Clair field came on production in February 2005. Clair is a large oil reservoir (>4 billion bbl STOIIP), 75 km west of Shetland. It is the largest naturally fractured reservoir developed in the UK, and is recovered by waterflood. Its 28 year appraisal period reflects the high reservoir complexity, relatively poor quality conventional seismic image, and uncertain impact of the conductive fractures.

This paper compares early Clair production with pre-drill expectations. Its primary focus is on reservoir uncertainty, and how the drilling sequence and data acquisition are managed to allow rapid and flexible response to learning, given an expectation of surprises. Clair is not pre-drilled, and the preferred well locations and design will change. A condition of responding rapidly is a hopper of well options, kept filled with diverse well types and locations, from which appropriate wells are selected using current knowledge. Key decision points in the sequence are identified in advance and options at each point evaluated against reservoir scenarios. This process allows planning for potential outcomes, generates the required hopper wells and identifies data requirements for future decisions.

Clair well-planning requires integration with reservoir surveillance and modelling. Early surveillance focuses on reservoir pressure response. All wells (including injectors) have a permanent down-hole pressure gauge, RFT data acquisition and a baseline production log. These data establish reservoir connectivity, injection response, and controls on productivity and injectivity. Also important is seismic data calibration against log and core information, to refine models of the natural fracture distribution. Water movement through fractures is expected and will become a surveillance focus.

Clair reservoir uncertainty is represented in sets of static and dynamic reservoir models which express the full uncertainty range consistent with data. At start-up, a wide range of fracture models, with diverse outcomes, can be consistent with the limited dynamic data. As new data arrive, the objective is to condense the range of models, which continue to be diverse, but must also be consistent with all historic data at the end of every month. These models inform drilling programme decisions, reflecting both the continuing uncertainty and rapid learning.

You can access this article if you purchase or spend a download.