Carbonate-ramp reservoirs modelling best solutions: Insights from a dense shallow well database in Central Italy

Despite data coming from fully developed oil fields are fundamental to validate reservoir models, they are rarely publishable because of confidentiality. In this work, we benefit from an exceptionally dense public dataset represented by 43 wells logs drilled in a carbonate-heavy oil-rich reservoir of the Majella Mountain (Central Italy) over a grid of 200 m × 200 m, with depths in the range of 90–200 m. We tested different modelling solutions, to assess the best modelling approach that fits the available data on hydrocarbon distribution. Both deterministic (using Kriging) and stochastic simulations (using Sequential Gaussian Simulation- SGS) were tested. The role of the variogram resulted as a fundamental parameter for both simulation methods. Wells-derived variogram agrees with the size and orientation of carbonate dunes that drive the hydrocarbon distribution highlighting the importance of an accurate depositional model. Kriging was faster in terms of computation time and better in maintaining the lateral continuity of the layers, however, it excessively smoothed the results. SGS gave a better distribution of computed values but the lateral continuity was not preserved. By adding a main vertical trend derived from the upscaled logs to the SGS, results become more accurate. We then applied the best model to estimate the total volume of hydrocarbon in place in an already exploited area starting from just one well. The computed reserves show a good fit with historical production data demonstrating the reliability of the model. •We benefit from a public dense well dataset to test carbonate-reservoir models.•Deterministic solutions are faster but mostly inaccurate.•Stochastic simulation with vertical trend improves simulations of single well profile.•Variogram shape is crucial to improve the 3D stochastic simulation results.•Wells-derived variogram agrees with size and shape of carbonate dunes.