This paper describes a new experimental technique developed to study the gas diffusion properties of coal using the transient method. It enables measurement of the diffusion coefficient, D, and estimate the variation in its value with changes in pressure and concentration. The preliminary conclusion of the study is that the diffusion of methane in coal cannot be described by a unipore model or a single diffusion coefficient. For simulation of long-term gas production from coal seams, D should be treated as a time dependent variable. The results obtained so far clearly show that the value of D decreases continuously with decrease in gas pressure. The most likely reason for the variation in the value of D is the change in the pore sizes in coal. This, in turn, might be resulting in a change in the mechanism of diffusion from regular to Knudson, or Knudson to surface.
With the onset of the energy crisis in the early 1970's, methane associated with coal was recognized to be a valuable natural energy resource. The commercial recovery of methane from coalbeds has since increased dramatically. The total U.S. coalbed methane production increased from 0.74 billion cu m (26 billion cu ft, BCF) in 1987 and 2.5 billion cu m (90 BCF) in 1989 to more than 9.2 billion cu m (324 BCF) in 1991, and 20.7 (730 BCF) in 1993 (...,1994). Interest in other countries - Australia, Canada, England and China - has been growing as well. Furthermore, there is also significant interest in recovery of gas from underground coal mining operations due to safety and environmental advantages. It is felt that the overall contribution of gas from coal seams will dramatically increase the worth's gas resource. The increase in coalbed methane production has led to the development of several computer models to simulate flow of gas in coalbed methane reservoirs and predict long-term production. According to King (1989), over thirty five distinct models have been established since 1958. These models '...differ by the assumptions used in the formulation, the degree of rigor used in the solution, and the capabilities considered by the model...'. However, a major hindrance to the effective use of any coalbed methane simulator is the unavailability of estimates of the required input parameters, and a knowledge about the variation in their values with time for recovery projections. Good examples of such parameters are permeability, diffusion coefficient, and cleat porosity of coal. These are, in turn, influenced by several factors like water content, in situ stress conditions and gas pressure, all of which change progressively during production, making the process of gas flow difficult to model. Due to the overburden and the associated stress, the permeability of coal - determined primarily by the cleats - is extremely low. It is generally accepted that the production from virgin coal seams is controlled by the permeability of coal. But gas recovery from coal mining operations, major production is from the gob, the area just behind the working face.
