Large-Scale Geological Carbon Sequestration - Introduction to Carbon Capture and Sequestration - page 459

0.8

Observed

0.7

0.6

Predicted

0.5

0.4

0.3

0.2

0.1

0.0

0

2 6

4

t (10 2 days after CO 2 injection)

Figure 10.2.1 Blind prediction of data from a pilot site
Blind prediction of fi eld-scale sequestration data from the CO2CRC Otway pilot site in
Australia, where 65 kilotonne of CO 2 was injected into a depleted gas reservoir. The
graph shows predicted and measured values of CO 2 saturation, S g , as a function of time
at an observation well along the path of the CO 2 plume. The increase in CO 2 saturation
indicates the arrival of the CO 2 plume at the observation well. Figure redrawn from
Underschultz et al. [10.1].

that existing fi eld-scale carbon sequestration models yield useful qualita-
tive information on the behavior of CO 2 in geological formations but large
uncertainties are manifested in their quantitative predictions of basic
properties, such as the velocity of the CO 2 plume or the rates of various
CO 2 trapping mechanisms.
The quality of fi eld-scale model predictions depends on two features:
the design of the model, which must account for all relevant physical
phenomena, and the input parameters of the model, which must accu-
rately describe fundamental properties such as the CO 2 -brine equation of
state and specifi c properties of the rock formations, such as their perme-
ability. A model may fail to predict the fate of CO 2 because of inaccura-
cies in its design or because of a poor choice of input parameters. In the
case of carbon sequestration, both potential sources of inaccuracy pre-
sent signifi cant challenges: geological formations are complex sites