Waterflooding can lead to substantial incremental oil production. Implementation of water-injection projects requires the project to fit into the risk (defined here as negative outcomes relative to defined project objectives) and uncertainty (defined here as the inability to estimate a value precisely) a company is willing to take.

One of the key risks for water injection into a shallow reservoir is injection-induced fractures extending into the caprock. If this risk is seen as “intolerable” in an as-low-as-reasonably-practicable (ALARP) analysis, a decision might be made not to proceed with the project. In this study, we evaluated caprock integrity by conducting simulations of long-term water injection that include the effects of formation damage caused by internal/external plugging, geomechanical stress changes, and fracture propagation in the sandstone and bounding shale.

The risk of fracture growth into the caprock was assessed by conducting Latin hypercube sampling considering a set of modeling parameters each associated with an uncertainty range. This allowed us to identify the range of operating parameters in which the risk of fracture-height growth was acceptable. Our simulations also allowed us to identify important factors that affect caprock integrity. To cover the uncertainty in geomechanical reservoir evaluation, the operating envelope is identified such that the risk to the caprock integrity is reduced. This requires introducing a limit for the bottomhole pressure (BHP), including a safety margin.

The limit of the BHP is then used as a constraint in the uncertainty analysis of water injectivity. The uncertainty analysis should cover the various development options, the parameterization of the model, sampling from the distribution of parameters- and distance-based generalized sensitivity analysis (dGSA) as well as probabilistic representation of the results. The results indicate that the time to reach the BHP limit varies substantially, dependent on the chosen development scenario. Injection of water (1000 m³/d), with total suspended-solids content ranging from 0.1 to 0.5 ppm by volume (ppmv) and particle size from 1 to 5 mm, into long horizontal wells (2000 m) results in injection times of more than 10,000 days even for the P10 percentile. However, injection of poor-quality water (injection rate 600m³/d, well length 600 m), with total suspended-solids content ranging from 0.5 to 5 ppmv and particle size from 10 to 30 µm, leads to the BHP limit of 10 (P10) to 740 (P90) days.

The dGSA can be used to determine which parameter has a stronger impact on the BHP and, hence, on the project, and should be measured if warranted by a value-of-information analysis. In the case reported here, dGSA showed that the filter-cake permeability has a big impact on the results and, hence, will be determined by laboratory measurements.

The final development option to be chosen depends on a traditional net-present-value analysis.