The key mechanisms of two-phase gas-water flow through coalbeds as utilized in a new three-dimensional. finite-difference simulator, COMETPC 3-D, are presented. The theory for ga transport thrOUgIL coal matrix, transfer of gas frosm matrix to fracture, and the adsorption isotherm boundary condition at the matrix-fracture interface is described. Several unique features of coalbeds which can affe.:t gas producibility including stress-sensitive permeability, matrix shrinkage compressibility, and gas readsorption are incorporated in the model. Applications to history matching production data from coalbed methane wells, studying changes in flow behavior in coals with gas desorption, and modelling horizontal wells are presented. INTRODUCTION Simulation of coalbed methane reservoirs is a more compleX and dat:a intensive process than for conventional gas reservoirs, This is because the primary means of gas storage in coalbeds is by adsorption of methane molecules on coal Surfaces.When a coalbed i!; subjected to pressure drawdown, the desorbed gas must move by diffusion through the extremely low permeability coal matrix in order to reach the natural fracture (cleat) system. Once in 119-1 the cleats, which normally have a high permeability relative to the matrix, gas and formation water flow according to Darcy's law. A number of approaches have been taken to simulate these complex mechanisms in coal. The methods vary from the early work of Price and Abdalla on equilibrium sorption models for degassing coal mines, to the unsteady state models of Ancell at a 1 2 and Kolesar et a 1 3 . An excellent review of numerical simulation work to date pertaining to coalbed methane has been given recently by King and Ertekin 4 The fully 3-D, multi-well model described here is based on the non-equilibrium, pseudo steady-state formulation discussed by King et a 1 5, and as such is an extension of an earlier 2-D models.The 3-D model was developed in conjunction with the Gas Research institute (GRI) and a thirteen company industry consortium. It has been validated against other industry simulators, and used in numerous coalbed reservoir studies, most notably in the Black Warrior and San Juan Basins. THEORY OF FLOW THROUGH COALBEDS Coal cleat-matrix represents a well-defined dual porosity system as described by Warren and Root The face and butt cleats constitute a well .; PETROLEUM

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractThis paper presents the results of using a shale-specific, finitedifference reservoir simulation model to history match and forecast production data from the Barnett Shale reservoir. The paper will illustrate the many uses of the model for vertical and horizontal wells including determining gas in place, recovery factors, optimal well spacing, drainage areas and drainage shapes, optimal fracture half-lengths and conductivities, infill evaluations, horizontal well modeling and optimal number of stimulation treatments, analysis of microseismic data, and compression evaluations. The model was developed in the early 1990s to incorporate all of the production mechanisms inherent in shales including matrix gas porosity, gas desorption isotherm, single-or two-phase flow of gas and water in the natural fractures, layers, complex hydraulic fractures, and variable flowing bottomhole pressures. The paper will discuss the methodology to incorporate all field data into the simulator including core, logs, well test, completion, stimulation, microseismic, and production data. Examples will be given using public datasets. We also show production comparisons between vertical and horizontal wells since this is of topical interest in the play's development history. Furthermore, we discuss the various types of data to collect, their importance to proper stimulation design, and the integration methodology to evaluate and complete shale reservoirs.

For many years gas has been produced naturally from Devonian shales throughout the Appalachian Basin at rates which made them commercially feasible. In 1975, because of the deteriorating gas supply situation in the United States and increasing demand for energy, the U.S. Department of Energy (DOE) established the Eastern Gas Shales Project (EGSP) as part of the Unconventional Gas Recovery (UGR) program to study and characterize and to develop economically feasible technology for further exploitation of Devonian shale gas reservoirs. Devonian shale gas reservoirs typically are characterized by a low storage, high flow-capacity natural fracture system fed by a high storage, low flow-capacity rock matrix. In this study analytical solutions are developed to analyze the basic fractured reservoir parameters that control well productivity. These parameters include fracture system porosity and permeability, matrix porosity and permeability, and matrix size. It is shown that the conventional well test method does not usually work for fractured Devonian shale gas reservoirs. For most cases, the semilog plot of the drawdown and buildup data does not show two parallel straight lines with a vertical separation. Numerical solutions are also used to include the Klinkenberg effect and desorption in the shale matrix. INTRODUCTION Fractured reservoirs have been studied for several decades. However, during the last three decades, most reservoir engineering studies have been directed toward homogeneous formations. The earliest discussions of fractured reservoir performance was the analysis of the Spraberry Field in West Texas.1 In 1959, Pollard2 presented a method to determine fracture volume from pressure buildup data. The Pollard method was extended by Pirson and Pirson3 to calculate the matrix volume of a fractured reservoir. One of the classic papers on fractured reservoirs by Warren and Root4 considered a dual porosity system which consisted of a fractured porous medium in which matrix blocks acted as a uniformly distributed source. Natural fractures are replaced by equivalent sets of horizontal and vertical fractures. Warren and Root presented an approximate analytical solution for naturally fractured reservoirs based on the Barenblatt and Zheltow formulation.5 They showed that the semilog plot of pressure drawdown or buildup data for an infinite reservoir displaces two parallel straight lines whose slopes are related to the flow capacity of the formation. The vertical separation of these straight lines is related to the relative storage capacity of the fractures. Warren and Root also showed that the Pollard2 and Pirson and Pirson3 techniques could lead to erroneous results in some cases. Odeh6 presented a model similar to that of Warren and Root4. His results did not displace two parallel straight lines. He concluded that fractured reservoirs cannot be distinguished from homogeneous ones. Later, Kazemi7 presented numerical solutions for fractured reservoirs. The Kazemi model consisted of a set of uniformly spaced horizontal matrix layers separated by fractures. This model can be considered a special case of the Warren and Root model. Kazemi did not use the Warren and Root assumption (i.e., that flow in the matrix blocks is quasi-steady state). Kazemi, et al.8 studied the pressure behavior of an observation well in a naturally fractured reservoir with an adjacent well producing at a constant rate. This study showed that the early time response was substantially different from that of an equivalent homogeneous reservoir. Crawford, et al.9 analyzed more than 20 field-measured pressure buildup curves in a reservoir known to be naturally fractured and concluded that the Warren and Root model adequately described the buildup response and is, therefore, useful in determining effective fracture permeability.

No abstract

An algorithm has been developed for computing the pressure response for a well with constant wellbore storage and non-Darcy skin factor across the completion. The algorithm has been used to generate type curves for drawdown and buildup tests. The builduppressure-derivative response for a well with non-Darcy flow across the completion exhibits a much steeper slope during the transition out of wellbore storage than that of a well with constant skin.No reservoir model with constant wellbore storage and skin can reproduce this steep derivative. Thus, if it is present in a buildup test, the well is experiencing either decreasing wellbore storage or decreasing skin factor, or both.With the new type curves, under favorable conditions, both Darcy and non-Darcy skin components may be estimated from a single buildup test following constant-rate production. The new algorithm also may be used to model a test sequence comprising multiple flow and buildup periods. IntroductionNodal production-system analysis 1 is one of the primary tools for optimizing production and predicting well performance. In nodal analysis, reservoir performance is described through the inflowperformance-relationship (IPR) curve. To construct accurate IPR curves for gas wells, both Darcy and non-Darcy skin components must be known.The non-Darcy skin is traditionally estimated by performing a multirate test. The effective skin factor is then graphed as a function of flow rate, allowing the Darcy and non-Darcy components to be determined from a straight-line fit through the data.When multirate tests are not available, the non-Darcy flow coefficient may be estimated from correlations. However, the resulting values may be in error by as much as 100%. 2 The behavior of a pressure-transient test with infinite-acting radial flow with constant wellbore storage and skin factor is well known. 3,4 Other authors have considered the case of variable wellbore storage with constant skin factor. 5,6 This paper examines the behavior of a well with constant wellbore storage and rate-dependent skin factor for drawdown and buildup tests. Buildup tests with non-Darcy skin factor exhibit a much steeper pressure derivative during the transition out of wellbore storage than those with Darcy skin factor, as seen in Fig 1. The new type curves provide three significant contributions to the industry: (1) they allow the test analyst to recognize non-Darcy flow from a log-log diagnostic plot of pressure change and pressure derivative; (2) they provide estimates of both Darcy and non-Darcy components of skin factor from a single buildup test, allowing construction of IPR curves based on well performance instead of correlations; and (3) they help identify the cause of high skin factors.High-rate wells with high skin factors represent excellent candidates for stimulation. 7,8 However, because of the risks involved in any workover, there often is a reluctance to stimulate such a well. Given that high-rate wells are the ones most likely to exhibit non-Darcy skin, the new type curves will...