SEQUENCE STRATIGRAPHIC EVALUATION AND PRODUCTION CHARACTERISTICS OF THE EOCENE MIRADOR FORMATION, CUSIANA FIELD, EASTERN COLOMBIA

Abstract

ABSTRACT The Cusiana Field (BP, Ecopetrol, Total and Triton) comprises relatively sand-rich reservoirs that range from late Cretaceous, passive margin deltaic to early Tertiary, early foreland basin estuarine-fluvial. The key reservoir is the Eocene Mirador Formation which comprises 60% or more of reserves. Currently (December 1996) the Mirador reservoir is providing nearly all of the 180,000 barrels of oil per day (bopd) of production from the Cusiana Field. During 1997 production is expected to rise to in excess of 300,000 bopd. The majority of this increased rate will be derived from additional Mirador producing wells. Due to the current importance of the Mirador Formation to Cusiana production and reservoir management, it has been the focus of detailed reservoir description and modeling and is the subject of this short paper. The Mirador Formation is between 340 and 460 feet in thickness (true stratigraphic) and comprises ~60% net effective sandstone facies; 4% or greater porosity and 0.1 millidarcy (mD) or greater horizontal permeability. The reservoir description of the Mirador Formation recognizes sandy incised valley fill deposits as the key reservoir components. The interpretation of incised valleys is based on an extensive core data set and biostratigraphic analyses. Valleys comprise 50% of the Mirador stratigraphy and therefore are the principal producing units in the reservoir. Reservoir pressure data collected during early appraisal and prior to production indicated all Mirador valley sandstones are well connected both vertically and laterally across the Cusiana Field. After the onset of production in 1992, subsequent pressure data confirmed good field wide connectivity of valleys, but also indicated that the Mirador Formation comprises two major compartments. In addition to the pressure data, production logs have shown that single valleys commonly behave as strongly layered systems. Both these production scale features of the Mirador reservoir owe their origin to key stratigraphic surfaces and associated textural variations in the reservoir. The two major compartments in the Mirador are separated by a major transgressive surface that was recognized in cores during the earliest stages of the reservoir description. More recent quantitative biostratigraphic analyses have shown this important stratigraphic surface to also be a major unconformity that separates the early Eocene Lower Mirador from the late Eocene/early Oligocene Upper Mirador. This important stratigraphic surface marks fundamental changes in the nature of the Mirador sedimentology and reservoir fabrics. The dramatic vertical changes in flow behavior within single Mirador valleys indicated by production logs occur across intra-valley flooding surfaces. Both the major unconformity in the Mirador reservoir and the intra-valley stratigraphic surfaces have implications for the reservoir management of the Cusiana Field. Accurate characterization of the major stratigraphic controls on production continue to contribute to predictions about field behavior during early field life. The reservoir description and characterization process in the Mirador reservoir is a blend of deterministic and stochastic modeling techniques. Reservoir layers for the static full field model are constrained by the sequence stratigraphic interpretation of the reservoir. This deterministic approach allows the general connectivity of valley sandstones to be strongly controlled in the modeling process and also makes the most of the valuable vertical data provided by the wells. Stochastic techniques, allied with empirical field data and analog information, are used to model widths, orientations and other facets of producing sandstones that are incompletely understood. Static models are upscaled into the dynamic simulator and history matched to evaluate various development options. The process of history matching each static realization lends insights to the prediction of reservoir properties, and is used interactively to refine the static prediction.