Calibration of Fractured Reservoirs with Dynamic Data

Abstract

Abstract Fractured reservoirs are reservoirs whose productivity and performance are controlled or strongly affected by the fracture network occurring in the field. Therefore, a correct characterization of the reservoir internal geometry and a suitable description of the effective fracture network are crucial for a predictive model to be reliably applied to performance prediction and production forecast. A new approach has been attempted to validate the reconstructed internal geometry of a selected fractured reservoir by reproducing the dynamic behavior of the reservoir monitored during MDT tests. The description of the reservoir fracture network was achieved by integration of relevant data that could be collected from wireline logs, especially imaging and sonic logs, conventional cores, specifically aimed at microfracture examination, small drilling mud loss analysis, and field scale observations of regional structural heterogeneities from outcrop analogs inspection. Fracture types, distribution, geometry, properties, interaction with matrix, and deformation history were thus defined. Fracture patterns were described in terms of distributions for orientation, size, shape, spatial location and intensity to stochastically generate a static model of the fractured reservoir. The dynamic behavior of the fractured system was reproduced by a finite element flow model able to simulate transient flow through three-dimensional rock masses with discrete fracture networks. Consistency was required between the observed and simulated pressure data also in terms of derivative pressure trend to assure that the reservoir geometry was adequately modeled. Analysis of the model response as a function of the assigned fracture parameters and comparison between the observed and simulated dynamic behavior allowed calibration of the fracture modeling parameters and achievement of a satisfactory description of the reservoir effective fracture network. Introduction An application of the most recent methodologies developed to characterise fractured reservoirs was attempted to generate a realistic model of the fracture network allowing hydrocarbon production from an otherwise low porosity and low permeability formation(1,2). The investigated reservoir is an oil-bearing fractured formation, mainly made up of massive, unstratified, tight calcareous dolomite, and its thickness at the well is in the order of 400 meters. The oil is strongly undersaturated at initial reservoir conditions and the oil density is 32°API. On the basis of all the available information a stochastic fracture distribution in the reservoir region surrounding the well at which all the data had been collected (Well A) was generated. The fracture pattern and aperture were statistically defined by integration of data from imaging logs recordings, conventional core analysis, drilling mud loss interpretation, and observations on outcrop analogs. Only the properties of the main fractures intercepted by the well bore were deterministically assigned to the model. Imaging log and cores from another nearby well (Well B) were also considered to verify the consistency on the fracture pattern characterization. Upscaling of the microfracture distribution observed at the wellbore and core scale was required to generate a representative model of the formation. A finite element model of the fracture network was then generated to properly describe the fluid flow in the reservoir whereas the flow in the matrix was simulated according to a generic matrix block approach(3,4). Simulations of the model dynamic behavior were performed to reproduce the pressure response recorded during the MDT tests. Comparison between simulation results and measured pressure data allowed verification of the model consistency and calibration of fracture network intensity and fracture permeability.