Impact of geological heterogeneity on CO2 sequestration: from outcrop to simulator

Abstract

Increased anthropogenic emission of carbon dioxide (CO2) into the Earth’s atmosphere since the industrial revolution has enhanced the greenhouse effect and contributed to global climate change. Controlling atmospheric CO2 emissions is thus essential to mitigate the environmental and socio-economic consequences related to these changes. Carbon capture and storage (CCS) was proposed as one possible option to control anthropogenic CO2 emissions, and is particularly viable at CO2 point sources such as coal-fuelled power plants. CCS was tested and applied globally in a variety of geological and top-side settings within the past decade, with varying success. In Longyearbyen, the main settlement on the Norwegian high-Arctic Svalbard archipelago, CO2 may be captured at the local coal-fuelled power plant and injected into an unconventional siliciclastic target aquifer. The target aquifer, within the Late Triassic to Middle Jurassic Kapp Toscana Group, comprises an up to 300 m thick sequence of tight, naturally fractured sandstones inter-bedded with siltstones and shales. During the Early Cretaceous, igneous intrusions, collectively classified as the Diabasodden Suite, were emplaced in the target aquifer. The pilot-scale Longyearbyen CCS project envisions only modest storage volumes of CO2, with the top-side CO2 storage requirements determined by the annual CO2 emissions from the local coal-fuelled power plant (c. 60 000 tons). As part of this PhD study, the geologically complex target aquifer was characterized and represented in a static geologic reservoir model. Fieldwork (e.g. structural and stratigraphic logs, geological mapping), borehole (e.g. drill core logs and plugs, wireline logs, water injection tests, vertical-seismic-profiling survey) and regional geophysical (e.g. 2D seismic, digital elevation model, magnetic data) data sets were used as input. Two main themes relating directly to the geological heterogeneity of the target aquifer were addressed in detail: (1) the natural fracture network, and, (2) the presence of igneous intrusions. Water injection tests, wireline logging and fracture mapping, in drill cores and at outcrops, all indicate that the tight heterolithic siliciclastic target aquifer is highly fractured, and that the pre-existing natural fracture network is critical for the injectivity of fluids. We integrated borehole (872 fractures measured along 302 m of drill core) and fieldwork data (7 672 fractures measured along > 1400 m of scanlines) to develop a conceptual model grouping reservoir intervals with similar mechanical and lithological properties into five litho-structural units (LSUs; classified as LSU AE). Fractures within the shale-dominated LSU A are predominantly low-angled and likely contribute to lateral fluid migration within the reservoir interval. In contrast, the predominantly high-angled fractures within the sand-dominated LSU C represent probable vertical intra-reservoir permeability pathways. Water injection tests indicate a linear flow pattern, particularly in the lower part of the aquifer (870-970 m depth). The orientation and configuration of the natural fracture network will thus ultimately control the migration direction and speed, and thus also the shape of the fluid plume. The highest overall fracture frequency is evident in LSU D (dolerite), but field observations suggest the majority of these fractures to be sealed by various types of cement (e.g. calcite), precipitated from percolating fluid in the transition zone between host rock and igneous intrusions. On a small scale, igneous intrusions form a contact metamorphic aureole in the surrounding host rock. This may significantly affect reservoir properties, even around relatively thin intrusions. We have studied such a thin (2.28 m thick) intrusion penetrated by the Dh4 borehole, and conclude that the total contact aureole is 160-195% the width of the sill itself. On a larger scale, igneous intrusions set up local-to-regional heterogeneities within the target aquifer, either as impermeable lateral to sub-vertical (sills and dykes, respectively) fluid flow barriers or as high-permeability pathways along fractured intrusion-host rock contact zones. We integrated numerous data sets to constrain the overall geometry of the igneous intrusions in Central Spitsbergen, and concluded that dykes and sills are the dominant geometries. Saucer-shaped intrusions were also mapped, but are located stratigraphically below the target aquifer. In general, igneous intrusions are most common in the lower one-third of the target aquifer, but in some cases dykes extend into the upper part of the aquifer, and even into the overlying cap rock. On a regional scale, igneous intrusions and sub-seismic faults are thus also likely to control the shape of the CO2 plume, along with the matrix properties and natural fracture network within the country rock. The current geological understanding of the unconventional target aquifer was incorporated into a scenario-based calculation of potential CO2 storage capacity. The wide range of low to high case (P90-P10) results, even for a given scenario with a deterministic areal extent, reflects the uncertainty attached to poorly constrained key parameters. These include the accessible segment size of the compartmentalized reservoir, the dominant CO2 phase at reservoir conditions and the storage efficiency factor. Reservoir simulations are required to constrain these parameters further. At this stage, the calculated storage capacity appears to be adequate to fulfil the stipulated requirement for the first phase injection of up to 200 000 tons of CO2. In summary, I present a collection of papers addressing the geological heterogeneity of an unconventional CO2 target aquifer on Svalbard. In addition, I use outcrop analogues of fracture corridors (from south-eastern Utah) and intrusion-host rock interfaces (from South Africa) to better understand processes acting on the target aquifer. This broad geological understanding is used to characterize the target aquifer, and is subsequently incorporated into a static geological model of the Longyearbyen storage site. The model may then be used as a base for extensive fluid flow simulations to optimize future well placement, injection rates and monitoring techniques. The learnings from this work can be applied directly to the Longyearbyen CO2 lab project, but may also be transposed as an analogue for storing CO2 in unconventional, naturally fractured reservoirs or even for producing hydrocarbons from similar geologic settings.