Well Productivity Decline During Oil & Water Production due to Fines Migration

Summary The aim of this work is to characterise well productivity decline during two phase production due to fines migration. Modelling is performed separately for homogeneous and heterogeneous reservoirs, allowing for quantitative predictions across a range of applications. In homogeneous reservoirs, progressive flooding of individual pores drives particle detachment, leading to permeability decline. Utilizing the pore size distribution allows for explicit modelling of the distribution of phases and the progressive formation damage during water-cut increase. For heterogeneous reservoirs, we extend the Dietz model for layer cake reservoirs to develop pseudo relative permeability functions that account for the formation damage occurring as water encroaches into the production zone. The Dietz model relies on the permeability distribution of the reservoir; however, a technique is developed to predict formation damage without requiring this information. The two models are compared with production data from two Malaysian fields. The result of both models are explicit formulae describing the decrease in well productivity as the water-cut increases. By presenting the data in a new diagnostic plot, impedance vs water-cut, the formation damage characteristics of the reservoir can be determined simply, and further calculations performed. Comparison with field data from Malaysian fields shows good agreement, validating the models. The models allow for straightforward extrapolation to higher water cuts for predictive analysis and well design, including the planning of well stimulation or remediation in cases of high formation damage. The heterogeneous case produces a linear curve on the new diagnostic plot, which is shown to be general for all permeability distributions. The models for the first time resolve quantitatively the widely reported phenomenon of formation damage due to fines migration occurring during periods of increasing water-cut. The models allow for history matching and predictive analysis.