Closing the Gap Between Reservoir Simulation and Production Optimization

Abstract

The operation of oil and gas fields requires that a multilevel decision hierarchy is used to address field optimization at all timescales. These decisions affect production volumes and costs, not only in their short-term outcomes but also over the life of the field. Depending on the time scale, several models with varying degrees of complexity and detail are typically used to characterize the reservoir (e.g. material balance, fullphysics model, and decline curve). However, the strategy to integrate and maintain these separate reservoir models (describing the same field) is often ad hoc and inconsistent. To make predictions suitable for short-term decisions, proxy-models (e.g. neural networks, response surface) have been proposed. However, these black-box models do not consider the underlying physics of the reservoir phenomena and are limited to the effects captured in the training data set. In this paper, we build upon our earlier work [1] on integration of full-field strategic models (physics-based) and short range operational models based on the moving-horizon, parametric identification approach for reservoir simulation. A short-range, reduced-order model structure is developed, and the model parameters are obtained from production history data. Because the model structure is motivated by the decomposition of a full-physics model, it is expected to be feasible to extrapolate outside the range of history data. The reduced-order model also increases the computational efficiency and effectiveness in carrying out the simulation objectives. The benefit of the proposed model is to assist in the short-term decision making in production operations. This paper provides a discussion of the methodology for identifying such physics-based parametric models for production operation workflows. It also presents case studies to illustrate the benefits of this method for real time production operations and closed-loop reservoir management. Introduction Global production optimization is a multi-time scale problem that requires integration and coordination of several work processes to make decisions on reservoir exploitation strategy, field development and production operations of oil and gas assets. The elements of these work processes span the organization and include several activities at multiple time scales, which pose challenging problems for real time production optimization (RTPO). The industry typically addresses this issue through the adoption of different technologies (for modeling and simulation) across the production supply chain, and through appropriate organization of its personnel. However, all such methods require the description of a reservoir by a model that can predict field production for making optimum decisions. The main purposes of reservoir modeling and simulation are to estimate production and to predict ultimate recovery of reserves under a variety of operating strategies. Such strategies may include various well configurations (e.g. horizontal and multilateral), well locations, well schedules and production rates, and fluid injection mechanisms. The approach to reservoir modeling is often influenced by the knowledge of the underlying phenomena and the purpose of the business work process requiring a model. The following scenarios summarize the different possibilities that arise in reservoir modeling:Determining the effects of measured causes with known causal relationship (e.g. production profiles or reservoir pressure forecasting),.Determining the cause from measured/expected effects with known causal relationship (e.g. production optimization).Determining the relationship between causes and effects, which are both observable and measured (e.g. history matching). The first approach is the conventional forward modeling problem, which is used to predict the evolution of a system output for analytical studies. The latter two approaches are classified as a class of inverse problems, in which the causes (boundary conditions) and the causal relationship (model) are determined from observations of operating data.

The Use of Model Predictive Control for Asset Production Optimization: Application to a Thin-Rim Oil Field Case

“…al [2,3] discretize and linearize the full model to arrive at a linear time dependent state space model:…”

The Promise and Challenges of Digital Oilfield Solutions—Lessons Learned From Global Implementations and Future Directions

“…• The curse of dimensionality is inherent in building proxy models through pure black-box methods and time series data that are used to optimize daily production operations. In this regard, parametric methods with model structures that are based on sound reservoir physics, data-driven models, and an adaptive scheme provide better alternatives (Nikolaou, Cullick, Saputelli, Mijares, and Sankaran, 2003;Awasthi, Sankaran, Nikolaou, Saputelli, and Mijares, 2007;. Another approach uses a control hierarchy based on multiple timescales (Fig.…”

Meeting the Challenges of Real Time Production Optimization— A Parametric Model-based Approach

“…The authors of this paper adopted this representation of the model to develop a parametric modeling methodology for Real-Time Production Optimization (RTPO) strategy [12,13]. A feasible approach to continuous model updating and short-term forecasting using this approach was presented in [14]. A feasible approach to continuous model updating and short-term forecasting using this approach was presented in [14].…”

“…Here, we build upon our previous work on building short-term parametric models [14]. The formulation of the structure of the parametric model is done starting from first principles -conservation of mass and constitutive equations (Darcy's law, compressibility equations and capillary pressure equations).…”

“…Bˆm T h T As discussed in [14], for short-periods of time, the time-dependence of the matrices , and in Eq. (1) is relatively weak.…”