CO₂ storage retrospective: Sleipner from 1996 to today

The site and the storage geology

• Location: the Sleipner area of the Norwegian North Sea, Block 15/9, operated by Equinor with its partners.
• Storage formation: the Utsira Sand, a Mio-Pliocene sand more than 400 km long from north to south and 50 to 100 km wide, about 26,000 km² in all. At Sleipner it is 200 to 300 m thick, its top about 800 m below sea level; it is clean, fine-grained and largely uncemented, with porosities of 0.30 to 0.42 and permeabilities of about 1 to 8 darcy (Chadwick et al. 2005; Chadwick and Eiken 2013).
• Top seal: a basin-restricted mudstone 50 to 100 m thick at the base of the Nordland Group, extending more than 50 km west and 40 km east of the plume, beneath about 700 m of overburden in all (Chadwick and Eiken 2013).
• Internal architecture: thin mudstones, typically about 1 m thick, and a thicker, laterally persistent “five-metre mudstone” near the top divide the sand. They are barriers to flow: the rising CO₂ collects beneath them in thin layers.
• Injection: a single deviated well, nearly horizontal where it is perforated, 1010 to 1013 m below sea level, close to the base of the sand and about 200 m below its top. The CO₂, with 1 to 2 per cent methane, goes in as a dense phase; the reservoir is at about hydrostatic pressure, about 29 °C at its top and 35.5 °C at the injection depth (Chadwick et al. 2005; Chadwick and Eiken 2013).
• Monitoring: a 3D seismic baseline in 1994 and repeat 3D surveys in 1999, 2001, 2002, 2004, 2006, 2008 and 2010 (Chadwick and Eiken 2013), then 2013 and 2016 (Furre et al. 2017) and 2020 (Furre et al. 2024); seabed gravity in 2002, 2005, 2009 and 2013; a high-resolution 2D survey in 2006, a controlled-source electromagnetic line and seabed imaging. There are no downhole measurements: the well lies beneath the plume, which keeps it clear of the CO₂ and the CO₂ out of its reach (Chadwick and Eiken 2013).

Exercise: read twenty years of surveys in Figure 9.6

1. Start at the 1994 baseline and step the survey to 1999, with the CO₂ mixed through every pore and 72 per cent of it in the layers. Nine bright reflections appear in (b) where the sand showed almost nothing, and the base of the sand sags beneath them: the model pushes it down by up to 61 ms above the injection point. Chadwick et al. (2005) mapped a push-down widely over 20 ms and locally over 40 ms in 1999. The model’s peak is higher because it stacks all nine layers over the injection point, where the real plume typically has five to seven at any one place, and puts all 28 per cent not in the layers into dilute CO₂, where Chadwick et al. found about 13 per cent; with 13 per cent dilute (a share of 87 per cent in the layers) the model gives 46 ms.
2. Change how the CO₂ shares the pores to patchy. The mass and its place are unchanged, but the push-down falls to 25 ms, and read with the uniform law it would imply only 0.7 Mt instead of 2.35 Mt. Plate (c) shows why: the dilute CO₂, at $S_{\mathrm{d}} = 0.066$, slows the sand a great deal when it is mixed through every pore and hardly at all when it is held in patches.
3. Back on uniform, move the share of the CO₂ in the layers to 100 per cent. The push-down falls from 61 ms to 25 ms with the mass unchanged: volume for volume, dilute CO₂ is the more effective push-down agent (Chadwick et al. 2004, 2005).
4. Set the share back to 72 per cent and step to 2016. More than 16 Mt had been injected, yet the model’s largest push-down has grown only to 74 ms: its symmetric stack fills the column above the injection point early, and the plume then grows sideways. The real push-down grew far more between 1999 and 2001, locally to over 60 ms (Chadwick et al. 2005). Four of the nine layers are now thicker than the 7.9 m tuning thickness in (d).
5. Switch (b) to the difference. Beneath the plume the difference is strong although no CO₂ is there: the push-down has moved every deeper reflection, which is why time-lapse processing measures time shifts as well as changes in amplitude.
6. Lower the peak frequency to 20 Hz. The tuning thickness rises to 13.8 m, and the layers blur together in (b). Every layer is now below tuning, where amplitude rises with thickness, but reading a thickness from it still needs the wavelet and the layer’s velocity.

What the monitoring has shown at Sleipner

More than twenty years of surveys have answered some questions firmly and left others open:

• The CO₂ has stayed in the Utsira Sand. Difference volumes through the caprock and overburden show only repeatability noise, with no systematic change over the plume, and seabed imaging and ROV video found no seepage (Chadwick and Eiken 2013). No migration out of the reservoir has been detected.
• “Not detected” has a size. The surveys are estimated to detect an accumulation of about 4000 m³ of CO₂, about 2800 t at the top of the reservoir and less at shallower depths, where CO₂ is lighter and more reflective (Chadwick 2010, in Chadwick and Eiken 2013). A clean survey rules out accumulations above that size, not every tonne.
• Where the CO₂ goes. It rises from the injection point through a chimney and spreads beneath the mudstones in up to nine thin layers, mostly thinner than 8 m, the topmost, under the caprock, growing fastest. The plume is about 200 m high; in plan it grew from about 1.8 km long in 1999 to about 4.5 km by 2010, with a northward extension since 2004 (Chadwick et al. 2005; Chadwick and Eiken 2013). By spreading the CO₂ through more rock, the mudstones promote its dissolution in the brine (Furre et al. 2024).
• Flow models match the shape, not the mechanism. A general match of the plume’s shape and growth is readily obtained, but whether the CO₂ crosses the mudstones because they are semi-permeable or through holes is still uncertain: both kinds of model reproduce it (Chadwick and Eiken 2013).
• The mass is the hardest number. The 1999 data were matched with a saturation model holding about 85 per cent of the injected CO₂, but only with low-saturation CO₂ between the layers that flow models find hard to explain, and the later surveys, with the deeper plume’s reflectivity fading, are harder still to model. The push-down integrated over the plume and the summed amplitudes do grow almost linearly with the injected mass, an empirical relation (Chadwick and Eiken 2013). Figure 9.6 shows why the conversion is uncertain: tuning, the dilute share and the mixing law each change the tonnes a survey implies.
• Gravity weighs the CO₂. Seabed gravity surveys gave an in-situ CO₂ density of 720 ± 80 kg/m³ and an upper bound of 18 per cent on the share of the CO₂ dissolved in the brine (Alnes et al. 2011).

The CCS template that Sleipner established

The workflow Sleipner established is the one later storage projects follow:

• Pre-injection characterization (Parts 5-7 territory): inversion to porosity, lithology, fluid content; capacity estimation; injectivity testing; integrity assessment.
• Caprock + seal integrity (Parts 7 + 8.1): demonstrate that the caprock is continuous, low-permeability, and geomechanically sound (Section 8.1 stress analysis). At Sleipner the caprock is a 50 to 100 m mudstone that extends tens of kilometres beyond the plume.
• Baseline 4D survey (Section 8.2): high-quality 3D seismic acquired before injection starts. This is the reference against which all future monitors compare. Sleipner’s baseline is the 3D survey of 1994, two years before injection began.
• Injection + monitoring program: regular monitor surveys (Sleipner: 1999, 2001, 2002, 2004, 2006, 2008, 2010, 2013, 2016 and 2020). Each monitor compared to baseline + previous monitors to track plume evolution.
• Plume model + history matching: reservoir simulator predicts plume geometry; observed 4D refines simulator; simulator + observation converge over time.
• Regulatory reporting: annual or biennial reports to national + EU authorities (under EU CCS Directive 2009/31/EC, Norway’s implementing regulation). 4D monitoring provides the quantitative verification.
• Long-term stewardship transfer: after injection ends + a “post-closure” monitoring period (under the EU Directive at least 20 years, unless the authority is satisfied sooner that the CO₂ is permanently contained), responsibility can pass to the state. Sleipner is still injecting.

Later projects, Norway’s Northern Lights among them, build their monitoring plans on Sleipner’s experience. The seismic interpretation skills you’ve built across this curriculum are the foundation of that protocol.

Where the seismic-interpretation toolkit goes from here

This textbook has covered seismic interpretation in the context of subsurface fluid systems. The toolkit you’ve built generalizes much further:

• Carbon capture and storage: directly, as Sleipner demonstrates. Projects in operation or development include Quest (Canada, injecting since 2015), Northern Lights (Norway), Greensand (Denmark), Bayou Bend (USA) and the UK’s HyNet and East Coast Cluster. All use the workflow + interpretation methodology built throughout this curriculum.
• Geothermal energy: enhanced geothermal systems (EGS) require characterization of fracture networks, in-situ stress (Section 8.1), and reservoir thermal properties. Seismic interpretation, especially the rock-physics + AVO + anisotropy parts, transfers directly. Major projects: Utah FORGE (USA), Cornwall Eden (UK), Iceland deep drilling, geothermal-EGS projects in Germany + Australia.
• Mineral exploration: 3D seismic is increasingly used to map ore bodies (especially for deep targets where surface geophysics is insufficient). Critical-mineral exploration (lithium, cobalt, copper, rare earths) for the energy transition uses similar methods. Seismic + geological interpretation skills transfer directly.
• Geohazard assessment: imaging shallow geological features (faults, shallow gas, unstable slopes) for engineering applications, offshore wind anchor design, subsea infrastructure routing, urban subsurface mapping. The high-resolution seismic interpretation skills are valuable.
• Hydrogeology + groundwater: aquifer characterization, contamination tracking, salt-water intrusion mapping. The full QI workflow applies with appropriate frequency-band adjustments.
• Hydrogen storage: emerging applications storing hydrogen in underground caverns + depleted reservoirs. Requires characterization analogous to CCS. Several pilot projects globally.
• Subsurface science generally: from earthquake hazard assessment to volcano monitoring to planetary geophysics, seismic-interpretation methodology travels.