| A fully implicit, three-dimensional, three-phase, compositional steam injection simulator has been developed in this study. Newton's iteration method has been employed to solve the set of nonlinear equations in this model. The entries to the Jacobian matrix are calculated analytically, and the constraint equations are eliminated at the matrix level. The model can operate for both block-centered and point-distributed grid systems. Apart from the natural ordering scheme, the D4 ordering scheme has also been implemented. To accomplish the versatility, the model has also been written in such a way that it can handle reservoir heterogeneities and irregularities. The well model is capable of simulating multi-block vertical, directional and horizontal wells with either bottom-hole pressure or flow rate specification.; The major objective of this study is to investigate the reservoir performance under steam injection using different well types and configurations. The first part of the investigation shows that recovery by the steam stimulation technique is about 15-20% of the recovery gained by the steam flooding technique, provided that the same pattern and number of wells are used in both cases. The steam flooding using horizontal wells yields the recovery of about 10-15% higher than those yielded by the operation using vertical wells in homogeneous reservoirs when the horizontal wells are located at the lower portion of the reservoir formation. The steam chamber technique is also promising; it has as much as 10% higher recovery than that obtained by the conventional steam flooding. However, in heterogeneous reservoirs with low vertical permeabilities, the vertical-well operation may excel the operation by the steam chamber method. Although it has never been reported in field practice, steam stimulation using horizontal wells seems to be preferable when there is no disadvantage in vertical flow in a reservoir. The optimum location of the horizontal well in the steam stimulation technique is one quarter of the reservoir thickness away from the reservoir bottom boundary.; The study of reservoir performances using different methods of heat loss modeling is also incorporated in this report. The analysis shows two groups of heat loss models; the first group gives instantaneous rate of heat loss, whereas the second one gives average rate of heat loss over the time step. This study recommends the use of the analytical solution with numerical approximation of the convolution integral for its accuracy, CPU consumption and storage requirement. |
