Preliminary Discussion of Net/Gross Distribution in Gulf of Mexico Deepwater Reservoirs

Shanley, K.W., presently with Basin Exploration, C.J. O'Byrne, R.R. Fisher, and T.M. Smith, BP Amoco, Houston, (Presently with Veritas DGC, Houston)

Abstract

In recent years, exploration and development portfolios have increasingly shifted towards deepwater reservoir systems, the majority of which are physically located in present-day deepwater settings. Because of their deep water setting, decisions regarding substantial capital expenditure are often required on the basis of 3D seismic coverage and extremely limited well data. The drilling of multiple appraisal wells to accurately delimit reservoirs often results in unacceptable erosion of financial value. These operational and financial constraints make early probabilistic assessments of project value increasingly important in decision analysis.

A critical component of any analysis of project value is an understanding of the distribution of reservoir parameters including reservoir net/gross. In an attempt to quantify variations in reservoir quality and to supplement seismic attribute analyses, cumulative probability distributions of reservoir net/gross, based on well penetrations, have been compiled for a suite of deepwater seismic facies. This approach of calibrating distinctive seismic facies provides a tool useful for risk assessment in both early phases of exploration as well as development planning. We recognize 10 seismic facies comprising 6 major facies classes in deep water deposits in the Gulf of Mexico. These seismic facies represent discrete 'elements' that share similar, observable (on seismic data) geometric characteristics. Similar to rock-based facies, these seismic facies elements form the building blocks of larger facies assemblages or associations. In this paper we present preliminary distributions for two common seismic facies - "Facies C/Sheets" and "Facies B/Concave-erosional". Each of these facies can be subdivided into two sub-facies, C1/C2 and B1/B2. Examination of the distributions of these data suggest that minimum estimates of net/gross (P₁₀ value, i.e., 90% of all data have a net/gross in excess of P₁₀) range from approximately 25% (C2) to 55% (C1), most likely estimates of net/gross (P₅₀ value, i.e. 50% of all data have a net/gross in excess of P₅₀) range from approximately 72% (C1) to 86% (B1), and maximum estimates of net/gross (P₉₀ value, i.e. 10% of all data have a net/gross in excess of P₉₀) range from approximately 92% (C2 and B2) to 97% (C1). These values vary somewhat from other data sources, notably Deepstar (1996) and Prather et al., 1998.