This is the first of two papers describing the application of simulator-optimization methods to a natural gas storage field development planning problem. The results presented here illustrate the large gains in cost-effectiveness that can be made by employing the reservoir simulator as the foundation for a wide-ranging search for solutions to management problems. The current paper illustrates the application of these techniques given a deterministic view of the reservoir. A future paper will illustrate adaptations needed to accommodate uncertainties regarding reservoir properties.
Although reservoir simulation is a well-established component of reservoir management throughout much of the petroleum industry, little use has been made of reservoir simulation coupled with systematic optimization techniques, i.e., simulation-optimization.
The main advantage of applying optimization tools, per se, to decision-making problems is that they are less restricted by human imagination than conventional case-by-case comparisons. As the number of competing engineering, economic, and environmental planning objectives and constraints increases, it becomes difficult for human planners to track complex interactions and select a manageable set of promising development strategies for examination. Using optimization techniques, the search can range over all possible combinations of variables, locating strategies whose effectiveness is not always obvious to planners.
The advantage of coupling the reservoir simulator to these optimization tools is that the search for strategies can be based on the simultaneous evaluation of reservoir performance measures and other economic, environmental, and policy considerations. It is no longer necessary to treat technical decisions driven by simulator forecasts of reservoir response and these other components of the decision-making process as separate steps.
The single biggest obstacle to the application of optimization techniques using a reservoir simulator as the forecasting tool is the computational time required to complete a single simulation. Even the examination of 10 variations on a field development plan becomes cumbersome when a single run requires hours to complete. Extending the use of these simulators into optimization regimes involving hundreds or thousands of runs poses a computational problem bigger than most organizations are willing or able to tackle.
The ANN-GA solution to this problem is to train artificial neural networks (ANNs) to predict selected information that the simulator would normally predict. A heuristic technique such as the Genetic Algorithm (GA) then searches for increasingly better strategies (for example, the most productive infill drilling pattern), using the trained networks to evaluate the effectiveness of each strategy in place of the original simulator. After analysis of the results of the search, the best-performing strategies are submitted to the original simulator to confirm their performance. The components of the methodology are illustrated in Fig. 1.
The ANN-GA methodology was first developed to address computational bottlenecks in applying simulation-optimization techniques to groundwater remediation applications. Studies employing 2-D flow-and-transport models of a contaminated groundwater Superfund site have documented the benefits of simulation-optimization both with1,2 and without3 the assistance of ANNs. These studies are part of the long-standing interest in the field of water resources in the use of simulators in formal decision-making contexts4.
The emphasis in the petroleum engineering literature, in contrast, has been on the evaluation of small sets of carefully selected scenarios, as exemplified by the work of Kumar and Ziegler5, Coskuner and Lutes6, and Kikani and Smith7. However, there are a few studies which have applied techniques which bear some resemblance to simulation-optimization methods.