Page 50

Page 51

Photo by: Cathy Braathen

Students Working with Seismic modelling

Modelling of the Longyearbyen CO2 sequestration site






The modelling work focused on building 3D geological models capturing the geometry of the target aquifer and key 
reservoir heterogeneities. These geo-models served as input for flow simulations aimed at forecasting dynamic res-
ervoir behaviour and providing estimates of storage capacity. 

Input data for the geo-modelling included structural, sedimentological and petrophysical data compiled from logs 
and drill-core material. This was supplemented by topographical data, mapping and logging of sedimentary and struc-
tural features in outcrops of the target formation, and a limited number of seismic lines from Adventdalen (Tveranger 
et al. 2011; Bælum et al 2012. Farokhpoor et al. 2010; Braathen et al. 2012; Ogata et al. 2012; 2014; Senger et al. 2014a, 
2014b; 2015). Due to the low-porosity nature of the target-formation (8-12%; mostly secondary), particular attention 
was paid to characterize fracture networks; which were considered a key contributor to permeability.

The importance of fracture permeability is confirmed by history matching of fall-off curves from injectivity tests, 
which suggest a fracture-controlled flow; linear in the lower part of the reservoir and more radial in the upper part 
(Larsen 2012).

Owing to practical/logistical constraints to data collection (e.g. terrain/topography) the spatial distribution of input 
data to the modelling is highly uneven; there is no direct data from the central, western and eastern part of the area; 
high-resolution data were available from wells logs, cores and seismic lines in Adventdalen in the south and from 
detailed descriptions and logs from the reservoir outcrops along the northern margin of the reservoir. Consequently, 
modelling of reservoir architecture, properties and geometries for the bulk of the planned target aquifer had to rely 
on interpolation and extrapolation over distances of kilometres to tens of kilometres.

In particular the lack of seismic data from the central part of the reservoir left several key issues, especially with re-
spect to the cause and configuration of compartmentalization of the reservoir evidenced by well pressure data, open 
to conjecture – significantly influencing the accuracy of gross rock volume estimations used to establish potential 
storage capacity.  

The scenario-based calculations using all available data yield , mean effective pore volume ranging from 0.0064 to 
64 million m3, equivalent to an estimated 0.004-3.9 million tons of low-density (61.17/kg/m3) CO2 or 0.052-52 million 
tons of high-density (807.76 kg/m3) CO2. A sub-hydrostatic initial reservoir pressure combined with a the fact that 
the depth of target aquifer decreases from about 1000m in Adventdalen towards sea level in the north, adds uncer-
tainty to what CO2 phase will be present in the aquifer. Although the storage capacity estimates span three orders 
of magnitude, the evaluated scenarios, apart from the most negative case for areal extent of the reservoir, fulfil a 
requirement of injecting 0.2-1.2 million tons of CO2 of a period of 20 years. 

From a modelling point of view, Improvement of these estimates of storage potential will to a large extent rely on 
improved mapping and understanding the segmentation mechanisms acting in the reservoir. In order to achieve this, 
an expansion of seismic data coverage in the central part of the reservoir is considered indispensable.  Further inves-
tigation on CO2 saturation, -density and storage efficiency factors is also recommended.

Figure Above: Summary figure from Senger et al. (2015) outlining the scenario-based approach used to estimate 
gross rock volume for storage capacity. Areal extent of the target aquifer linked to injectors in Adventdalen varies 
from 1 to 1015 km2.