Combining Geostatistics With Bayesian Updating To Continually Optimize Drilling Strategy in Shale-Gas Plays

Abstract

We present a new methodology to evaluate subsurface uncertainty during the development of shale-gas plays. Even after many wells are drilled, the average well production and the variation of well performance (economics) remain highly uncertain. The ability to delineate a shale play with the fewest wells and to focus drilling in the most-productive areas is a major factor in commercial success.The importance of probabilistic modeling in managing uncertainty in shale-gas plays is emphasized in several studies. The objective of this study is to develop a practical methodology that addresses these complexities and is dynamic, in the sense that the optimal drilling strategy can be continually updated as we learn the outcome of each well drilled.Maximizing the returns from a shale-gas play is essentially a problem of choosing well locations and number of wells to optimize production volumes and rates. Drilling policies must take account of many already-drilled locations, possible new drilling locations, spatial dependencies between performance at those different (possible) well locations, and the extent of uncertainty concerning whether a well will be economical. These factors cause typical valuation methodologies to be impractical because of the "curse of dimensionality."In this study, an unconventional play is divided into numerous segments. These segments are referred to as cells. In each cell, a fixed number of wells can be drilled. The chance of success (CoS) [of a well with a net present value (NPV) greater than zero] in any given cell is itself considered to be an uncertain variable. An initial probability distribution for the CoS of each cell is derived from analogous plays plus any available information about the specific play.The methodology proceeds as follows. First, as each new well (or group of new wells) is drilled, the outcome is used in combination with the prior probability distribution (with Bayes theorem) to create an updated probability distribution for the CoS of the relevant grid cell. Thus, our initial estimate can be continuously updated as more actual outcomes are realized. Second, the influence of the new CoS on the surrounding cells, because of spatial correlation, is updated by use of indicator Kriging (IK), a geostatistical technique. This combination of Bayesian update (BU) and IK is referred to as the BU-IK method.The methodology proposed in this study informs the development of drilling policies for shale-gas opportunities with a probabilistic model that accounts for the uncertainty in the CoS and its spatial dependency. The use of cells to represent a set of wells simplifies the analysis and greatly reduces the computing requirements. The methodology is applied to a well set from the Barnett shale in Texas.

Toward Production Forecasting for Shale Gas Wells Using Transfer Learning

“…Current shale gas well production and EUR prediction methods are mainly based on production decline curve analysis and its extension, − physics-based simulation prediction, − and data-driven machine learning methods. − Among them, the production decline curve analysis method is based on the fitting of production history curve to realize the prediction of production, which lacks direct consideration of physical process and is an empirical method. The physical-based simulation prediction methods require consideration of complex fracture networks and coupled flow-transport-deformation mechanisms of gas, which is a great challenge to the relevant data and modeling cost, making it difficult to establish an effective model to predict the production and EUR of shale gas wells. − Data-driven machine learning methods avoid the assumptions of physical models and can use data science and technology to gain a deeper understanding of what drives well production .…”

Fusion attention mechanism-based bidirectional long short term memory for carbonate gas well productivity prediction

“…Knowledge-driven approaches encompass graphical methods, numerical simulations, and analytical models, which simulate the intricate flow mechanisms of carbonate reservoirs, offering some understanding of the production process. However, accurately characterizing gas flow in carbonate reservoir modeling poses challenges due to the high heterogeneity, intricate fracture networks, and uncertain gas migration mechanisms encountered in carbonate reservoirs [3,4]. Furthermore, the detailed characterization of carbonate reservoirs is a demanding and time-consuming endeavor.…”

A practical approach for applying Bayesian logic to determine the probabilities of subsurface scenarios: Example from an offshore oilfield

“…There is a standard way of approaching such problems, using Bayesian updating (e.g. Van Wees et al, 2008;Willigers et al, 2013). Nevertheless, examination of the relevant literature indicates that this is not routinely applied in practice in subsurface teams.…”