Low permeability Coal Bed Methane (CBM) reservoirs are characterized by their low porosity and permeability, which in turn induce their complicated permeability characteristics of gas and water during production. Numerical experiments and field applications proved that in low permeability CBM reservoirs there exists percolation non-linearity and fluid muti-variability. The percolation of fluid needs to overcome threshold pressure gradient; while klinkenberg effects will restrict the gas permeability. Thus the classic percolation mathematical model is not valid for problems of low permeability gas reservoirs.
On this basis, a three dimensional, dual porosity, single permeability, non-equilibrium adsorption, gas-water two phase flow, pseudo-steady state mathematical model was developed, which reflects the concrete influence factors in low permeability CBM reservoirs, such as high velocity non-Darcy flow, threshold pressure and slippage effect. This was followed by a new theoretical formulation of permeability and porosity including the effect of matrix shrinkage.
This new model was solved by a fully implicit numerical method and the block pre-conditioning orthogonal minimisation algorithm. A computer programme called COAFOR has been developed for this purpose. Comparison of gas production rate between those from Eclipse software and the current programme was made. Three synthetic cases were studied, which proved that the developed algorithm works well for solving the problem stated above.
Those case studies have revealed that there exist three stages in coal de-watering and production processes. The existence of threshold pressure gradient can largely reduce the gas production rate, while the gas slippage factor can increase the bottom-hole pressure at the initial stage of production. The agreement between COAFOR and ECLIPSE under the same physical parameters was better than expected. Matrix shrinkage effect can largely increase the gas production rate and gas concentration owing to the enhanced permeability caused by the existence of matrix shrinkage.