Methane hydrate (MH) is being highlighted as next-generation hydrocarbon resources mainly because of its huge in place and cleanness. The Research Consortium for Methane Hydrate Resources in Japan (MH21 Research Consortium), which was organized to attain the exploration and exploitation of MH resources, has been implementing a variety of research projects. As part of such research projects, we have been developing a state-of-the-art numerical simulator (called ‘MH21-HYDRES’) for rigorously predicting MH dissociation and production performances. The main functions of this simulator and the efforts toward improving and verifying this simulator are introduced in this paper.The gas production from MH reservoirs is significantly different from that from conventional oil and gas reservoirs in terms of the mechanism and the phenomena, since (1) MH is a solid, (2) reservoir behaviors are associated with the chemical reactions such as MH dissociation/formation, and (3) reservoir properties, especially permeability, change drastically by MH dissociation. Therefore, it is impossible to predict MH reservoir performances by conventional oil and gas reservoir simulators, which lead to the development of the own numerical simulator specialized for MH reservoirs.Currently MH21-HYDRES can be applied to three-dimensional Cartesian and two-dimensional radial coordinate systems. This simulator can also deal with six components of methane, water, nitrogen, carbon dioxide, methanol and salt, and with five phases of gas, water, ice, MH and (precipitated) salt. The main features of this simulator are to calculate the kinetics of endothermic dissociation and exothermic formation reactions of MH as well as multi-phase flow behaviors resulting from these reactions. The simulator divides a target reservoir into multiple grid blocks, for which the pressure, temperature, water saturation, methanol and salt mass fractions, etc. are calculated solving the system of discretized non-linear equations for the component mass conservation and the overall energy conservation.To shorten the computational time, it is preferable to reduce the total number of grid blocks and hence to increase the size of grid blocks. On the other hand, larger grid blocks result in more significant numerical errors, which is more serious in a MH simulator than in a conventional oil and gas simulator. To resolve these problems inconsistent with each other, the Dynamic Local Grid Refinement (DLGR) function was incorporated into the simulator. DLGR enables a shorter computational time without reducing the accuracy of prediction, by allocating fine grid blocks only to the regions of importance where MH is being dissociated/formed at every time step. In addition, we are attempting to parallelize MH21-HYDRES utilizing the published MPI (message passing interface) and the parallelized solver for a linear equation system. Although the parallelized MH21-HYDRES is still in the prototype stage, using four processors, we could successfully achieve the computational speed that is about three times faster in comparison with serial processing.The international code comparison for MH simulators is being held, led by National Energy Technology Laboratory and U.S. Geological Survey. We are participating in this comparison to verify the performances of MH21-HYDRES. Although other simulators in this project show the deficiency for complex and large scale problems, MH21-HYDRES could provide reasonable solutions from the view point of the stability and robustness as well as the accuracy of calculated results, which may manifest the successful coding of this simulator.
Read full abstract