Large-Scale Geological Carbon Sequestration - Introduction to Carbon Capture and Sequestration - page 458

Section 2
Field-scale models
Geological carbon sequestration (GCS) operations rely on fi eld-scale
models to predict the behavior of injected CO 2 on length scales of tens of
kilometers and time scales of thousands of years. These models are
essential for optimizing CO 2 injection, predicting the economic cost of
carbon sequestration and the storage capacity of target formations, pur-
chasing CO 2 storage rights, planning potential remedial measures, and
securing public support and regulatory approval for carbon sequestration.
Key questions that the fi eld-scale models must accurately answer include:
How much CO 2 can be trapped in target formations? How rapidly can
CO 2 be injected? What will be the eventual extent of the CO 2 plume? How
will CO 2 injection impact regional hydrogeology? How long will monitoring
operations need to continue beyond the time of CO 2 injection?
Field-scale simulations also provide useful conceptual insights into
the infl uence of different phenomena on the performance of carbon
sequestration operations. These operations rely on predictions using
fi eld-scale models of multiphase fl ow and geochemistry in rock forma-
tions. In this section, we briefl y examine these fi eld-scale models. We
focus on two questions: how accurately do the fi eld-scale models predict
the fate of CO 2 in porous rocks? And how could these models be
improved through fundamental science?
Accuracy
The accuracy of fi eld-scale models can be evaluated in two ways. The
fi rst type of test compares the blind predictions of a geological carbon
sequestration model with existing fi eld-scale data ( Figure 10.2.1 ) [10.1].
This exercise is essential for building confi dence in the predictive ability
of models, but it is seldom carried out because of the limited availability
of appropriate experimental data. The second type of test probes the
sensitivity of model predictions to known uncertainties in model design
and input parameters ( Figure 10.2.2 ) and provides insight into the rela-
tive importance of different fundamental properties and processes [10.2,
10.3]. As illustrated in Figures 10.2.1 . and 10.2.2 , both approaches show