Formation compaction vs land subsidence to constrain rock compressibility of hydrocarbon reservoirs

Quantification of uncertainty is becoming increasingly important in any general modelling activity. In this study, the ensemble smoother, i.e., an ensemble-based data assimilation algorithm, is used to quantify and reduce the uncertainty associated with the geomechanical parameters of deep hydrocarbon reservoirs. The aim is at estimating the vertical uniaxial compressibility c(M) of the producing layers by assimilation of: (i) ground or seabed vertical and horizontal displacements measured with InSAR, multibeam surveys, and GPS; and (ii) reservoir deformation obtained from specific well logs (e.g., the radioactive marker technique) and extensometer stations. Usually subsidence measurements are characterized by large datasets (in both time and space) with a relatively low accuracy. Conversely, the compaction monitoring techniques provide more accurate measurements, although their availability is at limited points and over few time intervals. In this contribution, we test the capability of these two types of data to reduce the uncertainty associated to c(M) for a producing reservoir. Although dealing with a test case application, this investigation originates from the need of properly addressing and explaining the seafloor displacements observed over a real offshore gas field. The numerical tests are carried out with two different conceptual models for c(M), based on the common structure of gas fields. The first model considers a compressibility distribution varying with depth and effective vertical stress, but uniformly distributed within the reservoir. In this case, compaction measurements at the reservoir depth result very effective. However, when the reservoir is composed of several compartments bounded by faults and thrusts, the possible heterogeneity of c(M) among different blocks reduces the effectiveness of compaction measurements in data assimilation algorithms compared to that of surface displacements. (c) 2017 Published by Elsevier Ltd.