Integrated Multi-Well Reservoir and Decision Model to Determine Optimal Well Spacing in Unconventional Gas Reservoirs

Optimizing well spacing in unconventional gas reservoirs is difficult due to complex heterogeneity, large variability and uncertainty in reservoir properties, and lack of data that increase the production uncertainty. Previous methods are either suboptimal because they do not consider subsurface uncertainty (e.g., statistical moving-window methods) or they are too time-consuming and expensive for many operators (e.g., integrated reservoir characterization and simulation studies). This research has focused on developing and extending a new technology for determining optimal well spacing in tight gas reservoirs that maximize profitability. To achieve the research objectives, an integrated multi-well reservoir and decision model that fully incorporates uncertainty was developed. The reservoir model is based on reservoir simulation technology coupled with geostatistical and Monte Carlo methods to predict production performance in unconventional gas reservoirs as a function of well spacing and different development scenarios. The variability in discounted cumulative production was used for direct integration of the reservoir model with a Bayesian decision model (developed by other members of the research team) that determines the optimal well spacing and hence the optimal development strategy. The integrated model includes two development stages with a varying Stage-1 time span. The integrated tools were applied to an illustrative example in Deep Basin (Gething D) tight gas sands in Alberta, Canada, to determine optimal development strategies. The results showed that a Stage-1 length of 1 year starting at 160-acre spacing with no further downspacing is the optimal development policy. It also showed that extending the duration of Stage 1 beyond one year does not represent an economic benefit. These results are specific to the Berland River (Gething) area and should not be generalized to other unconventional gas reservoirs. However, the proposed technology provides insight into both the value of information and the ability to incorporate learning in a dynamic development strategy. The new technology is expected to help operators determine the combination of primary and secondary development policies early in the reservoir life that profitably maximize production and minimize the number of uneconomical wells. I anticipate that this methodology will be applicable to other tight and shale gas reservoirs