As part of the GeoCquest Field Validation project differences in the representation of lithological heterogeneity in geo-models and the respective predicted CO₂ migration was explored and validated. Two different geo-modelling workflows were applied to the Paaratte Formation at the Otway International Test Centre, the CO₂ migration between an injection well (CRC-3) and a monitoring well (CRC-8) predicted and the predictions compared to the observed arrival of CO₂ at the monitoring well. 

Both geo-model types were based on the same facies model. Conventional geo-models were developed by conditioning porosity to the facies and permeability was conditioned to facies using porosity as a trend. In contrast, in the GeoCquest geomodelling workflow pre-defined homogeneous and composite (heterogeneous) rock types and their properties were conditioned to the facies utilizing horizontal and vertical variography. Subsequently, porosity, horizontal and vertical permeability were generated using random data sampling based on Gaussian distribution, considering minimum, maximum, mean, and standard deviation values for each composite rock type. 50 realizations of the conventional and 50 realizations of the GeoCquest geo-models, both at a resolution of 3.3 x 3.3 x 1 m resolution, were used to simulate CO₂ migration from the injection to the monitoring well using the GEOS simulator. 

The two geo-model types resulted in large differences in the prediction of CO₂ arrival at the monitoring well. The GeoCquest approach predicted a mean arrival of CO₂ after 12-13 days, which closely matches the observed arrival using PNL data. Furthermore, the range in the predicted arrival was very small (about 10 days). In contrast, simulations using the conventional geo-model predicted a mean arrival after 50 days and a much larger range in the predicted arrival. These results suggest a much more accurate CO₂ migration prediction using the GeoCquest workflow, where rock properties for homogeneous and heterogeneous rock types are accounted for.