The extent and size of the fracture network connected to the wellbore are among the most uncertain parameters for multi-fractured horizontal wells. The uncertainty has a profound impact on production under primary depletion and has even a bigger impact on the recovery during huff-n-puff (HnP) gas injection. In this paper, we quantify the uncertainty in the enhanced oil recovery (EOR) for a pair of wells in the Permian basin, as well as report the lessons learned from modeling HnP EOR using reservoir simulation.
We create a base sector model for two wells completed in Wolfcamp B and C formations, where fractures are incorporated using embedded discrete fracture model. For each well, we identify 13 uncertain parameters (fracture properties and compaction, initial conditions, and relative permeability) and use a Bayesian assisted-history-matching (AHM) method to match primary production data. We shed some light on two issues that impact the recovery significantly. One issue is the grid refinement; we create three models with varying grid size. We study the effect of gas molecular diffusion and report the added recovery using different diffusion coefficients. The third issue is the size of the fracture network, which is dynamic as suggested from evidence of communication between the wells. Moreover, the fracture area that becomes accessible during injection is bounded by a large uncertainty range. We enable the inter-well interference by activating an additional fracture network during injection. We create four scenarios within the range in the literature (approximately 0.5, 1, 2, and 4 million ft2/stage) and their properties are tuned to match the historical HnP data. Production from the AHM solutions is extended beyond the first four HnP cycles to yield probabilistic predictions.
We show that grid refinement has a small effect on the production under primary depletion, but it has a very large impact on HnP recovery. We explain that numerical dispersion causes artificial mixing of gas in the large grid blocks, leading to artificial optimistic recovery. Molecular diffusion of gas in the liquid phase has negligible impact on the recovery. The contribution becomes appreciable only if diffusion coefficients greater than 10-3 cm2/s are used, which is far greater than the reported values for diffusion of the considered hydrocarbons. We conclude that molecular diffusion is of 2nd order importance compared to advection and should be neglected. The contacted-fracture area is emphasized as an important factor that is usually set too low in the reservoir-simulation practice. Production for 37 filtered AHM solutions is extended under the four fracture-area scenarios. Based on our new analysis, we report that the HnP results in an average of 76% improvement in oil recovery compared to natural depletion.
