Abstract A numerical investigation of the effect of pattern configuration on oil recovery by in situ combustion is presented. The patterns considered were inverted 5, 7, 9 and 13 spot. The data used for the base case were typical of the Athabasca area of Alberta, Canada. The sensitivity results are presented for changes in well density, pay thickness, initial oil saturation, oil viscosity, absolute permeability, water-oil ratio and air injectivity. All reslts are compared at the time of steam breakthrough.
The numerical results presented in this paper show that hexagonal (7 and 13 spot) patterns perform better than square (5 and 9 spot) patterns for most situations. As the oil mobility is reduced, the relative performance of square patterns tends to improve.
The results presented should be of value in the design of in situ combustion projects.
Introduction Interest in in situ combustion as an enhanced oil recovery process has been slowly increasing since the mid 1930's. Reviews by Nicolls and Luhning, Farouq Ali and Chu indicate that over one hundred field tests have been conducted with varying degrees of success. These reviews also indicate that a number of different patterns are being tried in the field, among them 5, 7, 9 and 13 spot inverted patterns are most common. Unfortunately, the literature does not provide clear guidance on the selection of pattern for given reservoir characteristics. Because of the variability of reservoir and fluid characteristics and the complexity of the recovery process, no single pattern is likely to be optimum under all situations. It is, however, useful to have some rough indication of the relative performance of various patterns for typical situations before a detailed study is attempted. Ideally comparisons of patterns should be based in field pilots, unfortunately this approach is virtually impossible due to the time, cost and technical difficulties involved. A state-of-the-art thermal simulator is the best tool available for conducting sensitivity and comparative studies. While studies of this type have been published for micellar/polymer flooding by Kazemi and MacMillan and steamflooding by Chu, no such study is available for in situ combustion.
In this paper we present a comparison of various patterns for in situ combustion when applied to a typical Alberta oil sand reservoir. The sensitivity of results to some important reservoir and operational parameters is also included.
The numerical model used in this study is described in detail by Rubin and Buchanan. It is a four-phase (oil, water, gas, coke), multicomponent, fully implicit simulator with appropriate features to conduct a study of this type. Reference 6 contains results of several validation runs for this model, in addition a comparison with a two-dimensional physical model of 1/6 of a 7-spot inverted pattern was also made in this study (Figure 1). All of these results show that the model being used is capable of simulating the in situ combustion process with appropriate selection of grid and other data. While a fine grid can be used to simulate laboratory experiments, the application of simulators to field studies makes it necessary to use grid blocks that are large compared to the thickness of the burning zone. Consequently, the narrow high temperature combustion zone is smeared over the blocks containing it. This leads to high sensitivity of results to grid type and size. In this study the grid was selected to minimize this effect.