Pre-development fracture modelling in the Clair field, west of Shetland

Abstract

The Clair oilfield is a large fractured sandstone reservoir lying 75 km west of Shetland on the UK continental shelf. Fracture analysis and modelling was carried out in preparation for the phase 1 development, which started production early in 2005. Fracture clusters and discrete fluid inflows observed in wells are associated with faults and other localized deformation features tens or hundreds of metres apart. The reservoir has moderate to good matrix permeability, but well flow rates and profiles are fracture-dominated. Full-field geological models were built using conventional object modelling approaches for matrix and discrete fracture networks for fractures, and upscaled to populate a reservoir simulation grid. Dual-porosity, dual-permeability dynamic modelling (full-field and well-test) was undertaken to understand the fracture and matrix flow contributions and their interaction. Fracture models were conditioned to wells and to seismic data, including coherency and multi-azimuthal velocity information from a four-component, ocean bottom cable three-dimensional seismic survey. At this early stage in field development, there is insufficient calibration to select a single fracture model. Instead, well and depletion plans have been tested against multiple fracture models chosen to encompass a wide range of plausible outcomes.

Structurally complex reservoirs: an introduction

“…Obviously that requires a meaningful definition of 'anomalously low', and may not be possible where there are large matrix velocity variations due to fluid or lithology effects. Fracture predictions based on seismic anisotropy and coherency-like attributes have been presented by, for example Bloch et al (2003) and Barr et al (2007). Calibration against core or image log data is advisable (e.g.…”

“…This scenario almost certainly applies to stratabound joints even in circumstances where they have a component of shear displacement and may display some fractal characteristics (e.g. Odling et al 1999;Barr). Finally, the mapped fault tip locations are themselves limited by seismic resolution.…”

“…This leads to the interesting observation that 'how fractured' a field appears depends not just on the conductivity of its fracture network, but also on that of the matrix properties. For example, the West Sole field (Barr) behaves more like a 'classical' fractured reservoir than does the Clair field (Barr et al 2007), despite having an effective fracture permeability an order of magnitude smaller. That is because it has two or three orders of magnitude less matrix permeability.…”

“…Fractured reservoirs typically have very heterogeneous porosity and permeability distributions (Matthäi et al), which result in characteristic patterns of well performance with most production coming from the best few wells (e.g. Nelson 2001;Barr). The worst wells have failed to intersect connected, conductive fractures and a large financial benefit would flow from an ability to target only the best well locations.…”

“…(b) Models of the fractured Clair reservoir, UKCS ( fig. 12 of Barr et al 2007). On the left are multiple stochastic realizations of a discrete fracture (conductive fault) network.…”