AAPG Bulletin, Vol.91, No.6, 847-876, 2007
Compartmentalization and time-lapse geochemical reservoir survelance of the Horn Mountain oil field, deep-water Gulf of Mexico
Oil is produced at the Horn Mountain field (Gulf of Mexico, Mississippi Canyon blocks 126 and 127) from middle Miocene reservoirs M and J. Reservoir facies are characterized as sand-filled channels and associated overbank deposits and are positioned in combination structural and stratigraphic traps. Prior to initial production, several barriers and baffles were identified in both reservoirs by integrating geological, geophysical, petrophysical, pressure, PVT (pressure-volume-temperature relationships), and geochemical data and petroleum-filling history. A compartmentalization risk matrix was developed to facilitate and visualize the integrated evaluation of compartmentalization. During production, in addition to traditional surveillance technologies, we applied time-lapse geochemistry (TLG) to visualize petroleum sweep by monitoring changes in fluid composition and fingerprints across reservoirs. In this technology, appraisal and preproduction fluid samples are first analyzed to map fluid types across a static reservoir. Then, a surveillance program in which fluid samples are taken from producing wells at regular time intervals is designed and executed. The obtained production samples are geochemically fingerprinted and compared with preproduction fluids from the same well and surrounding wells. At Horn Mountain, interpretation of geochemical data allowed us to infer oil movement across reservoir M and helped to reevaluate reservoir models and reduce risks in managing reservoir performance. In reservoir J, an untapped compartment was identified, and an additional producer was justified for future drilling. Time-lapse geochemistry results were consistent with and complimentary to other surveillance data available to date. Our study demonstrates that TLG is a safe and cost-effective technology, which reduces uncertainties associated with other reservoir surveillance methods and appears to be valuable for reservoir management.