In conventional reservoirs with bottom-water and no natural fractures, recovery can be predicted from known designed well spacing and completion placement. However, in naturally-fractured reservoirs (NFR), there is a considerable intrinsic uncertainty of such predictions because well’s completions may randomly intercept fractures (fracture-well) or be accidently placed in the matrix blocks (matrix-well) of NFR. Moreover, the recovery prediction is even more difficult when dual-completed wells with downhole water sink (DWS) are used for water coning control — with a second completion placed in the aquifer beneath the oil-water contact. The study compares recovery of the fracture and matrix-wells in the same NFR developed using single-completed and dual completed wells. The study also proposes a probabilistic method for finding well’s recovery in NFR having distributed fracture network and uncertain wells’s location in the network.
Two patterns of fracture network are considered: densely-distributed natural fractures, and sparsely-distributed fracture clusters. DPDP Cartesian flow model is verified for estimation of recovery performance in each of the two networks using water cut patterns and ultimate recovery estimates. The model proves to be sufficiently accurate to be used in this study. The study considers ultimate recovery after 20 years of production, or when the water-cut value is 97%, for known or unknown well completion placement in the fracture network. In the latter case (uncertain completion placement), expected value of well’s recovery is computed for the (known) fracture corridor size distribution from seismic surveys and borehole image logs. The distribution also provides the fractional area of NFR occupied by the fracture corridors and exclusion (matrix) zone, which helps estimate the probability of on/off-fracture completion.
In this study, we consider two (on-fracture/off-fracture) locations of well completion for conventional and DWS wells. When the location is certain (known) and fractures are densely distributed, placement of well’s completion has no effect on recovery performance. However, our results also show that in the fracture corridor network, single completed matrix wells are superior to fracture wells. Similarly, DWS wells dual-completed in a exclusion-zone would maximize the recovery.
Probabilistic assessment of recovery is demonstrated for a field case NFR with known size distribution of fracture corridors. In case the reservoir is developed with single-completed wells the expected value of recovery factor is 33% with 42% and 58% of wells randomly completed in fracture corridor and exclusion-zone respectively. When DWS wells are used, probabilistic placement of two-completions is the same but oil recovery increases to 41%. The study demonstrates the need and a method for probabilistic assessment of recovery from bottom-water NFRs with sparsely-distributed fracture networks using conventional and dual-completed wells.