CO2 interaction with aquifer and seal on geological timescales: the Miller oilfield, UK North Sea

Abstract

Carbon Capture and Storage (CCS) has been identified as a feasible technology to reduce CO2 emissions whilst permitting the continued use of fossil fuels. Injected CO2 must remain efficiently isolated from the atmosphere on a timescale of the order of 10000 years and greater. Natural CO2-rich sites can be investigated to understand the behaviour of CO2 in geological formations on such a timescale. This thesis examines the reservoir and seal on one such oilfield. Several hydrocarbon fields in the South Viking Graben of the North Sea naturally contain CO2, which is thought to have charged from depth along the western boundary fault of the graben. The Miller oil field which contains ~ 28 mol% CO2, of isotopic composition δ13C = -8.2‰. The Upper Jurassic Brae Formation reservoir sandstones and the Kimmeridge Clay Formation (KCF) seal have been exposed to the CO2 accumulation since its emplacement. Rock samples from the reservoir sandstone and bottom of the seal mudrock were examined using multiple techniques, including XRD, SEM, fluid inclusion and carbonate stable isotope analyses. The sandstones show no features directly attributable to abundant CO2 charge. SEM analyses reveal significant heterogeneities in diagenesis within the KCF. The silt/sand lithologies of the KCF have undergone a diagenetic history similar to that of the Brae Formation sandstones. In contrast, the KCF shales display a distinctly different diagenesis of dominant dissolution of quartz and feldspar with little evidence of mineral precipitation. In both the Brae Formation and the KCF, pore-filling kaolinite, illite and carbonates are relatively late diagenetic events which can be associated with CO2-induced feldspar dissolution. Mudrock X-ray diffraction mineralogical data reveal abrupt vertical mineralogical variations across the reservoir crest in the Miller Field, while such variations are absent in a low-CO2 control well in the same geological settings. This suggests that reactions induced by abundant CO2 dissolved feldspar and produced kaolinite, carbonates and quartz in the seal, while oil emplacement inhibited the reactions in the oil leg. However, petrographic evidence and comparison between different sections argue against CO2 reactions as the sole cause for such large mineralogical variations, especially for quartz. The vertical mineralogical variations to a certain extend represent original sedimentary heterogeneity. Linear variations of carbonate δ13C with depth were discovered in both shale and silt/sand lithologies of the KCF in a 12m zone immediately above the reservoir. These features are absent in the low-CO2 control well. These trends are interpreted as dissolution of original carbonates by CO2 slowly ascending from the reservoir. New carbonates precipitated from a carbon source with upwards decreasing δ13C due to mixing between three carbon sources with different C isotopes at systematically varying ratios. The isotopes in the reservoir and the bottom of the seal suggests initial CO2 charge at about 70-80 Ma. CO2 infiltration rate is estimated at about 9.8×10-7g·cm-2·y-1. Geochemical modelling was applied to reconstruct the reservoir fluid evolution by calibrating it to mineralogy, fluid chemistry, diagenesis and fluid inclusion data. The modelling suggests that CO2 migrated into the reservoir together with a saline basinal fluid derived from the underlying evaporites at ~ 70 Ma. The CO2 and basinal water charge imposed an important influence on the mineral reactions and fluid chemistry. This study suggests that the KCF has formed an excellent CO2 seal, with no substantial breach since its charge at 70-80 Ma.