Summary Sorption-induced strain and permeability were measured as a function of pore pressure using subbituminous coal from the Powder River basin of Wyoming, USA, and high-volatile bituminous coal from the Uinta-Piceance basin of Utah, USA. We found that for these coal samples, cleat compressibility was not constant, but variable. Calculated variable cleat-compressibility constants were found to correlate well with previously published data for other coals. Sorption-induced matrix strain (shrinkage/swelling) was measured on unconstrained samples for different gases: carbon dioxide (CO2), methane (CH4), and nitrogen (N2). During permeability tests, sorption-induced matrix shrinkage was demonstrated clearly by higher-permeability values at lower pore pressures while holding overburden pressure constant; this effect was more pronounced when gases with higher adsorption isotherms such as CO2 were used. Measured permeability data were modeled using three different permeability models that take into account sorption-induced matrix strain. We found that when the measured strain data were applied, all three models matched the measured permeability results poorly. However, by applying an experimentally derived expression to the strain data that accounts for the constraining stress of overburden pressure, pore pressure, coal type, and gas type, two of the models were greatly improved. Introduction Coal seams have the capacity to adsorb large amounts of gases because of their typically large internal surface area (30 to 300 m2/g) (Berkowitz 1985). Some gases, such as CO2, have a higher affinity for the coal surfaces than others, such as N2. Knowledge of how the adsorption or desorption of gases affects coal permeability is important not only to operations involving the production of natural gas from coalbeds but also to the design and operation of projects to sequester greenhouse gases in coalbeds (RECOPOL Workshop 2005). As reservoir pressure is lowered, gas molecules are desorbed from the matrix and travel to the cleat (natural-fracture) system, where they are conveyed to producing wells. Fluid movement in coal is controlled by diffusion in the coal matrix and described by Darcy flow in the fracture (cleat) system. Because diffusion of gases through the matrix is a much slower process than Darcy flow through the fracture (cleat) system, coal seams are treated as fractured reservoirs with respect to fluid flow. However, coalbeds are more complex than other fractured reservoirs because of their ability to adsorb (or desorb) large quantities of gas. Adsorption of gases by the internal surfaces of coal causes the coal matrix to swell, and desorption of gases causes the coal matrix to shrink. The swelling or shrinkage of coal as gas is adsorbed or desorbed is referred to as sorption-induced strain. Sorption-induced strain of the coal matrix causes a change in the width of the cleats or fractures that must be accounted for when modeling permeability changes in the system. A number of permeability-change models (Gray 1987; Sawyer et al. 1990; Seidle and Huitt 1995; Palmer and Mansoori 1998; Pekot and Reeves 2003; Shi and Durucan 2003) for coal have been proposed that attempt to account for the effect of sorption-induced strain. Accurate measurement of sorption-induced strain becomes important when modeling the effect of gas sorption on coal permeability. For this work, laboratory measurements of sorption-induced strain were made for two different coals and three gases. Permeability measurements also were made using the same coals and gases under different pressure and stress regimes. The objective of this current work is to present these data and to model the laboratory-generated permeability data using a number of permeability-change models that have been described by other researchers. This work should be of value to those who model coalbed-methane fields with reservoir simulators because these results could be incorporated into those reservoir models to improve their accuracy.

This paper describes the derivation of a new equation that can be used to model the permeability behavior of a fractured, sorptive-elastic media, such as coal, under variable stress conditions commonly used during measurement of permeability data in the laboratory. The model is derived for cubic geometry under biaxial or hydrostatic confining pressures. The model is also designed to handle changes in permeability caused by adsorption and desorption of gases from the matrix blocks. The model equations can be used to calculate permeability changes caused by the production of methane from coal as well as the injection of gases, such as carbon dioxide, for sequestration in coal. Sensitivity analysis of the model found that each of the input variables can have a significant impact on the outcome of the permeability forecast as a function of changing pore pressure; thus, accurate input data are essential. The permeability model can also be used as a tool to determine input parameters for field simulations by curve-fitting laboratory-generated permeability data. The new model is compared to two other widely used coal permeability models using a hypothetical coal with average properties. IntroductionDuring gas production from a coal seam, as reservoir (pore) pressure is lowered, gas molecules, such as methane, are desorbed from the matrix and travel by diffusion to the cleat (natural fracture) system where they are conveyed to producing wells. Fluid movement in coal is controlled by slow diffusion within the coal matrix and described by Darcy flow within the fracture system, which is much faster than the contribution of diffusion. A coal formation is typically treated as a fractured reservoir with respect to fluid flow; meaning that the sole contributor to the overall permeability of the reservoir is the fracture system and the contribution of diffusion through the matrix to total flow is neglected. Coalbeds are unlike other non-reactive fractured reservoirs because of their ability to adsorb (or desorb) large amounts of gas, which causes swelling (or shrinkage) of the matrix blocks.Coal has the capacity to adsorb large amounts of gases because of their typically large internal surface area, which can range from 30 m

Sorption-induced strain and permeability were measured as a function of pore pressure using subbituminous coal from the Powder River basin of Wyoming, U.S.A. and high-volatile bituminous coal from the Uinta-Piceance basin of Utah, U.S.A.We found that for these coal samples, cleat compressibility was not constant, but variable.Calculated variable cleat-compressibility constants were found to correlate well with previously published data for other coals.Sorption-induced matrix strain (shrinkage/swelling) was measured on unconstrained samples for different gases: carbon dioxide, methane, and nitrogen.During permeability tests, sorption-induced matrix shrinkage was clearly demonstrated by higher permeability values at lower pore pressures while holding overburden pressure constant; this effect was more pronounced when gases with higher adsorption isotherms such as carbon dioxide were used.Measured permeability data were modeled using three different permeability models that take into account sorption-induced matrix strain.We found that when the measured strain data were applied, all three models poorly matched the measured permeability results.However, by applying an experimentally derived expression to the strain data that accounts for the constraining stress of overburden pressure, pore pressure, coal type, and gas type; two of the models were greatly improved. Introduction Coal seams have the capacity to adsorb large amounts of gases because of their typically large internal surface area (30 m2/g to 300 m2/g)[1]. Some gases, such as carbon dioxide, have a higher affinity for the coal surfaces than others, such as nitrogen.Knowledge of how the adsorption or desorption of gases affects coal permeability is important not only to operations involving the production of natural gas from coal beds, but is also important to the design and operation of projects to sequester greenhouse gases in coal beds. [2] As reservoir pressure is lowered, gas molecules are desorbed from the matrix and travel to the cleat (natural fracture) system where they are conveyed to producing wells.Fluid movement in coal is controlled by diffusion in the coal matrix and described by Darcy flow in the fracture (cleat) system.Because diffusion of gases through the matrix is a much slower process than Darcy flow through the fracture (cleat) system, coal seams are treated as fractured reservoirs with respect to fluid flow.However, coalbeds are more complex than other fractured reservoirs because of their ability to adsorb (or desorb) large amounts of gas. Adsorption of gases by the internal surfaces of coal causes the coal matrix to swell and desorption of gases causes the coal matrix to shrink.The swelling or shrinkage of coal as gas is adsorbed or desorbed is referred to as sorption-induced strain.Sorption-induced strain of the coal matrix causes a change in the width of the cleats or fractures that must be accounted for when modeling permeability changes in the system.A number of permeability-change models [3,4,5,6,7,8]for coal have been proposed that attempt to account for the effect of sorption-induced strain.Accurate measurement of sorption-induced strain becomes important when modeling the effect of gas sorption on coal permeability. For this work, laboratory measurements of sorption-induced strain were made for two different coals and three gases.Permeability measurements were also made using the same coals and gases under different pressure and stress regimes.The objective of this current work is to present these data and to model the laboratory-generated permeability data using a number of permeability-change models that have been described by other researchers.This work should be of value to those who model coalbed methane fields using reservoir simulators as these results could be incorporated into these reservoir models to improve their accuracy.

Waterflooding is by far the most widely applied method of improved oil recovery. Crude oil/water/rock interactions can lead to large variations in the displacement efficiency of waterfloods. Laboratory waterflood tests and single-well tracer tests in the field have shown that injection of low-salinity water can increase oil recovery, but work designed to test the method on a multi-well field scale has not yet been undertaken. Historical waterflood records could unintentionally provide some evidence of improved recovery from waterflooding with lower salinity water. Numerous fields in the Powder River basin of Wyoming have been waterflooded using low salinity water (about 1000 ppm) obtained from the Madison limestone or Fox Hills sandstone. Three Minnelusa formation fields in the basin were identified as candidates for waterflood comparisons based on the salinity of the formation and injection water and reservoir characteristics. Historical production and injection data for these fields were obtained from public records. Field waterflood data were manipulated to display oil recovery in the same format as laboratory coreflood results. Recovery from fields using lower salinity injection water was greater than that using higher salinity injection water-matching recovery trends for laboratory and singlewell tests. IntroductionAlmost without exception, at the start of a waterflood, water from the cheapest source (usually different in composition than the initial formation water) is used as the injection water, provided injectivity is not adversely affected by formation damage. Historically, little consideration has been given in reservoir engineering practice to the effect of the composition of the salt in the injection water on waterflood displacement efficiency or to the possibility of increased oil recovery through manipulation of the injection water composition. Most laboratory relative permeability tests and displacement tests are done using synthetic formation water as both the for-

Domain-specific generators will increasingly rely on graphical languages for declarative specifications of target applications. Such languages will provide front-ends to generators and related tools to produce customized code on demand. Critical to the success of this approach will be domain-specific design wizards, tools that guide users in their selection of components for constructing particular applications. In this paper, we present the P3 ContainerStore graphical language, its generator, and design wizard.