Abstract
Accumulated field empirical observations suggest that water injected to displace heavy oils forms in the reservoir channel-like communication paths from the injectors to the producers. The evidence comes from mass balances and, more recently, from 4D seismic monitoring of heavy oil waterfloods. The reasons for this are multifold, including unconsolidated sand formation dilation about injectors due to slow pressure diffusion in heavy oils, reservoir heterogeneities in permeability and saturation, sand production from the reservoir, and the instability of the displacement interface due to the high mobility ratio between water and heavy oil. Once formed, the channels can degrade further economic recovery of the heavy oil as the water oil ratios increase significantly. This study reports on initial results from a laboratory program to test the optimal reservoir management response upon formation of such communication paths in heavy oil waterfloods.
To physically simulate reservoir waterflood behavior under the existence of a communication path, a large scale ‘big can’, five feet long with a ten inch by ten inch cross section, was designed and constructed that allowed for the creation of a highly reproducible communication path from the injection to production end of the can. This was a mandatory requirement for accurate comparison between alternative reservoir management strategies whose differences would otherwise be hidden by variations in random communication path formation. The design has proven to be highly successful. Our first objective was to test whether the industry paradigm and the regulatory mandated practice of maintaining a voidage replacement ratio (VRR) of one throughout the entire waterflood is optimal. Live 18.6 API Alaska North slope oil was used to saturate four Darcy sand that filled the big can. Upon creation of the communication path, three VRRs were tested: 1.0 (conventional waterflood), 0.7 (hybrid waterflood/solution gas drive), and 0.0 (conventional solution gas drive). The VRR=0.7 run outperformed the conventional VRR=1.0, suggesting that periods of under injection may improve heavy oil waterflood response upon formation of injector-producer communication.
