

The goal of a field development optimization process, or workflow, is to investigate various options and determine a course of action that will deliver the largest expected value from an asset. The analysis is often complicated by uncertainty in important inputs. Ideally, operators desire workflows and tools that integrate reservoir engineering and optimization principles in a fast-solving model that can be used to explore the full range of the uncertain inputs. This need is acute in the screening and concept selection stage where the primary objective is to determine the sensitivity of competing concepts to the sources of uncertainty. In these early stages, model results can be used to determine whether additional information should be collected, and to narrow down the number of competing options.The objective of this research is the development of a workflow and tool that integrates reservoir response surfaces within a project optimization model that contains real options. The incorporation of real options is critical because a static view of capital investment and facility constraints causes a systematic undervaluation and can introduce error to development decisions. The new workflow and integrated reservoir-economic optimization tool developed in this research leverage methods and engineering work products that are already known to industry, for example experimental design (ED) and response surface methods (RSM).A demonstration is provided for a gas flood project using a stylized reservoir. Specifically, we investigate the selection of initial well configurations and injection capacities while simultaneously accounting for the (real) option to update these decisions after production information is acquired in the early periods of production. The workflow is used to optimize the development of a gas flood. As a second step, the workflow is used to solve a value of information problem. BibliographyLake, Larry W. 1989. Enhanced Oil Recovery, Englewood Cliffs, New Jersey: Prentice Hall.

A data-driven smart proxy model for a comprehensive reservoir simulation

“…Dejea [36]has presented a procedure using response surface methodology to manage the uncertainties in production prediction. Purwar [37]has built a workflow for the optimization of the gas flood using response surface methodology. Rezo (2008) study proposed the use of response surface methodology to secure a reliable initial guess for nonlinear inversion and for understanding the separate contributions of the various measurements of the formation tester to specific inversion parameters.…”

Application of Machine Learning and Artificial Intelligence in Proxy Modeling for Fluid Flow in Porous Media

“…In general, the more complex the mathematical function the more computational effort is required to obtain a proper function which can accurately describe the system behavior. In petroleum engineering, this technique has been applied in several types of studies such as history matching [1], upscaling geological models [2], field development planning [3], optimization [4], and many other studies. However, the most common area where this methodology has been applied is in uncertainty analysis of a particular process such as oil production [5,6] and CO 2 sequestration [7].…”

Developing a Grid-Based Surrogate Reservoir Model Using Artificial Intelligence

“…In the area of petroleum engineering, this technique has been applied in several types of studies such as history matching (Slotte and Smorgrav 2008), upscaling geological models (Narayanan, et al 1999), field development planning (Carreras, Turner and Wilkinson 2006), optimization (Purwar, Jablonowski and Nguyen 2010), and many other studies. However, the most common area where this methodology has been applied is in uncertainty analysis of a particular process such as oil production (Ahmed, et al 2013) (Al Salhi, et al 2005 and CO2 sequestration (Yeten, Castellini, et Uncertainty analysis requires identification of the involved uncertain parameters that affect the desired output of a system and evaluation of their degree of influence.…”