Innovative Approach for Building and Calibrating Multiple Fracture Network Models for Fractured Carbonate Reservoirs

Abstract

AbstractThe productivity of wells drilled into naturally fractured reservoirs is often unpredictable due to the highly heterogeneous nature of the reservoir. In such reservoirs, productivity is a function of complex processes involving the density and aperture of fractures, their interconnectivity and the quality of the surrounding rock matrix. For reducing the risks associated with drilling of new wells for field development, one approach consists in creating multiple equiprobable reservoir models constrained to both static and dynamic data. We further propose to associate with each of these models a synthetic indicator, represented by a single scalar 3D property accounting for modelled fracture properties and correlated with the expected productivity of future wells. This workflow was applied to a group of fractured carbonate reservoirs in Kuwait which helped in designing new horizontal wells.The developed methodology focuses on capturing accurately the lateral and vertical heterogeneity of the fracture network. Image log data analysis shows that the vertical changes of fracture density are dominated by facies variations, whereas areal contrasts are primarily related to the location of the main faults and folds. In some of the modeled scenarios, the orientation of secondary fractures, developed during the latest tectonic phases, is assessed using paleo-stress modeling. Fracture length parameters are inferred from the comparison between percolating volumes extracted from detailed connectivity analysis of discrete fracture networks and qualitative well test interpretations. Fracture aperture and permeability are finally calibrated against production history in small scale sector models. Once the initial sectors have been prepared by a reservoir engineer, this "big-loop" calibration process, which involves modifying the geological fracture model, upscaling it and running the flow simulation, is performed directly by the geologist, ensuring the consistency of the final simulation model. For better understanding the impact of matrix properties and fracture porosity, permeability and shape factor on well productivity, thousands of simulations are run on a small-scale synthetic model, yielding a multi-dimensional lookup table giving the potential productivity as a function of the fracture network characteristics.Six different scenarios involving changes of fracture densities, orientations and aperture are explored. These scenarios are consistent with fractures interpreted on image log data, with the major structural elements and with historical production data. Using the big-loop calibration, most of the history matching work is performed in the static model. The planning of new wells was then achieved by targeting zones which have been identified as highly productive in all the modeled scenarios.Innovative techniques are used for decoupling the vertical and lateral heterogeneity modeling for deducing fracture length from percolation analysis and for performing aperture calibration to production data. The methodology we propose for computing a productivity indicator enables rapid assessment of the flowing potential of fractured reservoirs.